OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

Kant’s Modal Metaphysics

OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

Kant’s Modal
Metaphysics

Nicholas F. Stang

1
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Nicholas F. Stang 2016
The moral rights of the author have been asserted
First Edition published in 2016
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2015942949
ISBN 978–0–19–871262–6
Printed in Great Britain by
Clays Ltd, St Ives plc
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

Preface

The Unicorn and the Narwhal

The cover of this book reproduces a plate from Museum Museorum (1704), a kind of
‘cabinet of curiosities’ in book form by the eighteenth-century naturalist Michael
Bernhard Valentini.1 It depicts, on the upper right-hand side, a unicorn. Below that,
it depicts a narwhal, a species of whale (Monodon monoceros) that lives mainly in
Arctic waters near Canada, Greenland, and Russia, recognizable by the helical ‘horn’
that protrudes from the foreheads of adult males.2 There is a natural visual associ-
ation of unicorns and narwhals, reinforced by their German names: Einhorn and
Seeeinhorn.
The left-hand side of the drawing depicts a skeleton unearthed in 1663 by lime-
miners in Quedlinburg, near the city of Mecklenburg, now in the state of
Mecklenburg-Vorpommern. It became known as the ‘Quedlinburg unicorn’, and
over the next hundred years came to the attention of many of the great minds of
Europe. Its discovery is described by Otto von Guericke, the mayor of Mecklenburg
and a noted seventeenth-century naturalist, in a work on vacuum tubes.3 From von
Guericke’s account Leibniz learned of the Quedlinburg unicorn, while working in the
Harz Mountains in the 1690s; a drawing of it (not by his hand) is included in
Protogaea, or A Dissertation on the Original Aspect of the Earth and the Vestiges of
Its Very Ancient History in the Monuments of Nature, a work of natural history
Leibniz began during this period, but never finished. Interestingly, Leibniz himself
seems to have taken it to be the remains of an actual unicorn. The Quedlinburg
unicorn is even mentioned in an article by Baron d’Holbach in Diderot’s
Encyclopedia.4
While Leibniz followed Thomas Bartholin’s influential debunking of unicorn horns
in De unicornu observationes novae (1645) and accepted that most ‘unicorn horns’ are
narwhal tusks, he seems to have believed that the Quedlinburg skeleton was that of an
actual unicorn: “However, we should not hide the fact that a quadruped unicorn of the
size of a horse can be found in Abyssinia, if we have to believe the Portuguese

1
Its complete title is Museum Museorum, oder vollständige Schau-Bühne aller Materialen und Specer-
eÿen, nebst deren natürlichen Beschreibung, Election, Nutzen und Gebrauch.
2
In fact, the ‘horns’ of narwhals are not technically horns but tusks (essentially, very long teeth), but
I will ignore that complication.
3
Experimenta nova (ut vocantur) Magdeburgica de vacuo spatio (1672). Amsterdam.
4
A more complete history of the Quedlinberg unicorn can be found in the editor’s comments in Leibniz
(2008), 103; see also Cohen (2002) and Ariew (1998). I first learned of ‘Leibniz’s unicorn’ from the blog of
John McKay; see McKay, J. (2013). “Leibniz's Unicorn.” Retrieved February 20, 2015, from http://
mammothtales.blogspot.com/2013/06/leibnizs-unicorn.html.
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

vi PREFACE

Hieronimus Lupo and Balthasar Tellesio; and similarly, the skeleton extracted from
limestone in 1663 on Mount Zeunikenberg, next to Quedlinburg, looked more like a
terrestrial animal than anything else.”5 This interest in the Quedlinburg unicorn is just
one part of a larger early modern project of finding a place for ‘unicorn horns’ and
other buried animals remains (e.g., mammoths) within a changing picture of the
natural world. If these are unicorn remains, then a species of animals has gone extinct,
which means that the natural world is not a static creation but something that can
change and thus have a history. If they are not unicorn remains, then some other
explanation must be given of their provenance (e.g., that they are horns of narwhals,
or the tusks of elephants, etc.).
Kant would have been well aware of this literature. He possessed another book by
Valentini, Armentarium artis et naturae (1709), to which is appended a summary of
recent scholarly literature, including a capsule summary of De unicornu fossili (1666)
by Johann Lorenz Bausch. Bausch (who draws on Bartholin) argued that buried
‘unicorn’ horns, like that found in Quedlinburg three years earlier, are the fossizilized
tusks of mammoths and narwhals, not the remains of actual unicorns or the sui
generis products of geological processes, as some had maintained.6 It is thus note-
worthy that in his 1763 work The Only Possible Ground of Proof [Beweisgrund] for a
Demonstration of the Existence of God, Kant uses the narwhal and the unicorn as,
respectively, his examples of what does and does not exist:
[ . . . ] when existence occurs as a predicate in common speech, it is a predicate not so much of
the thing itself, but of the thought which one has of it. For example: existence belongs to the
narwhal [Seeeinhorn] but not to the land-unicorn [Landeinhorn]. This simply means: the
representation of a narwhal is an empirical concept; in other words, it is the representation of
an existing thing. (OPG, Ak. 2:72f)

In writing this passage, Kant likely had in mind cases like that of the Quedlinburg
unicorn and what, at the time of his writing, was a scientific ‘discovery’ of fairly
recent provenance: buried ivory is the tusks of narwhals or other animals, never the
fossilized remains of unicorns.7 In addition to the many criticisms Kant levels against
Leibniz’s metaphysics in that work, then, we can add another, more zoological
objection: unicorns, unlike narwhals, have never existed.
The Only Possible Ground of Proof, which I will refer to as ‘Beweisgrund’ for short,
will be a main focus of this book. But before we descend to the details of Kant’s
arguments, it is worth pausing to reflect on the many differences between narwhals
and unicorns: narwhals are aquatic, unicorns are not; narwhals have little hair, while
unicorns have flowing manes; narwhals eat fish, while unicorns are herbivores.

5
Leibniz (2008), 101.
6
Valentini (1709), on p. 15 of the supplementary volume, Historia literaria; see Warda (1922), 30.
7
Within a few years Kant would have known of the Quedlinburg unicorn itself, if he did not already in
1763, for he owned Hans Blumenbach’s 1769 work Handbuch der Naturgeschichte (Warda (1922), 27),
which contains an account of it, as well as a reference to Leibniz’s Protogaea.
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

PREFACE vii

However, one difference between them stands out: narwhals exist, while unicorns do
not. Intuitively, though, unicorns are possible; from a twenty-first-century perspec-
tive, if the evolutionary history of horses had been different, or if the ancestors of
narwhals had evolved into land mammals, then there would have been unicorns, and
some fossilized remains, like the Quedlinberg unicorn, might have been skeletons of
actual unicorns.8 How should we understand the difference between what actually
exists (e.g., narwhals) and what does not actually exist but is possible (e.g., unicorns)?
Is this a difference between objects that have, and objects that lack, a property, the
property of ‘existence’? Or is it wrong to say that there are non-existent objects in the
first place? What does it mean to say that unicorns do not actually exist but possibly
could have? And if unicorns are possible, how do we know this? Merely by imagining
them, or do we need to know something about the laws of nature to know whether
unicorns are possible? This book is the story of the development of Kant’s thinking
about these, and other, issues about existence and possibility. In other words, this is a
book about the difference between narwhals and unicorns.

8
However, the possibility of unicorns is not universally granted. In fact, it is the subject of a famous
debate in the semantics and metaphysics of modality. Saul Kripke denied that unicorns are possible
(Kripke (1980), 24, 156–8), while Michael Dummett upheld their possibility (Dummett (1983)).
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

Contents

Acknowledgments xiii
List of Figures xvii
List of Tables xix
Notes on the Text xxi

Introduction 1

Part I. Kant’s Pre-Critical Modal Metaphysics

1. Logicism and Ontotheism 13
1.1. Introduction 13
1.2. Logicist Metaphysics 14
1.3. Logicist Epistemology 21
1.4. Ontotheism 26
1.5. Possibilism 31
1.6. Real Predicates 36
2. Is Existence a Real Predicate? 43
2.1. Introduction 43
2.2. Descartes’ Ontological Argument* 43
2.3. Leibniz’s Ontological Argument* 49
2.4. Baumgarten’s Ontological Argument* 56
2.5. The Argument in Beweisgrund: Leibniz 65
2.6. The Argument in Beweisgrund: Baumgarten 72
2.7. The Argument in the Critique of Pure Reason 74
2.8. Absolute Positing 77
3. Real Conflict, Real Grounds, Real Possibility 80
3.1. Introduction 80
3.2. Logical Grounds 82
3.3. Real Grounds 88
3.4. Real Conflict 91
3.5. The Argument for Real Incompatibility 93
3.6. Real Possibility 97
4. Grounding Possibility 99
4.1. Introduction 99
4.2. Two Requirements on Possibility 100
4.3. Harmony, Power, and Intellect 107
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

x CONTENTS

4.4. “This thought rises far higher than a created being can reach” 116
4.5. Possibility, Thought, and Content 118
5. Kant’s Modal Argument 121
5.1. Introduction 121
5.2. Absolute Necessity 122
5.3. The Only Possible Ground of Proof 128
5.4. Prior Replies to the Plurality Objection* 135
5.5. E Pluribus Unum 138
5.6. How (Not) to Represent God 144
5.7. Kant’s Pre-Critical Modal Metaphysics 147

Part II. Kant’s Critical Modal Metaphysics

6. Real Possibility and the Critical Turn 153
6.1. Introduction 153
6.2. Modal Epistemology in the Prize Essay 154
6.3. Modal Epistemology in the Inaugural Dissertation 156
6.4. Relation to an Object 158
6.5. Intuition, Existence, and Possibility 162
6.6. Transcendental Philosophy and the Concept of an Object 166
6.7. A priori Cognition of Phenomena 171
6.8. No a priori Cognition of Noumena 183
7. Three Kinds of Real Possibility 197
7.1. Introduction 197
7.2. Critical Real Possibility 197
7.3. Formal Possibility 200
7.4. Empirical-Causal Possibility 214
7.5. Noumenal-Causal Possibility 219
8. Nomic Necessity 228
8.1. Introduction 228
8.2. What Nomic Necessity Is Not 230
8.3. Essences: Real and Logical 234
8.4. Essences, Natures, and Laws 238
8.5. Essentialism and Constructivism in Metaphysical Foundations 244
8.6. From Forms to Essences 252
8.7. Nomic Possibility as Real Possibility 254
8.8. Back to the Beginning 257
9. The Unity of Kant’s Modal Metaphysics 260
9.1. Introduction 260
9.2. Categories and Real Definitions 263
9.3. Unschematized Modal Categories 267
9.4. Absolutely Necessary Existence 272
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

CONTENTS xi

9.5. Postulates of Pure Theoretical Reason 280

9.6. The Necessary Ends of Theoretical Reason 288
10. The Antinomy of Kant’s Modal Metaphysics 297
10.1. Introduction 297
10.2. Intuitive Intellect, Intellectual Intuition 300
10.3. Some (Unsuccessful) Attempts to Resolve the Antinomy* 302
10.4. Resolving the Antinomy 307
10.5. Noumenal Freedom without Noumenal Modality 313
10.6. Representing Noumenal Modality 317
10.7. Absolute Positing, Existence, and Actuality 321
10.8. Back to Beweisgrund 325

Note on Sources and Key to Abbreviations and Translations 329

Bibliography 335
Index 351
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

Acknowledgments

I would like to thank the publishers of the following journals for permission to use
some material from three previously published papers:
• Stang, N. (2010). Kant’s Possibility Proof. History of Philosophy Quarterly, 27(3):
275–99.
• Stang, N. (2011). Did Kant Conflate the Necessary and the A Priori? Noûs, 45(3):
443–71.
• Stang, N. (2015). Kant’s Argument that Existence is not a Determination.
Philosophy and Phenomenological Research 91:3, 583–626.
I have been working on this project, in one form or another, for over ten years;
consequently, I have incurred a number of debts, great and small, to people who have
read and commented on various versions, numerous audiences to whom I have
presented various parts of various chapters, and all of the people who have helped
and encouraged me along the way.
It began its existence ten years ago as my Ph.D dissertation at Princeton University,
under the supervision of Béatrice Longuenesse and Desmond Hogan. I am especially
grateful to Béatrice, who has supported this project ever since I proposed, in my
second semester at Princeton, to write my thesis on Kant on modality. I continued to
work on this book during my six years at the University of Miami, where I developed
most of the arguments and interpretations in these pages. Two consecutive Chairs of
the Miami Department, Harvey Siegel and then Otávio Bueno, spurred me along in
this process, constant sources of encouragement and advice. Since coming to the
University of Toronto, Arthur Ripstein has been generous enough to read the whole
manuscript and give me invaluable advice on how to organize and structure the final
product to give readers a ‘way in’, though I am certain I have not done nearly enough
on that score. Thanks also to Ian Drummond for some excellent proofreading that
caught many of my numerous typos. Those that remain are my fault, for which I beg
the patience of the reader.
Along the way many people have been generous enough to read and comment on
drafts of various chapters. Thanks to Rae Langton, Eric Watkins, Lanier Anderson,
Beau Madison Mount, and Catharine Diehl for their comments on various parts of
the manuscript. Karl Schafer has read and commented on various chapters over the
years (including, thanks to Karl’s excellent comments, a rejected version of
Chapter 6). Lucy Allais gave me much-needed feedback on most of the chapters of
the book; by allowing me to read the page proofs of Manifest Reality before it was
published, Lucy also helped me massively improve the discussion of concepts and
intuitions in Chapter 6. Three anonymous readers from Oxford University Press
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

xiv ACKNOWLEDGMENTS

gave me extensive notes on the whole manuscript, which formed the basis of an
extensive revision beginning in the spring of 2014. Finally, Colin Marshall and Mike
Raven were generous enough to read, and extensively comment upon, almost the
whole manuscript of the book; thanks to Colin and Mike for helping me make the
whole work much better. I would also like to thank various audiences to whom I have
presented this material over the years, both for the opportunity and for their
insightful questions: grad seminars at UMass and Toronto; Departmental colloquia
at UMass, Irvine, Santa Barbara, Miami, Florida-International, Oxford, Stanford,
Michigan, Toronto, McMaster, and Western Ontario; as well as conferences in São
Paulo, Berlin, Atlanta, and Vancouver.
I would especially like to thank Tobias Rosefeldt, who generously invited me to
present the ideas from this book at the Berliner Kant-Kurs in July 2015. I learned an
enormous amount from the insightful comments and questions of the course parti-
cipants, but the book was already in production by that point. Where possible I have
tried to add a note or two responding to an objection, or adding a note of clarifica-
tion. I would like to thank everyone who came, but especially Andrew Stephenson,
Max Edwards, Noam Hoffer, Catharine Diehl, Karl Schafer, Lucy Allais, Stefanie
Grüne, Jim Kreines, Damian Melamedoff, Manish Oza, and Tobias himself for their
excellent comments.
The last ten years have been a particularly exciting time to be writing on Kant’s
theoretical philosophy, due to a generation of (I hesitate to call us ‘young’ anymore)
scholars whose work and whose feedback on my ideas has been a constant source of
inspiration and stimulation. I have been benefited in particular, over the years, from
reading the work of, and conversing with, Andrew Chignell, Des Hogan, Anja
Jauernig, Dina Emundts, Tobias Rosefeldt, Stefanie Grüne, Colin Marshall, Colin
McLear, Karl Schafer, Lucy Allais, James Messina, Dai Heide, Jessica Leech, Uygar
Abaci, Tim Rosenkoetter, and Clinton Tolley.
In addition to my intellectual debts, I would also like to acknowledge the many
friends who helped and encouraged me along the way. Eli Chudnoff, Brad Cokelet,
and Whitney Bauman offered friendship, good cheer, and someone to complain to,
over the years. Thanks to everyone who read and commented on the more extensive
version of the ‘Narwhals and Unicorns’ preface: Jennifer, Catharine, Tim, Dino,
Omer, Sanmay, Radu, and others. The first draft of this book was written during a
sabbatical year, when I lived at 75 Stewart Avenue in Bushwick, Brooklyn, where
I was fortunate enough to live with the best roommates of my life: Nancy, Jack, Meg,
and Barry. Thank you for all the great times, for letting me get away with never
cleaning the dishes, and for throwing me a surprise party when I ‘finished’ this book
in early 2013 (ha ha). By far the best thing that happened to me during my year in
Brooklyn, though, was meeting Michael Moss. I have been working on this book for
all the time I have known Michael and I offer my deepest thanks for all the love and
patience he has shown at every stage in its writing, rewriting, revision of the
rewriting, complaining about the revision of the rewriting, etc.
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

ACKNOWLEDGMENTS xv

I would also like to acknowledge several scholars whose work either appeared, or
became known to me, too late for me to engage with it in this book. Lanier
Anderson’s The Poverty of Conceptual Truth: Kant’s Analytic/Synthetic Distinction
and the Limits of Metaphysics (Oxford University Press, 2015) appeared while this
book was already in production. Omri Boehm’s Kant’s Critique of Spinoza (Oxford
University Press, 2014) covers several topics (esp. Beweisgrund) of this book from a
different direction; I became aware of the relevance of Boehm’s work too late to
include discussion of it here. Giuseppe Motta is the author of two highly significant
works on Kant’s theory of modality—Die Postulate des empirischen Denkens über-
haupt. Kritik der reinen Vernunft, A 218–235/B 265–287. Ein kritischer Kommentar
(De Gruyter, 2012) and Kants Philosophie der Notwendigkeit (Peter Lang, 2007)—
that deserve to be better known to English-language scholars. However, I found I
could not discuss Motta’s work in the depth it deserves without making this book,
already rather long, much longer. I have engaged critically with Anderson’s and
Boehm’s books in the pages, respectively, of the Australasian Journal of Philosophy
and the online journal Critique. I look forward to engaging with Motta’s work in
detail in the near future.
Finally, this book is dedicated to Charles Parsons, who taught my first class on the
Critique of Pure Reason in the Fall of 1999. I did not know at the time how much of
my intellectual life would be devoted to that book, but I am enormously grateful for
the start he gave me. Prof. Parsons has always been, and remains, my model of a
scholar and a philosopher.
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

List of Figures

4.1. The space of all possible predicates 106

4.2. Chignell’s space of all possible predicates 108
5.1. An infinite ascending chain of maximal beings 141
5.2. A series of maximal beings terminating in a GARP 141
5.3. Extramundane grounds of possibility 142
5.4. Extramundane grounds with a common ground 143
6.1. The table of nothings (organized by logical precedence) 168
7.1. Kinds of real possibility in Kant’s Critical system 201
8.1. The real essence of K 236
8.2. The Kantian system of sciences 245
8.3. Real essence of matter 252
8.4. Real essence of matter 252
8.5. Conformity to forms of experience 253
8.6. Conformity to forms of experience 253
9.1. Kinds of real possibility in Kant’s Critical system 271
9.2. Modes of validation 283
10.1. Negative and positive noumena 303
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

List of Tables

6.1. Kant’s disjunctive premise 186

6.2. Kant’s initial alternatives 187
6.3. Complete set of alternatives 187
9.1. Absolute necessities for immanent real possibility 278
10.1. The non-modal noumenal correlates of possibility and actuality 320
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

Notes on the Text

Sources. At this end of this book you will find a Note on Sources, which explains the
abbreviations I use to refer to the works of Kant and other important primary
sources, as well as a complete Bibliography.
Translations. Quotations from Kant are mainly from published translations (where
available), with some modifications by me; quotations from other thinkers are a
mixture of published translations and my own. The Note on Sources also contains
information on which translations were used.
Emphasis. The Critique was printed originally in Fraktur, using roman letters for
Latin phrases (like ‘noumenon’) and Fettdruck for emphasis. When quoting from
Kant, I have followed the editors of the Cambridge Edition in rendering the former in
italics and the latter in bold. Within quoted Kantian texts, underlining indicates my
emphasis.
Concepts and propositions. I use angle brackets to denote the concepts that would
normally be expressed by the italicized expression within, e.g., <substance> is the
concept of substance. I use italicized sentences to denote propositions; the sentence
‘Gold is a yellow metal’ expresses the proposition Gold is a yellow metal. I refer to the
relation between an object and a concept whose extension includes that object as
‘instantiation’, for example, particular substances instantiate <substance>. I realize
this will strike some readers as odd, since instantiation is usually taken to be a relation
between objects and properties (universals). However, the expression usually used by
Kant scholars for the relation between an object and a concept (an object is said to
‘fall under’ a concept), leads to many awkward constructions, and talk of ‘subsuming’
an object under a concept brings in a misleading reference to the agent who does the
subsuming.
Optional sections. Certain sections are not necessary to understand the larger argument
of this study; they are usually further developments of a line of thought, or consideration
of further texts. I have indicated such sections with an asterisk (*) in the Table of
Contents.
Supplementary. In order to keep this book at a reasonable length, I have left out certain
additional material, largely engagements with other scholars and more involved
technical developments of my own ideas. These can be found on my website:
https://sites.google.com/site/nickstang/.
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

Introduction

1. From Ontology to Transcendental Philosophy

This book is a study of Kant’s theory of possibility, starting from the so-called ‘pre-
Critical’ period of the 1750s and 1760s, through to the Critical system of philosophy
Kant published in the 1780s and the early 1790s. It discusses Kant’s confrontation
with the ontological theory of all possible beings shared by his predecessors (mainly
Leibniz, Wolff, and Baumgarten), his own positive theory of possibility in the 1760s,
and his distinctively Critical theory of possibility in the 1780s. Its guiding orientation
will be the idea that the way to understand both the change and the continuity
between Kant’s pre-Critical and his Critical theory of possibility is that he comes to
believe that the ontological question about possibility must be transformed into a
question in ‘transcendental philosophy’: how do we represent possible objects
a priori? My goal will be to explain Kant’s theory of possibility before he reoriented
his theory around that question, why he believed that reorientation was necessary,
what this question means, and his own answer to it. Possibility cannot be studied in
isolation, for it becomes one of the three ‘categories of modality’ in the Critique, along
with actuality/existence and necessity. Consequently, I will also investigate Kant’s
views about what distinguishes actuality from mere possibility, his theory of the
closely related notion of existence, and his theory of necessity, including, crucially,
his theory of necessary existence. Throughout, however, possibility, and the transi-
tion from ontology to transcendental philosophy, will provide the unifying thread
that runs through this whole study. However, since that transition occurs at a
particular point in the book (Chapter 6), I want to begin by explaining what ‘ontol-
ogy’ and ‘transcendental philosophy’ are and preparing the reader for the transition
from the former to the latter.
As every reader of the Critique of Pure Reason knows, that work is an attempt to
answer the question, how is metaphysics possible? Kant undertakes a ‘critique of pure
reason,’ an investigation into the nature and scope of our cognitive capacities, in
order to determine the scope of what we can cognize independently of sensory
experience, and whether it includes the concepts and claims of metaphysics. It is
also well known that Kant makes a traditional distinction between general metaphys-
ics and special metaphysics. General metaphysics, sometimes called ontology, is the
science of all possible beings qua possible; ontology studies the properties shared by
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 INTRODUCTION

all possible beings as such. Special metaphysics, by contrast, is the science of

particular kinds of beings, and divides into three specific sub-disciplines: psychology
(metaphysics of souls), cosmology (metaphysics of the universe), and theology
(metaphysics of God). Clearly, possibility is an important topic in ontology, for it is
something all possible beings share, and an adequate ontology will include some
account of what possibility is (what all possible beings have in common in virtue of
being possible) and what principles govern possibility. The delineation of the proper
subject matter of ontology requires us to distinguish possible beings (e.g., thinking
substances) from impossible ones (e.g., round squares), and the major metaphysics
textbooks of the eighteenth century make precisely such a distinction.1
Kant regularly lectured from such textbooks of metaphysics, mainly Baumgarten’s
Metaphysica, and some of his students transcribed these lectures.2 In the metaphysics
lecture transcripts available to us Kant makes this traditional separation between
metaphysica generalis (ontology) and metaphysica specialis. But he often begins his
discussion of ontology by claiming that the distinction of possible from impossible
beings is a distinction within a more general concept, the concept of an ‘object’
[Gegenstand, sometimes the Latinate Object]. One such instance comes from the
Mrongovius transcripts from the winter semester 1782–3, between the publication of
the two editions of the Critique:
[ . . . ] if two concepts are opposed—here [in Baumgarten’s Metaphysica], for example, possible
and impossible—then they always stand under a higher concept, for opposition always
presupposes a disjunctive proposition. Now there must be a divided concept that has the
opposing concepts as members of the division, and this is a higher concept. Object [ . . . ] is
surely the highest concept in ontology. The possible we call ‘thing’ [Ding], ‘something’ [Etwas],
and opposed to it is the impossible, nothing [Nichts]. (MM, Ak. 29:8113)

This remark, and others like it, is mirrored by a short passage in the Critique that is
easy to overlook:
The highest concept with which one is accustomed to begin a transcendental philosophy is
generally the division into the possible and the impossible. Since all division, however,
presupposes a concept that is divided, a yet higher concept must be given, and this is the
concept of an object in general [Gegenstand überhaupt] (taken problematically, and unspeci-
fied whether it is something or nothing). (A290/B346)

I take the concept <object> here to be the concept <object of representation>. I will
argue for this in the course of this study; for now it will constitute an interpretive
hypothesis.
In the Critique Kant describes <object> as the highest concept of “transcendental
philosophy.” That the readers of the Critique can be “accustomed” to something in

1 2
See Ch. 1.2 for specific references. Ak. 28–9.
3
Cf. MV (Ak. 28: 410 f.), MvS (Ak. 28: 477–9 f.), ML2 (Ak. 28: 542 f.), and esp. MK3 (Ak. 29: 960).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

INTRODUCTION 

“transcendental philosophy” means that the latter phrase does not refer to the
distinctive philosophical project inaugurated by that work (as it does in other
contexts), but to some more general type of philosophy, of which both pre-Kantian
ontology and the Critique are instances. Kant defines this generic notion of tran-
scendental philosophy in the Introduction to the A edition: “I call all cognition
transcendental that is occupied not so much with objects but with our a priori
concepts of objects in general” (A11).4 This applies to pre-Kantian ontology, since
the distinction between the possible and the impossible is not a distinction between
two kinds of objects (i.e. round squares are not a kind of object) but between two
concepts: the concept of the possible and the concept of the impossible. From a
Kantian perspective, the metaphysics textbooks of Wolff and Baumgarten are cata-
logues of a priori concepts, principal among them <possible> and <impossible>.
In addition to this general idea of transcendental philosophy (the study of a priori
concepts of objects, not directly of objects themselves), Kant also uses this term more
specifically to refer to the philosophical project inaugurated by the Critique itself:
“not every a priori cognition must be called transcendental, but only that by means of
which we cognize that and how certain representations (intuitions or concepts)
are applied entirely a priori, or are possible” (A56/B80). The guiding question of
transcendental philosophy properly so called is: how is it possible (for us) to represent
an object a priori (independently of sensory experience)? This is what Kant means
when he says that the concept <object of representation> comes before the concepts
<possible> and <impossible> in transcendental philosophy: we must first inquire into
our capacity to represent objects, specifically, our capacity to represent them a priori.
The ontology of Wolff, Baumgarten, and others erred, according to Kant, in directly
considering the concepts of possibility and impossibility, without first inquiring into
the cognitive capacities by which we represent objects in the first place.
But the difference between ontology and its successor, Kantian transcendental
philosophy, is not merely that the second considers an antecedent question about a
more general concept (<object of representation>). By prefacing its inquiry into
possibility with that question, transcendental philosophy transforms both the ori-
ginal question of ontology (the nature of all possible beings as such) and its basic
distinction (between possibility and impossibility). By subordinating <possible> and
<impossible> to the more general concept <object of representation>, Kant transforms
these into the concepts <possible object of representation> and <impossible object of
representation>. Consequently, the fundamental question of ontology becomes the
following question in Kantian transcendental philosophy: how is it possible for us to
represent possible objects a priori? Ontological questions about the nature of possible

4
This sentence is emended in the B edition to: “I call all cognition transcendental, which is not
concerned so much with objects but with our mode of cognition of objects, insofar as this is supposed to
be possible a priori” (B25–6). This is the specific concept of transcendental philosophy that does not apply
to e.g., Wolff and Baumgarten.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 INTRODUCTION

beings as such are transformed, in Kant’s hands, into cognitive-semantic questions

about our representation of possibility. It will take the rest of this book for me to
explain why Kant thinks this transformation is necessary, and its larger implications
for his theory of possibility.

2. Breakdown of Chapters
I begin, in Part I, with Kant’s confrontation in the 1750s and 1760s with the ontology
of Wolff and Baumgarten. Wolff and Baumgarten share with Leibniz, their great
predecessor in the German rationalist tradition, a view about possibility which I call
logicism: anything that is not logically contradictory is possible. Consequently, the
principle of non-contradiction is the highest principle of ontology. The logicists also
hold a corresponding principle about necessity: a necessary truth is one whose
negation entails a contradiction. The space of possible being, for the logicists, is a
logical space, consisting of logically consistent concepts of possible beings and logical
relations among those concepts.
The logicists hold the traditional view that God exists necessarily. Combined with
the logicist view of necessity, this entails: the proposition God does not exist logically
entails a contradiction. The reason this proposition logically entails a contradiction is
that existence is contained in God’s essence, his mere possibility. God is the unique
being whose mere possibility entails his existence. I refer to the doctrine that God
exists in virtue of his essence as ontotheism. A term coined by Kant himself,
ontological argument, refers to arguments that purport to derive God’s existence
from his essence.
Chapters 1 and 2 focus on Kant’s critique of ontotheism and ontological argu-
ments. Although Kant’s objections to the ontological argument in the Critique of
Pure Reason are well known, they are a continuation of a line of argument he had
been pressing, in one form or another, since 1755. After some preliminary back-
ground on the metaphysics and epistemology of logicism, I explain why logicism,
paired with the assumption that God exists necessarily, commits these thinkers to the
doctrine that God exists in virtue of his essence. I then show that Kant rejects
ontotheism: there can be no ontological arguments because there cannot be a being
whose essence grounds its existence. His objection to ontological arguments—that
existence is not a ‘real predicate’ or ‘determination’—is one of his most famous, but
also one of his least well understood, doctrines. I argue that this doctrine means that
existence is not a predicate that applies to a subset of objects: objects do not divide
into the existent and the non-existent. Existence can be used as a predicate of objects
themselves, in which case it necessarily applies to all objects there are, or it can be
applied to concepts, in which case it applies to concepts that are instantiated by at
least one object. I argue that ontotheism as such is committed to denying this; the
ontotheistic explanation of why God necessarily exists—his essence contains
existence—entails that existence is a ‘real predicate’ that potentially applies only to
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

INTRODUCTION 

a subset of objects. If Kant is right that existence is not a real predicate, then
ontotheism is false. If ontotheism is false, then logicisim is false, for it entails
ontotheism (given the assumption of divine necessity). If Kant can show that God
does not exist with logical necessity, then he can show that there is a distinction
between what is necessary (e.g., that God exists) and what is logically necessary, and,
correlatively, between what is possible and what is merely logically possible (e.g., that
God does not exist).
In the first half of Chapter 2 I reconstruct in detail the ontological arguments of
Descartes, Leibniz, and Baumgarten to show they are implicitly committed to the
doctrine that existence is a real predicate. Although Descartes is not a logicist, he is an
ontotheist, and, since his ontological argument was deeply influential on Leibniz, and
hence on Wolff and Baumgarten, I include him in my discussion of (otherwise
logicist) ontological arguments. In the second half of Chapter 2 I reconstruct in detail
Kant’s arguments that existence is not a real predicate, in both his 1763 work The
Only Possible Ground of Proof [Beweisgrund] for a Demonstration of the Existence of
God and the better-known arguments in the Critique of Pure Reason. I argue that
none of Kant’s arguments should trouble a sufficiently sophisticated ontotheist. No
matter how intuitive many of us post-Kantians may find this doctrine, Kant fails
to prove that existence is not a real predicate. However, once Kant adopts this view, it
requires a new metaphysics of existence, one in which possibility-facts and existence-
facts are insuperably different: no being exists in virtue of its essence (its possibility).
This divorce between possibility and existence, between concept and object, will have
far-reaching consequences for Kant’s philosophy.
Chapter 3 concerns Kant’s criticism of another key logicist doctrine. The logicists
hold the natural view that certain pairs of predicates (properties)5 are metaphysically
incompatible: it is impossible, for instance, that one and the same being thinks and is
extended. Logicism entails that if a pair of predicates is metaphysically incompatible,
they are logically inconsistent (their co-instantiation entails a contradiction).
I reconstruct Kant’s sophisticated argument in Attempt to Introduce the Concept of
Negative Magnitudes into Philosophy (1763) that there is non-logical metaphysical
incompatibility between predicates. That argument depends upon a prior Kantian
argument, that the relation between a cause and its effect cannot be modeled as a
logical relation. I critically reconstruct both arguments and show how Kant’s argu-
ment that there are metaphysically incompatible but logically consistent predicates
offers further support to his anti-logicist doctrine that there are logical possibilities
that are not possible simpliciter, or as we might now say, not metaphysically possible
(e.g., that an extended thing thinks).
Kant’s term for ‘metaphysical’ possibility as distinct from logical possibility is real
possibility. Chapters 4 and 5 concern Kant’s own positive theory of real possibility in

5
On the relation between predicates, properties, and concepts see Ch. 3.1.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 INTRODUCTION

the 1750s and 1760s, focusing on Beweisgrund. The core idea of Kant’s pre-Critical
theory of real possibility is that real possibility requires not only the logical consist-
ency of its elements, but that the atomic predicates of which real possibilities are
composed be possible in their own right. This in turn requires that there be an existing
being that grounds or makes possible these atomic predicates. Kant argues at length
in Beweisgrund that God is the ultimate such ground of possibility, without which
nothing would be really possible. Before reconstructing that argument, though,
I examine in Chapter 4 what it means to say that God is the ground of real possibility.
I canvass various interpretations that have been offered in the secondary literature
and argue that none of them is compatible with Kant’s texts. I do not, however, go
on to offer my own positive interpretation, for Kant explicitly states that we cannot
understand the grounding relation between God and real possibility. In fact, Kant
expresses puzzlement about our ability to so much as think about (mentally repre-
sent) this relation. In Chapter 5 I reconstruct in detail Kant’s argument that there is a
being that grounds all real possibility. I argue that, although he explicitly claims we
cannot understand this grounding relation, his argument implicitly requires us to
think of God as grounding real possibilities through his powers. We are simultan-
eously required, and forbidden, to represent God as grounding real possibility
through his powers.
Part II concerns Kant’s modal theory in the Critical period, and it begins, in
Chapter 6, with an account of the role of modality in the ‘Critical turn.’ I point out
that in Beweisgrund, and the nearly simultaneous ‘Prize Essay,’ Kant presupposes
that real possibilities are ‘given’ to us (they are cognitively available to be thought
about by us) but fails to develop an adequate philosophical account of how and why
they are given. This further compounds a problem already diagnosed in Beweis-
grund: Kant lacks an account of how we can mentally represent the grounding
relation between God and real possibility. Kant’s pre-Critical modal theory lacks
any adequate account either of the epistemology or cognitive semantics of real
possibility (how we can mentally represent it). But the lack of such an epistemology
(or semantics) is not a minor flaw, for it entails that Kant lacks an adequate account
of how we can mentally represent his own modal metaphysics, much less know it
to be true.
A key step on the road to the Critical philosophy comes in the Inaugural Disser-
tation (1770), when Kant distinguishes between two fundamentally different kinds of
representation: intuitions and concepts. An intuition is an immediate presentation of
a singular object, while a concept is a general representation that may or may not
have any instances. This requires a reformulation of the questions in modal seman-
tics and epistemology from above. Since intuitions immediately ‘give’ us objects
(make them available for singular thought), the questions about possibility become:
how do we know a priori which of our concepts are really possibly instantiated (by an
object)? And what is the content of the thought that one of our concepts is really
possibly instantiated (by an object)?
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

INTRODUCTION 

In a famous 1772 letter to Marcus Herz, Kant poses a question (“what is the
ground of the relation of that in us which we call ‘representation’ to the object?”) that,
according to one popular narrative, motivates the project of the Critique. I argue that
this is a question in modal semantics and epistemology: what explains our possession
of a priori concepts that are really possibly instantiated by objects? And, how do we
know a priori that our concepts are really possibly instantiated? The epistemic
question becomes especially pressing when applied specifically to the categories
(the fundamental concepts of metaphysics), because Kant shares with his ontologist
predecessors the view that metaphysics concerns possible beings and that we must be
able to prove that its fundamental concepts are possibly instantiated. If we cannot
know a priori that the categories are concepts of possible objects, then metaphysics is
impossible. Kant answers this question by distinguishing between concepts and
intuitions, and dividing all objects überhaupt into those that can be objects of sensible
intuition (phenomena) and those that cannot be such objects (noumena).
I reconstruct Kant’s positive explanation of how we can have a priori knowledge
that the categories are possibly instantiated by phenomena and his negative argu-
ment that we cannot have such knowledge of noumena. Consequently, there can be a
priori metaphysics but its generality is limited: it applies only to phenomena. Because
there can be no a priori metaphysics of noumena, metaphysics of all possible beings
(ontology, metaphysica generalis) is impossible for discursive intellects like ours
(intellects that sensibly, or passively, intuit objects).
Chapter 7 treats Kant’s Critical theory of possibility, necessity, and contingency
within the empirical world. Through a close reading of the ‘Postulates of Empirical
Thinking in General,’ I argue that Kant implicitly distinguishes the formal possibility
of an empirical object from its causal possibility. An empirical object is formally
possible just in case it is compatible with the forms of our experience. An empirical
object is causally possible just in case there is an actual object that can causally bring
it about. However, Kant’s transcendental idealist doctrine that empirical objects are
appearances of noumena entails that they have two kinds of grounds: other empirical
objects, and the noumena that affect us in experience. Thus, an empirical object is
empirically-causally possible if and only if another empirical object can causally
produce it, and is noumenally-causally possible if and only if noumena can cause
that object to appear. The clearest case where the modalities diverge, I argue, is Kant’s
theory of freedom. If I perform an action, empirical determinism entails that the
omission of that action is empirically-causally impossible. If empirical objects had
only empirical grounds, we would lack freedom. I am free to omit an act I actually
perform because it is noumenally-causally possible for me to cause that act not to
appear: as a noumenon, I have the power to cause that act to be omitted.
Chapter 8 examines in greater detail Kant’s theory of modality as applied to the
empirical world. Kant claims both that empirical causal laws (e.g., the inverse-square
law of attraction) are not formally necessary and that all laws as such are necessary.
In what sense, then, are laws necessary? I call this kind of necessity ‘nomic necessity’
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 INTRODUCTION

and examine, and reject, various alternatives proposed by earlier commentators.

Through an analysis of Kant’s theory of matter in Metaphysical Foundations of
Natural Science, I argue that Kant accepts the existence of natural kinds (constituted
by objects that share an essence) and that nomically necessary truths are truths
grounded in the essences of natural kinds of empirical objects. In particular, the laws
of motion are grounded in the essence of matter, the natural kind picked out by the
concept <matter>, the most general empirical concept, which applies to all outer
objects. I defend this interpretation in light of Kant’s apparent skepticism about our
knowledge of the essences of empirical objects and their kinds.
In Chapter 9 I return to Kant’s pre-Critical modal theory and show that it is largely
retained in the Critical system, bringing an unexpected degree of historical and
systematic unity to Kant’s modal metaphysics. His modal theory is historically
unified because he retains within the Critical system the pre-Critical view that God
is the absolutely necessary ground of all real possibility, but demotes its epistemic
status: what he considered an object of a priori rational knowledge in Beweisgrund is
now an article of ‘necessary theoretical belief.’ We are rationally required to postulate
such a first ground of real possibility in order to explain why anything is really
possible, but we do not thereby come to know that such a being exists. This historical
continuity in Kant’s modal metaphysics also shows the systematic unity of that
metaphysics in the Critical system. The various different kinds of real possibility he
accepts are unified by a common scheme: real possibilities are grounded in an
absolutely necessary first ground. For each kind of real possibility, there is a corres-
ponding conception of an absolutely necessary ground: space (formal possibility), the
laws of nature (empirical possibility), real essences (nomic possibility), and God
(noumenal possibility). Kant retains his pre-Critical conception of real possibility,
generalizes from it to a set of general conditions on any concept of real possibility
whatsoever, and then instantiates it at the level of various particular kinds of
possibility. Far from being a mere aggregate of different, or even incompatible,
conceptions of modality, Kant’s Critical modal metaphysics is unified by his con-
ception of possibility as grounded in an absolutely necessary being.
In Chapter 10 I diagnose and resolve what I refer to as the ‘antinomy’ of Kant’s
Critical modal metaphysics. His theory of freedom rests on attributing modal
properties to the wills of finite rational agents (considered as noumena), specifically
the possibility that they act otherwise than they actually do, but in his influential
discussion of the ‘intuitive intellect’ in }76 and }77 of the Critique of Judgment, he
claims that an intuitive intellect would not cognize its objects as possibly being
otherwise than they actually are. If there is a God, he has an intuitive intellect, and
what is more, an omniscient intuitive intellect: he cognizes all of the properties of
noumena as such; it follows that noumenal wills lack modal properties. In this
chapter I resolve this modal antinomy by arguing that, although God does not
cognize modal properties in noumena (hence they do not have such properties), he
does cognize a ‘noumenal correlate’ of modal properties: dependence on his essence
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

INTRODUCTION 

and dependence on his will. These noumenal correlates are not ‘in themselves’
modal, but we represent them modally as possibility and actuality, respectively,
when we apply the concepts of noumenal possibility and actuality to them.
I develop a theory of these correlates and the content of the concepts of noumenal
real possibility and actuality that explains how those concepts represent those
correlates. I argue that we can retain the freedom of noumenal wills while holding
that possibility is an artifact of our modalized representation of them by developing a
non-modal account of noumenal freedom as the power to act otherwise. I explain
how this non-modal power to act otherwise is represented modally by us as the
ground of the possibility that we act otherwise.
I then use this account of noumenal correlates and our concepts of noumenal
modality, retrospectively, to clarify a tension that arose in Kant’s pre-Critical modal
metaphysics: Kant denies that God grounds possibility through his powers (because
his powers are among the possibilities that need grounds), but implicitly thinks of
God as grounding possibility causally through his powers. From a Critical perspec-
tive, there is no modality at the purely noumenal level, so the possibility of God’s
powers does not need to be accounted for; we modalize these powers by represent-
ing them as the ground of the possibility of the properties they are the powers to
cause to be instantiated. What is more, we resolve another outstanding issue from
Beweisgrund: our ability to so much as represent God as the ground of all real
possibility. We can represent God as the ground of possibility because possibility
is an artifact of our modalized representations, not something in noumena ‘in
themselves.’ To return full circle, this completes Kant’s transition away from
modal ontology (metaphysica generalis) to the study of modal representation:
possibility does not characterize noumenal beings in themselves but is an artifact
of how we represent them. The first question in modal metaphysics is not: What is a
possible being? but: What is it to represent an object as possible?
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

PART I
Kant’s Pre-Critical Modal
Metaphysics
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

1
Logicism and Ontotheism

1.1. Introduction
As Kant developed his own positive views in modal metaphysics in the pre-Critical
period, he did so primarily in reaction to the ‘school metaphysicians’ Christian Wolff
and Alexander Gottlieb Baumgarten1 and the available writings, such as they were, of
Gottfried Wilhelm Leibniz.2 Kant’s modal theory was also influenced by Christian
August Crusius, but since Crusius was himself critical of what he saw as the excessive
rationalism of Leibniz and Wolff, I discuss him later, in Chapters 3 through 5. While
Leibniz, Wolff, and Baumgarten differed significantly over central metaphysical
issues like the possibility of inter-substantial causation, the nature of the ‘pre-
established harmony,’ and the relation between simple substances and extended
bodies, they shared a set of core doctrines about modality that I will call logicism.3
Since logicism will form the background of the next three chapters, I devote the
first two sections of this chapter to exploring the logicists’ metaphysics (section 2)
and epistemology (section 3) of modality. Section 4 introduces a key pair of concepts:
ontotheism and ontological arguments. Ontotheism (my term) refers to the view,
held by many early modern thinkers, that God exists in virtue of his essence. An
ontological argument is one that purports to derive the existence of God from his
possibility, from his essence. Ontotheism entails therefore that there is a sound
ontological argument, an ontological proof.4 The logicists accept the traditional view

1
For instance, in his critical discussion of philosophical theories of existence in Beweisgrund Kant
mentions only Wolff, Baumgarten, and Crusius (Ak. 2: 76–7).
2
When discussing Leibniz in this historical context, it is important to bear in mind that only a fraction
of Leibniz’s writings were available in the eighteenth century. Kant and his contemporaries engaged with
the Leibnizian doctrines presented in those texts. Texts available in this period include the Theodicy, New
Essays on Human Understanding, the Clarke correspondence, Monadology, “Meditations on Knowledge
Truth and Ideas,” “A New System of Nature,” Part I of the Specimum Dynamicum, “On an Emendation of
First Philosophy,” and “On Nature Itself.” We must therefore keep separate Leibniz as it would have been
reasonable for Kant or his near contemporaries to read him and Leibniz as we now know him.
3
For an overview of this period in Kant’s development, see Schönfeld (2000) and Watkins (2005); see
esp. Watkins’s discussion of the differences between Wolff and Leibniz in his (2005), 46–50, as well as his
(2006). For a general overview of German philosophy in the centuries and decades before Kant, see Beck
(1969) and Wundt (1939) and (1945).
4
Kant typically speaks of the ‘ontological proof [Beweis]’, but since proof entails soundness I want to
separate ontological proofs (which, according to Kant, are impossible) from ontological arguments, which
are putative ontological proofs.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

that God exists necessarily, and this commits them, given their logicist account of
necessity, to the claim that God necessarily exists in virtue of the logical containment
of existence in his essence. In section 5 I argue that ontotheism entails that there are, or
could be, merely possible but non-existent objects. I call this view possibilism. In
section 6 I argue that Kant’s famous objection to the ontological argument—that
existence is not a real predicate—means that ‘exists’ is a predicate that necessarily
applies to every object there is: it is not possible for there to be non-existent objects. If
Kant is right that existence is not a real predicate, then ontotheism is false and there
can be no sound ontological arguments, for nothing, not even God, can exist in virtue
of its essence. If ontotheism is false, then logicism is false, for there is a necessary truth
(that God exists) that is not logically necessary.

1.2. Logicist Metaphysics

Wolff and Baumgarten conceive of metaphysics in modal terms. They divide this
philosophical science into general metaphysics (ontology), and special metaphysics,
consisting of the rational sciences of theology, cosmology, and psychology. Following
Aristotle’s definition of ‘first philosophy,’ they define ontology as the science of
beings qua beings.5 Wolff writes, in the preface to the Ontologia: “ontology is the
science of a being insofar as it is a being [scientia entis in genere, seu quatenus ens
est].”6 Baumgarten begins the ‘Ontology’ section of Metaphysica with this definition:
“Ontology (metaphysica, metaphysica universalis, philosophia prima) is the science of
the most general predicates of a being [entis].”7 ‘Being’ (Ding, ens) refers to any
possible entity.8 Thus, a science of “all possible beings [Dinge], insofar as they are
possible,” as Wolff defines philosophy [Welt-Weisheit] in the so-called ‘German
logic’ is also a science of beings, insofar as they are beings, that is, “the science of
the most general predicates of beings.”9 ‘Being’ is the highest genus in ontology. It

5
See Metaphysics 1026a16. Note, though, that while Aristotle defines first philosophy as the science of
being qua being, Wolff and Baumgarten define ontology as the science of beings (entia) qua beings, a
difference that would later be emphasized by Heidegger; see Heidegger (1975), }3.
6
Ont. }1. 7
Meta. }4.
8
See Dt.Met. }16 and Meta. }8. As Dirk Effertz notes in his translator’s introduction to Wolff (2005),
Wolff treats the German Ding as equivalent to the Latin ens (Eberhard follows him in this in his German
translation of Baumgarten, Baumgarten (1783)); compare the definition of Ding in Dt.Met. }16 (“alles was
seyn kann, es mag wirklich oder nicht”) to the definition of ens in Ont. }134 f. (“quod existere potest
consequenter cui existere non repugnat”). This is somewhat unfortunate, because in later German philoso-
phy (including, in some but not all passages, Kant himself) Ding refers specifically to an ens insofar as it has
concrete reality and (in Kant’s case) forces. It is also unfortunate, because, to the English ear, thing (the
cognate of Ding) has a more concrete connotation than being (ens). Cf. Honnefelder (1990), 315. Thanks to
Abe Stone for drawing my attention to these issues, and to Corey Dyck for further discussion.
9
Dt.Log. }1. Compare Wolff ’s definition of philosophy in Discursus Praeliminaris: the “science of the
possibles, insofar as they can be” (}1). The equivalence of these definitions is confirmed by Johann August
Eberhard, a contemporary of Kant, who gives the following gloss on Baumgarten’s definition of ontology:
“because the predicates which appear in ontology apply to beings [Dinge], not because they belong to a
particular kind, but because they are beings: so ontology can be explained by Aristotle, the Scholastics, and
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

includes everything there is: God, finite substances, their modes, space, time, etc. To
be a being is to be possible; whatever is not possible is a non-being, and does not fall
within the purview of ontology, or any science for that matter. Ontology is the science
of possibles qua possible, and therefore the most general of all sciences. Kant himself
quotes this definition of ontology and of ‘being’ approvingly; in the Mrongovius
metaphysics lectures, he says “we now begin the science of the properties of all beings
[Dinge] in general, which is called ontology” (Ak. 29: 784).10
Wolff and Baumgarten share with Leibniz a core set of views about modality I will
call logicism. First, they distinguish between what is possible in itself (in se) and what
is possible in connection with what actually exists (hypothetice). I will focus almost
exclusively on the former notion, which corresponds with Leibniz’s idea of strict or
metaphysical possibility; unless otherwise noted, possibility (necessity) refers to in se
possibility (necessity), not possibility (necessity) hypothetice.11
Wolff and Baumgarten define (in se) possibility, impossibility and necessity as
follows:
Wolff:
[ . . . ] one can further see that whatever contains [enthält] nothing contradictory in itself is
possible, that is, whatever not only can itself exist next to other things that are or can be, but
also contains in itself only those things that can exist next to each other is possible.12
When what is opposed to a thing contains in itself something contradictory, the thing is
necessary. Now since what contains something contradictory in itself is impossible (}12), that
which is opposed to something necessary is impossible. And if what is opposed to a thing is
impossible that same thing is necessary.13

Baumgarten:
Every thing that is representable, that does not contain a contradiction, which is not A and
not-A, is POSSIBLE.14
NECESSARY is that whose opposite is impossible; the non-necessary is CONTINGENT.15

However, these definitions of possibility and necessity are hard to interpret. While they
purport to be definitions of what it is for any ‘being’ (Ding, ens) in general to be possible
or necessary, they are given in terms of the (non-)containment of a contradiction in it,

Wolff as the science of beings in general, insofar as they are beings” (see Eberhard’s editorial comments to
Baumgarten (1783)).
10
Cf. Kant’s announcement for his lectures for the winter semester 1765–6: “I shall then proceed to
ontology, namely, the science which is concerned with the most general properties of all beings [Dinge]”
(Ak. 2: 309).
11
For their theories of possibility hypothetice and its role in their defense of contingency, see Dt.Met. }}
574–5 and AzDM }195, 197; and Meta. }}15–18, 102–5. Baumgarten and Wolff inherit from Leibniz the
distinction between what is ‘in itself ’ or ‘strictly’ possible, and what is possible ‘in connection’; see “On
Freedom and Possibility” (AG 19–23).
12
Dt.Met. }12. Translation from Watkins (2009), 10. Cf. AzDM }6.
13
Dt.Met. }36. Translation from Watkins (2009), 13. 14
Meta. }8. 15
Meta. }101.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

but only concepts, or other logically structured items, can contain contradictions.16
A concept contains a contradiction just in case it contains both a concept C and its
negation ~C.17 Wolff and Baumgarten appear to be claiming that:
(PossibilityC) A concept is possibly instantiated if and only if it contains no
contradictions.
However, their definitions of necessity are given in terms of ‘opposites’ and only
propositions have ‘opposites’—their negations. These are most naturally read as
definitions of the necessity of a proposition:
(NecessityP) For any proposition p, it is necessary that p if and only if ¬p entails a
contradiction.18
Entailing a contradiction is the natural propositional equivalent of a concept’s
containing a contradiction.19 How then do we unify this definition of possibility
and this definition of necessity? There is a natural way of converting a claim about a
concept’s possible instantiation into a claim about the possibility of a proposition
being true: C is possibly instantiated if and only if the proposition C is instantiated is
possibly true. So we might define possibility for propositions as follows:
(PossibilityP) For any proposition p, it is possible that p if and only if p does not
entail a contradiction.20
One might wonder how well this agrees with the original definition of possibility in
terms of the instantiation of concepts, (PossibilityC). I think they are co-extensive. If
the proposition C is instantiated entails a contradiction, this must be because the

16
It is somewhat unclear what these thinkers took to be the exact ontological status of concepts,
propositions, and other logical entities. For ease of exposition, I will be assuming they all share the
following views, which are derived from (a certain reading of) Leibniz. First, they are nominalists:
everything that exists is an individual (including individual particulars, and individual properties, or
modes). The thoughts of an individual mind are modes of that mind, not shareable by any other mind.
So my token concept of water is numerically distinct from God’s token concept of water. However, we can
talk about ‘the’ concept of water by talking about what all token concepts of water have in common in
virtue of being token concepts of water. Since all minds express the complete truth about the world
(universal harmony), this means that all token concepts of an object will have the same content (but vary in
their clarity and distinctness). Talk of concepts is a compendium loquendi for talk of individual token
concepts in individual minds. Likewise, talk about ‘the proposition’ that p is a compendium loquendi for
talk about individual propositionally structured thoughts inhering in individual minds: what all such token
thoughts have in common in virtue of having the same content. See Mates (1986), 170–88, for discussion.
17
I use ~ for the negation of a concept and ¬ for the negation of a proposition or judgment. I explain
why this difference might be significant in Ch. 2.4. For more on this distinction, see Stang (2012).
18
Baumgarten defines the impossible as “that which contains [involvens] a contradiction, or which
implies [implicans] a contradiction” (Meta. }7); given that the necessary is that whose ‘opposite’ is
impossible (}101), the definition given in the body of the text follows.
19
After all, the German word ‘enthalten’ can mean both ‘contain’ and ‘entail.’
20
Cf. Wolff ’s definition of the possible as that from which nothing impossible follows [fliesst] (Dt.Log.
}1.35).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

concept C itself contains a contradiction.21 Conversely, if concept C contains a

contradiction—it contains, say, concept C1 and ~C1—then the proposition C is
instantiated entails a contradiction, for it entails that there is something that is
both C1 and ~C1. If this is correct, then a concept C is possiblyC instantiated if and
only if the proposition C is instantiated is possibleP. I propose therefore that we
interpret Wolff and Baumgarten as accepting the definitions of possibility and
necessity in terms of propositions, (PossibilityP) and (NecessityP).
These are definitions of what Baumgarten and Wolff call ‘in se’ or ‘an sich’
possibility, and they correspond to Leibniz’s category of ‘strict’ or ‘metaphysical’
possibility, of beings possible in ‘themselves’ or ‘in their own nature’:
Provided also that it is understood that necessity and possibility, taken metaphysically and
strictly, depend solely upon this question, whether the object in itself or that which is opposed
to it implies contradiction or not.22

Leibniz’s definition is more straightforward than Wolff ’s or Baumgarten’s because he

defines possibility and necessity in terms of ‘implying’ a contradiction. Since only
propositions (judgments, etc.) can imply anything, this suggests that possibility and
necessity are to be defined in terms of the possibility and necessity of propositions,
and that the possibility or necessity of the instantiation of a concept should be
assimilated to that model: concept C is possibly instantiated if and only if the
proposition C is instantiated is possible (i.e. it does not entail a contradiction).
Leibniz too accepts (PossibilityP) and (NecessityP).
Some readers might wonder what precise notion of ‘entailment’ is involved in
these definitions of possibility and necessity. While it is reasonably clear that for these
thinkers the relevant notion of entailment is syntactic rather than semantic, this
distinction is not precisely drawn in this period. Wolff analyzes ‘inferences’ [Schlüsse]
in terms of their agreement with the traditional Aristotelian syllogistic forms, which
are syntactically defined.23 Leibniz worked extensively on what he called a ‘universal
characteristic,’ a universal symbolic language for all reasoning. The symbolism of the
universal characteristic is so constructed that whether an argument is valid can be
mechanically checked, without attending to the meanings of the individual symbols

21
The other potential source for contradictory consequences of the proposition C is instantiated, aside
from contradictory content in C, would be if the instantiation of a C violated some other necessary truth. In
the context of logicist metaphysics, the main such example would be that the existence of a less than
maximally perfect possible world violates the necessary truth that God creates the most perfect possible
world. However, Wolff and Baumgarten both maintain that the proposition a less than maximally perfect
world exists is in se possible (possible in its own nature), but impossible in nexu with God’s perfect will. For
their theories of possibility in nexu and its role in their defense of contingency, see Dt.Met. }}574–5; AzDM
}}195, 197; and Meta. }}15–18, 102–5.
22
Th. }367. See also Th. }}44, 45, 228, 230–2, 234, 235; NE 497; AG 19, 44, 95, and 98. In this book I am
primarily concerned with what is possible and necessary ‘in itself ’ or in its own nature, not with possibility
in nexu.
23
See Dt.Log. }}4.165–72; esp. his emphasis on “förmliche Schlüsse.”
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

(which concepts they stand for). In terms of the contemporary distinction, this is the
idea of a purely syntactic definition of provability.24
The definitions of possibility and necessity given so far apply to whole propositions
(dicta); in contemporary terms, they give an account of de dicto modality. What is de
re possible and necessary for a being (ens) is determined by its essence. An essence is
a logically complex concept that determines what it is to be that being, and which
explains, at least partly, all of the other predicates of the being (which concepts it falls
under).25 Leibniz, Wolff, and Baumgarten identify the possibility of a being with its
essence.26 A being has a logically consistent essence. The predicates contained in its
essence, its essential predicates, are necessarily had by that being. For instance, a
triangle is necessarily three-sided and necessarily a plane figure. The predicates that
follow from a being’s essence, but are not contained in it, are its attributes. The
predicates of a being that do not follow from its essence are its accidents. Attributes
and essential predicates pertain necessarily to a being, and accidents pertain only
contingently. The essence constrains the range of accidental predicates a being can
have, because the accidents must be consistent with the essence. Consequently, any
predicate possessed by a being is at least partially explained by its essence: wholly
explained in the case of essential predicates and attributes, and partly explained in the
case of accidents.27 Essences of finite beings are incompletely determinate; they
underdetermine the accidents of beings, which are partly grounded in their essences
(which constrains the range of accidents beings can have) and partly grounded in the
(causal) relations beings bear to one another. Those very beings could (in the ‘in se’
sense of possibility) exist with different accidents.28

24
Cf. “Samples of the Numerical Characteristic” (AG 10–18, esp. 18), “On the General Characteristic”
(L 221–7, esp. 224–5), and “Two Studies in the Logical Calculus” (L 235–47); cf. P 42–3.
25
As I am using them, ‘predicate’ and ‘concept’ are co-extensive terms with different senses: every
concept is a predicate because it can be the predicate of some judgment, and every predicate is a concept.
One refers to a concept as a predicate in a context where that concept is being predicated of an object or is
one of the marks (constituents) of a larger concept. Thus, predicates are not linguistic entities (because
concepts are not). For more on Kant’s use of the term ‘Prädicat’ see Ch. 3.1.
26
Dt.Met. }35, Meta. }40, and NE 293. For more on Leibniz’s identification of the possibility of a thing
with its essence, see AG 21, L 146 f., and A II.i.390; for discussion, see Adams (1994, 136–8).
27
Dt.Met. }}33–8, 44, 175, 176; Meta. }}41, 42, 50, 52, 65, 195; and A VI.iii.574. See also Ont. }}143, 144,
153, and 168.
28
The question of the determinacy of possibilities is one on which Wolff and Baumgarten depart from
Leibniz. Leibniz famously held that all of the predicates that apply to an individual substance follow from
its complete individual concept, and no individual substance can exist with a different complete concept
than it actually has (AG 11, 30, 31, 40–1, 95, 98; L 226, 231–2; and NE 486). He distinguishes the essential
predicates of an individual substance from its contingent ones—both of which are contained in its
complete concept—by claiming that the former follow from a finite analysis of the concept, while the
latter require an infinite analysis. See “On Contingency” (AG 28–30) and “Primary Truths” (AG 30–4).
Cf. Mates (1986), 108–17; Sleigh (1990); and Adams (1994), 25–30. In my discussion of Leibniz, therefore,
whenever I talk about ‘entailment’ or ‘demonstration’ I mean entailment in a finite number of steps and
demonstration by finitely many steps. Without this qualification, Leibniz’s view is that every truth (even
contingent ones) about an individual substance can be demonstrated from its complete concept (though it
may require an infinite demonstration), and the negation of every truth (even contingent ones) about an
individual substance entails a contradiction (although it may require an infinite demonstration to uncover
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

The contemporary reader might be puzzled by talk of the logical consequences

of predicates; after all, one might think, only propositionally structured entities stand
in logical relations, and predicates are not propositionally structured (i.e. they cannot
have truth values). However, this is not how thinkers in this period see matters.
Leibniz, Wolff, Baumgarten, and Kant frequently talk about non-propositional
entities—especially predicates (concepts)—entailing one another. This is not an
insuperable barrier to dialogue. We can translate the earlier idiom into the contem-
porary one by understanding talk of logical relations between concepts (predicates)
as shorthand for talk of logical relations between propositions that predicate those
concepts of objects. For instance, we can say that the predicate <triangular> entails
the predicate <trilateral> because, for any object x, the proposition x is triangular
entails x is trilateral.
In some texts Leibniz defines necessity slightly differently:
(NecessityDem) For any proposition p it is necessary that p if and only if there is a
(finite) demonstration of p from identities and definitions.29
One might wonder whether these definitions of necessity are equivalent. In particu-
lar, one might wonder whether, if there is a demonstration of p from identities and
definitions, ¬p entails a contradiction. I will assume that these definitions of necessity
are equivalent. If this is correct, then we can also define possibility in terms of
demonstration, using the interdefinability of possibility and necessity:30
(PossibilityDem) For any proposition p, it is possible that p if and only if there is
no (finite) demonstration of ¬p from identities and definitions.
If I am correct, this definition of possibility is equivalent to the earlier definition of
possibility in terms of entailment. Sometimes it will be convenient to adopt the
definitions of these modal notions in terms of demonstration.
It is crucial to note, though, that the identities and definitions that can be appealed
to in a demonstration of a putatively necessary proposition (or the negation of a
putatively possible one) are limited to the identities and definitions of the concepts
involved in that very proposition. This is because we are concerned with what is
necessary (and what is possible) ‘in its own nature.’ In other words, we need to restate
these principles slightly:
(NecessityDem) For any proposition p it is necessary that p if and only if there is a
(finite) demonstration of p from identities and definitions of the constituents of p.

this contradiction). For critical discussion of Leibniz’s views on the determinacy of possibilities, see
Mondadori (1973), (1975), (1985); Mates (1986), 137–51; Sleigh (1990); Adams (1994), 53–110;
Nachtomy (1997), 51–103; and Cover and O’Leary-Hawthorne (1999), 87–142.
29
“On Contingency” (AG 28–30), “Primary Truths” (AG 30–1), and “The Source of Contingent
Truths” (AG 98–101). Cf. L 226, AG 21, and 95. Cf. M. Wilson (1990).
30
□p $ ¬◇¬p. Cf. Meta. }101.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

(PossibilityDem) For any proposition p, it is possible that p if and only if there

is no (finite) demonstration of ¬p from identities and definitions of the constitu-
ents of p.
If there is some being or beings whose concepts are not involved in p, but from whose
definition there is a demonstration that p, p can still count as ‘in se’ contingent (not
necessary) as long as there is no demonstration of p from the definitions of the very
concepts contained in p.31 Translating this point back to the earlier principle
(PossibilityP) we can impose the requirement that a proposition p entails q only if
this relation is grounded in the syntactic structure (see previously) of both proposi-
tions and the concepts contained in them (not in any ‘outside’ concepts).32
The principle of non-contradiction (PNC)—that no being is both A and not-A—is
the principle that governs all of possibility, all beings, and hence is the highest
principle of ontology. Since ontology is the highest and most general science, the
PNC is the highest principle of all sciences and all thinking whatsoever.33 Possibility,
for Leibniz, Wolff, and Baumgarten, ultimately reduces to logical self-consistency.
The space of possibilities is a logical space, governed by logical relations of compati-
bility and entailment.34 This also means that the space of possibilities is governed by a
form of the principle of sufficient reason. If it is possible that p, there is an explanation
of why p is possible: it does not entail a contradiction. Likewise for necessities: if it is
necessary that p, this is so in virtue of the fact that ¬p entails a contradiction. There are
no brute, ungrounded facts about possibility or necessity.
Before continuing, I want to pause and note that logicism is a deeply plausible view
about the metaphysics of modality. One core idea behind logicism is that whenever
two predicates are incompatible, there is a reason for this incompatibility, and this
reason lies in the logical structure of the predicates in question. For example, many
thinkers in the early modern period held that thinking and extension are incompat-
ible; it is impossible for an extended being to think. This much is familiar from
Descartes’ Sixth Meditation. But whereas Descartes appeals to our intuitive know-
ledge of a brute incompatibility between the natures of thought and extension,
logicism holds that wherever there is such an incompatibility, there is a logically
sufficient explanation. On the logicist view, the natures of thought and extension

31
This is crucial to secure the ‘in se’ possibility of non-actual worlds. For if it is necessary that God
creates the best possible world (which, in some contexts, all three of these thinkers are willing to admit)
then there is a demonstration of this truth; if this demonstration can be appealed to in any demonstration,
then there will be a demonstration, of any other possible world, that it is not actual. See previously for the
distinction between ‘in se’ and ‘hypothetice’ possibility.
32
This also means that ‘in se’ necessity does not obey the K rule: if □p & □(p⊃q) then □q. The
antecedent might be true, while the concepts contained in q are not sufficient to demonstrate q ‘on their
own.’
33
Dt.Met. }10, Meta. }7, and AG 19.
34
Cf. Adams (1994), 12–15; Mates (1986), 105–7; and Pichler (1910), 32–48. For a discussion of texts in
which Leibniz suggests that there are contingent facts about possibles qua possible, i.e. facts about
possibility not governed by the principle of non-contradiction, see Adams (1994), 30–4.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

must be ultimately composed of logically incompatible components, although we

may not possess sufficient insight to discover them.
Logicism should not be confused with the once widespread view that analyticity
and necessity are co-extensive.35 Although it is a matter of some controversy in
contemporary philosophy what exactly analyticity is, and, indeed, whether there is an
analytic/synthetic distinction at all, it is relatively uncontroversial that, if there is an
analytic/synthetic distinction, analyticity is a semantic property of sentences.36 Logi-
cism, on the contrary, is a view about the modal status of propositions, not sentences;
although these early modern thinkers do not draw this distinction in those terms,
they would acknowledge that logicism is not a view about bits of human language.37
Instead, logicism holds that a proposition is necessary just in case its truth is
grounded in the natures or essences of the objects the proposition is about. Logicism
is not a view about the meanings of propositions, if meanings are understood as
something that we are in a position to grasp in virtue of semantic competence with
our language.

1.3. Logicist Epistemology

Having briefly outlined the logicist modal metaphysics, we now turn to the account
given by Leibniz, Wolff, and Baumgarten of how we attain knowledge of possibility
and necessity. In their modal epistemologies, Wolff and Baumgarten closely follow
Leibniz, especially the theory presented in “Meditations on Knowledge, Truth, and
Ideas,” first published in 1684 and recommended to Wolff by Leibniz in their
correspondence.38 In this essay, Leibniz classifies concepts according to four prop-
erties: clarity, distinctness, adequacy, and whether one’s grasp of them is intuitive or
symbolic. Concepts have these properties relative to minds; a concept that is dis-
tinctly understood by one mind might be indistinctly understood by another.39
I clearly understand a concept just in case I know whether a given object falls
under the concept or not; for instance, I clearly understand the concept <red>
because I can tell, of a visible object, whether it is red or not.40 I distinctly understand
a clear concept if, in addition, I can give necessary and sufficient conditions for an
object to fall under it; I understand the concept <red> only indistinctly, or obscurely,

35 36
E.g., Ayer (1936). For an overview of these issues, see Boghossian (1996).
37
Cf. the discussion of the ontological status of logical entities like propositions in Mates (1986), 47–68.
38
Cf. Wolff ’s very similar epistemology of ideas in Dt.Log. }}1.1–23 and Log. }}78–95. Leibniz also
refers to “Meditations” in other writings available to eighteenth-century readers, e.g., NE 297. Cf. Wundt
(1945), 140, 153–8; and Honnefelder (1990), 304–5.
39
More perspicuously: token concepts (modes of individual minds) have these properties simpliciter.
When we talk about ‘the’ concept of X, though, we are talking about what all token concepts of X have in
common in virtue of being token concepts of X. Since the clarity, distinctness, etc. of token concepts differ
from mind to mind, we need to specify which mind, and thus which token concept, we are talking about.
For instance, the concept <red> is indistinct for me because my token concept of <red> is indistinct.
40
Cf. Dt.Log. }}1.9–12.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

because I cannot give a set of marks that are necessary and sufficient for something to
be red.41
To understand the notion of adequacy we need to understand more about the
structure of concepts. A concept is either simple (if it has no parts) or complex (if it
has parts).42 A complex concept is composed of other concepts, called the ‘marks’ of
that concept, which it is said to ‘contain.’ The marks of one concept can themselves
contain other concepts, and concept containment is transitive: if mark C1 is con-
tained in (is a mark of) concept C2, and C2 is contained in (is a mark of) concept C3,
then C1 is contained in (is a mark of) C3. Since concept containment is transitive, the
marks of a concept may not be immediately or directly contained in it; finding its
marks may require unpacking the immediate constituents of the concept, and their
constituents, etc.
Leibniz defines the adequate understanding of concepts as follows:
One has distinct knowledge of an indefinable notion, since it is primitive, or its own mark, that
is, since it is irresolvable and is understood through itself and therefore lacks requisites [ . . . ]
[W]hen everything that enters into a distinct notion is, again, distinctly known, or when
analysis has been carried to completion, then knowledge is adequate.43

The ultimate constituents are atomic concepts, which are “understood through
themselves,” that is, they are not composed of further concepts. I adequately under-
stand a complex concept just in case I can decompose it into its constituent concepts,
and can resolve those concepts into their constituent concepts, and so on, to the
atomic marks that compose the original concept, which are by definition adequate.44
When thinking with complex concepts, we typically do not hold in mind simul-
taneously all of the primitive constituents of the concept and their complex manner
of arrangement. Instead, we mentally assign some symbol to one of the concepts that
compose the larger concept. In such cases, our grasp of the concept is symbolic. If we
mentally distinguish all of the constituents of the concept, down to its primitive
constituents and the complex manner in which they compose the concept, our grasp
of the concept is intuitive. All thought about primitive concepts is intuitive, because
they have no conceptual constituents.45
Given the logicist conception of possibility from section 2, a concept is possibly
instantiated if and only if it contains no contradictory marks. When we have an
intuitive and adequate grasp of a concept, we have insight into its logical structure
and we can see whether there are any contradictions present in it. Since primitive
concepts are logically atomic, they are also logically independent; primitive concepts
stand in no relations of logical entailment with one another. Therefore, the only

41
See ‘DM’, }8 (AG 41); NE 262, 293–5, and 346; as well as Dt.Log. }1.15.
42
See Dt.Log. }1.17. 43
AG 24.
44
Cf. Mon. }35; NE 120; L 160, 167; and Dt.Log. }1.16. 45
AG 25.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

source of contradiction within concepts is their complex composition out of primi-

tive parts.
This epistemology of concepts is related to Leibniz’s distinction between nominal
definitions and real definitions. A nominal definition provides a set of marks that are
individually necessary and jointly sufficient for an object to fall under it. I distinctly
understand a concept just in case I can give a nominal definition of it. A real
definition, by contrast, shows why it is possible for objects to fall under it; a real
definition decomposes the concept into its logically atomic constituents and shows
that there is no contradiction among them. As Leibniz writes in “Meditations”: “we
also have a distinction between nominal definitions, which contain only marks of a
thing to be distinguished from other things, and real definitions, from which one
establishes that a thing is possible” (AG 26).46 Only if we have an intuitive and
adequate understanding of a concept can we give a real definition of it and have
insight into the possibility of an object falling under it.
This, in turn, is related to Leibniz’s distinction between a posteriori and a priori
knowledge.47 In a posteriori knowledge, gained paradigmatically through experience,
we know only that something is the case. In a priori knowledge we know the reasons
why.48 If we experience an object that instantiates it, then we can know that a
given concept is possibly instantiated.49 However, this can only explain our
knowledge of actualized possibilities, and it does not give us a priori knowledge
of possibility: we know only that the concept is possibly instantiated, because
actually instantiated, but not why this is the case. To know why a concept is
possibly instantiated we need a real definition of that concept; we need to analyze
the concept into its primitive constituents and see that there is no logical
inconsistency among its marks. In this case, we have adequate intuitive under-
standing of the concept.50 A real definition of a concept also gives us knowledge
of non-actual possibilities; whether or not the concept is actually instantiated,
if we know that it is logically consistent then we know that it is possibly
instantiated.51
This conception of a priori knowledge as knowledge ‘from the grounds’ should be
distinguished from the more familiar ‘justificatory’ definition of a priori knowledge,
as follows:
(A PrioriG) S knows a prioriG that p if and only if G is a sufficient ground of the
fact that p and S derives p from G.
(A PrioriJ) S knows a prioriJ that p if and only if (i) S knows that p and (ii) the
justification in virtue of which S knows that p does not come from experience.

46
Wolff draws a similar distinction in Dt.Log. }}1.41–9; see also Log. }}141, 152, 179, and 191.
47
Cf. Dt.Met. }}77, 368, and 372.
48
For more on Leibniz’s use of the terms ‘a priori’ and ‘a posteriori’ see Adams (1994), 109–10.
49
Cf. Dt.Log. }1.34. 50
AG 26 and NE 294. 51
See NE 162, 266, 254, 311, 352, and 346.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

This distinction will be a continuing theme in this book, for, as I will argue, Kant
often has the ‘from the grounds’ conception in mind when he discusses a priori
cognition (Erkenntnis) and knowledge (Wissen).52 These are only preliminary char-
acterizations; the details of the definition of a prioriG knowledge will depend on what
kind of ground is in question (e.g., efficient-causal grounds, or the formal grounds
contained in a thing’s essence). Leibnizian real definitions provide a prioriG and a
prioriJ knowledge; since our grasp of a real definition is purely intellectual, it does not
depend on sensory experience (a prioriJ) and it acquaints us with the grounds of the
possibility of the concept’s object (a prioriG). From now on, unless otherwise noted,
by ‘a priori’ knowledge I mean a prioriG knowledge.53
Leibniz makes this elaborate classification of concepts in “Meditations” in order to
diagnose, and recommend a cure for, a major obstacle to cognition: errors about what
is possible. The paradigm instance of a modal error, for Leibniz, is taking a covertly
contradictory concept to be a concept of a possible object. Leibniz’s example of a
concept with a concealed contradiction is that of ‘the fastest motion,’ an apparently
consistent concept, hence apparently a concept of a possible object.54 Leibniz puts
such importance on rooting out modal error because he holds that we can derive no
knowledge from self-contradictory concepts. He writes: “For we cannot safely use
definitions for drawing conclusions unless we know first that they are real definitions,
that is, that they include no contradictions, because we can draw contradictory
conclusions from notions that include contradictions, which is absurd.”55 Leibniz’s
reasoning appears to be as follows. If S is an impossibly instantiated concept, then it
contains a contradiction between two of its constituents, call them A and ~A. If any
proposition involving S as the subject concept were true, then the proposition S is A
would be true, because in this judgment we merely predicate one of S’s components
of S. But, by parity of reasoning, S is ~A would be true. Hence, a contradiction arises if
we admit as true any propositions with contradictory subject concepts. Because
propositions with self-contradictory concepts cannot be true, propositions with self-
contradictory concepts cannot be known. In the case of a prioriJ knowledge, where we
cannot rely on experience to show that our concepts are actually instantiated (hence
possibly instantiated), we need real definitions of our concepts: proofs that they are
logically consistent, hence possibly instantiated. Since metaphysical knowledge is para-
digmatically a prioriJ, knowledge in metaphysics requires a prioriJ knowledge of the
possibility of the instantiation of its most basic concepts (e.g., substance, force, etc.)56

52
See Smit (2009).
53
For the Wolffian version of these doctrines, see Dt.Met. }372 and Log. }498.
54
Leibniz uses the same example, for the same purpose, in his correspondence with Elizabeth (AG 238)
and DM }23.
55
AG 25.
56
For instance, in the case of the ontological argument (see the rest of this chapter), we must first prove that
the concept <most perfect being> is possibly instantiated. As Leibniz writes in “Meditations,” “the fact that we
think about a most perfect being is not sufficient for us to assert that we have an idea of it” (AG 25).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

In fact, according to the view Leibniz defends in “Meditations,” contradictory

concepts are not genuine concepts. He writes:
However, at first glance we might seem to have the idea of a fastest motion, for we certainly
understand what we say; yet we certainly have no idea of impossible things. And in the same
way, the fact that we think about a most perfect being is not sufficient for us to assert that we
have an idea of it.57

When we think about a concept that contains a hidden contradiction, e.g., ‘the fastest
possible motion,’ we take ourselves to be entertaining a thought with content. But we
are mistaken; for Leibniz, concepts with concealed contradictions are not really
concepts at all. To show that we have a concept of the most perfect being, we must
show that this concept is a concept of a possible being.58
We saw earlier that the logicists hold that all necessary truths can be demonstrated
from identities and definitions:
(NecessityDem) For any proposition p it is necessary that p if and only if there is a
(finite) demonstration of p from identities and definitions of the constituents of p.
The definitions in question are real definitions. If we allowed nominal definitions,
like ‘Julius Caesar is the individual that figures in most of the examples in this
book,’59 this principle would entail that it is necessarily the case that Julius Caesar
figures in most of the examples in this book, which is absurd. So the principle should
be formulated as:
(NecessityRealDef) For any proposition p, it is necessary that p if and only if
there is a (finite) demonstration of p from identities and real definitions of the
constituents of p.
The domain of the possible, for Leibniz, Wolff, and Baumgarten, is a logically
structured domain of logically consistent concepts of possible objects, propositions
about those possible objects, and their logical relations. By analyzing concepts into
their more fundamental constituents, we gain insight into the logical structure of
modal reality and attain a priori knowledge of possibility. A domain of objects is
intelligible, for the logicists, to the extent that a priori knowledge of that domain is
possible. Whether some domain is a priori knowable is a metaphysical matter:
is there a sufficient ground for all facts about objects in this domain?60 Since the
domain of possibility is a logical domain, for the logicists, possibility is fully a priori

57
AG 25.
58
Leibniz’s views on contradictory concepts, and the truth conditions of propositions involving them,
are less fixed than he presents them in “Meditations.” See Mates (1986), 67–8.
59
This is a nominal definition because it provides necessary and sufficient conditions for being Julius
Caesar. It does not give the meaning of ‘Julius Caesar’ (so descriptivism about proper names is not intended
here).
60
Mon. }28 (AG 216–17) and AG 209; cf. Dt.Met. }77.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

knowable. Although we may not have the cognitive resources to analyze all concepts
and all objects and determine their essences, the a priori intelligibility of the space of
possibility consists in the existence of logically sufficient explanations of why con-
cepts are possibly instantiated and why propositions are possibly true.
Readers may have noticed that, whereas in section 2 I phrased everything in terms
of ‘beings’ (Dinge, entia), Wolff ’s and Baumgarten’s term for the highest genus of
ontology (the possible as such), I have switched in this section to talking about
‘objects.’ The reason is that in this section we have been primarily concerned with
concepts of beings and the relation between beings and the concepts they fall under.
I am using the term ‘object’ to refer to beings insofar as they fall under concepts, that
is, insofar as they constitute the extensions of concepts. In contemporary terms, they
are admissible values of bound first-order variables. We have seen that the principle
of non-contradiction holds for all beings (Dinge, entia). Consequently, our quantifier
expressions should be restricted to beings, for otherwise the principle of contradic-
tion would have counter-examples (i.e. there would be objects that are not beings,
Undinge, and for which the PNC would not hold). Conversely, the science of
ontology (the science of all beings insofar as they are beings) presupposes that we
can quantify over all beings, so every being is an object (of quantification). Hence-
forth, I will often refer to beings as ‘objects’; by calling them ‘objects’ I mean to
remind the reader that they are objects in the Quinean sense: admissible values of
bound first-order variables. I will sometimes refer to them as q-objects.61
This also explains an awkwardness in my characterization of ‘being’ from section
2. I defined a ‘being’ as ‘any possible entity whatsoever,’ but if the reasoning of the
previous paragraph is correct, then no definition of being by genus (e.g., entity) and
differentia (e.g., possible) is possible; there is no wider genus of which ‘being’ is a
species. So my characterization of beings as possible entities should be not under-
stood as a definition but as an indication of how wide the class of beings is. Whether a
non-circular definition of ‘being’ (as a count noun) is possible is obviously a difficult
question in its own right; I will not attempt to address it here.

1.4. Ontotheism
All of the logicists, and at least the pre-Critical Kant, share the orthodox view that
God exists necessarily.62 Julius Caesar exists only contingently, like all finite created
substances, but God exists necessarily. What accounts for this difference? Logicism,
recall, is the view that:
(NecessityRealDef) For any proposition p, it is necessary that p if and only if there is a
(finite) demonstration that p from identities and real definitions of the constituents of p.

61
See Quine (1948). Whether concepts themselves are beings and/or objects is a complicated matter;
see previous notes for a discussion of the ontological status of concepts.
62
For Kant’s pre-Critical view of divine necessity, see Ch. 5; for his Critical view, see Ch. 9.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

Since God exists necessarily, there is a demonstration of his existence from his real
definition (and perhaps from the real definition of <existence>). God’s essence either
contains existence as a component, or contains predicates that logically entail his
existence. I will abbreviate this by saying that God’s existence is logically grounded in
his essence.
Since this point will be crucial for the next two chapters, it warrants further
development. As we have seen, the logicists hold the natural view that objects
necessarily have certain predicates in virtue of their essences (essential properties,
and attributes), while their contingent predicates are merely compatible with their
essences (accidents). The logicists extend this model of the modal status of predicates
to the case of existence. God exists necessarily in virtue of the fact that <existence> is
logically grounded in his essence (either by being contained in it, or logically entailed
by it). God exists necessarily because his essence alone makes it the case that he exists.
By contrast, Julius Caesar exists contingently, if he exists at all, because <existence> is
not logically grounded in his essence. Caesar exists contingently because his essence
alone does not make it the case that he exists.
This is a commitment that each of our logicists endorsed:
Leibniz: For if there is reality in essences or possibles, or indeed, in eternal truths, this reality
must be grounded in something existent and actual, and consequently, it must be grounded in
the existence of the necessary being, in whom essence involves existence, that is, in whom
possible being is sufficient for actual being.63
Wolff: God exists through his essence, or his existence is essential.64
Baumgarten: From God’s possibility, it is valid to draw the conclusion that he exists, i.e. his
existence is sufficiently determined by his very essence.65

Descartes and Spinoza also agreed that God exists in virtue of his essence.66
Kant introduced the term ontotheology to describe the part of theology that
attempts to determine the predicates God has merely as a possible being (ens).
Ontotheology, he says in his theology lectures, “considers God merely through
concepts of possible beings in general [mögliche Dinge überhaupt]” (Volck.RT,
Ak. 28: 1142).67 The divine predicates discussed in ontotheology include his possi-
bility, essence, necessity, substance, simplicity, immateriality, etc.68 Consequently, an

63
Mon. }45. This text is important because Leibniz asserts that God exists necessarily in virtue of his
essence “involving” existence, even though he does not give the ontological argument in the Monadology;
even where Leibniz does not use the ontological argument to prove God’s existence, he still uses the
underlying metaphysics to explain God’s necessary existence.
64
TN }27. Cf. Dt.Met. }}929–30 and Ont. }308. 65
Meta. }820.
66
See CSM II: 46–7 and Ethics Ip7d (a substance’s “essence necessarily involves existence, or [sive] it
pertains to its nature to exist”). See also Eberhard: “a being is necessarily actual if the ground of its actuality
is contained in its essence” (Ak. 28: 561). I discuss Descartes in Chapter 2.2, but I forgo further discussion
of Spinoza in this book.
67
Cf. the virtually identical formulations in Pöl.RT (Ak. 28: 1002 f.). Cf. A632/B660, MV (Ak. 28: 451),
MK2 (Ak. 28: 777), and MK3 (Ak. 28: 824).
68
Danz.RT (Ak. 28: 1252).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

ontotheological argument for God’s existence is one that purports to prove the
existence of God merely using modal concepts. Kant refers to ontotheological
arguments using his more familiar coinage: ontological arguments.69 Anselm, the
discoverer of onto(theo)logical arguments, is also described as the founder of
ontotheology.70
It is often overlooked that Kant sometimes uses ‘ontological argument’ in a broad
sense71 to include two very different kinds of onto(theo)logical argument: (i) a
prioriG ontological arguments that infer from God’s possibility (his essence) to his
existence, and (ii) a posterioriG ones, which infer from the possibility of other things
to God’s existence.72 Both kinds of onto(theo)logical argument are a prioriJ because
they do not depend upon experience. This broad sense of ‘ontological argument’
corresponds to the broad sense of ‘ontotheology’ defined above: theology using
concepts of possible beings as such. In this broad sense of ‘ontotheology’ Kant
himself is engaged in ontotheology in Beweisgrund and gives an a posterioriG
argument he himself calls “ontological.”73 Throughout his philosophical writings
he rejects type (i) arguments because they are based on a false metaphysical view
(God exists in virtue of his essence). His attitude to type (ii) arguments is more
complex, and it will take me the rest of this book to explore it completely.
The term ontological argument has had a significant career after Kant, and has
come to refer specifically to a prioriG (type (i)) arguments.74 In fairness to the
tradition, Kant often uses that term (and ‘ontotheology’) in the more specific
sense.75 To avoid confusion, I will follow this convention; henceforth, ontological
argument will refer only to a prioriG arguments that purport to derive God’s existence
from his essence. Likewise, I will use ontotheism to refer to the specific ontotheolo-
gical doctrine that God exists in virtue of his essence. Ontotheism, as I will use that
term, is the view that there is an a prioriG ontological argument for the existence of
God. It is not a commitment of ontotheism, however, that, given the limitations of
our intellect, we can ever give an ontological argument for the existence of God, only
that there is one to be given.
Kant rejected ontotheism and (a prioriG) ontological arguments almost from the
moment he began publishing philosophy. In his 1755 Nova dilucidatio he writes:
To say that something has the ground of its existence in itself is absurd [ . . . ] I find, indeed, the
view repeatedly expressed in the teachings of modern philosophers that God has the ground of
His existence in Himself. For my part, I find myself unable to support this view [ . . . ] I know

69 70
MV (Ak. 28: 454) and ML2 (Ak. 28: 599). Volck.RT (Ak. 28: 1143).
71 72
Logan (2007) is a notable exception. OPG, Ak. 2: 155.
73
OPG, Ak. 2: 160, 162. See also Refl. 6027 (Ak. 18: 427).
74
Though it is not clear that it applies to the arguments discussed in Malcolm (1960), Plantinga (1967),
82–94, and Plantinga (1974), 197–221.
75
MV (Ak. 28: 454), ML2 (Ak. 28: 598–9), Pöl.RT (Ak. 28: 1027), Volck.RT (Ak. 28: 1173), and Danz.RT
(Ak. 28: 1260).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

that appeal is made to the concept of God; and the claim is made that the existence of God is
determined by that concept. (ND, Ak. 1: 394)76

Variants of this claim—that no being can exist in virtue of its essence or its possibility—
are found throughout the pre-Critical and Critical writings.77 I will argue in sections 5
and 6 of this chapter that Kant’s famous objection that “existence is not a predicate or a
determination of a thing” (Ak. 2: 72) entails that nothing whatsoever can exist in virtue
of its essence. This same claim is repeated almost word for word in CPR and through-
out Kant’s lectures on metaphysics and theology.78 Kant’s rejection of ontotheism is
one of his most consistent doctrines. Given the historical continuity of this Kantian
doctrine, it will be appropriate, in reconstructing Kant’s arguments for it, to consider
both Critical and pre-Critical texts, although the historical context of this chapter is
Kant’s pre-Critical Auseinandersetzung with logicism.
Throughout these writings, Kant levels many objections to specific ontological
arguments, but my aim in this chapter (and Chapter 2) is to reconstruct his reasons
for rejecting (a prioriG) ontological arguments as such, that is, his reasons for
rejecting ontotheism tout court. Getting the level of generality of Kant’s rejection of
ontological arguments right is important for several reasons. First, it affects what will
count as success on Kant’s part. He does not merely claim that particular ontological
arguments of Leibniz, Baumgarten, etc. are invalid or unsound (though he does also
claim this); he claims that ontological proofs as such are impossible. Kant raises
various objections to specific ontological arguments, but those are not my concern
here;79 my concern is reconstructing Kant’s objection to ontological arguments as
such. It is also important because it shows that Kant’s ultimate objection to onto-
logical arguments is not directed to them as arguments. Kant’s claim is not that the
premises of an ontological argument do not validly entail its conclusion, although his
view does entail that. Nor is Kant’s point the epistemic point that ontological
arguments fail to give sufficient reasons to accept that God exists, although his
view does entail that. Kant’s claim is that ontological arguments presuppose a false
metaphysical view about the source or ground of necessary existence, what I am
calling ‘ontotheism.’80 As Kant puts it succinctly in his lectures on metaphysics, “God
himself cannot know his own existence through concepts” (MK2, Ak. 28: 784). Even

76
Cf. MV: “no intellect can have insight into the absolute necessity of a being” (Ak. 28: 455).
77
MH (Ak. 28: 13, 55, 131, 133–4), MV (Ak. 28: 455), MK3 (Ak. 28: 824), MM (Ak. 29: 814), and Pöl.RT
(Ak. 28: 1027).
78
ML1 (Ak. 28: 313), MV (Ak. 28: 413, 455), ML2 (Ak. 28: 598), MK2 (Ak. 28: 783), and MM (Ak. 29:
822).
79
For instance, Kant points out at A598/B624 that Leibniz’s (and, following him, Wolff ’s and
Baumgarten’s) proof that <ens realissimum> is logically consistent does not suffice to prove that it is
possibly instantiated. Cf. Ak. 28: 455.
80
Kant introduces his discussion of existence in Beweisgrund by writing that the concept of existence
must be treated with special precision because in the case of necessary existence “hat eine subtilere
Nachforschung aus einem unglücklich gekünstelten, sonst sehr reinen Begriff irrige Schlüsse gezogen,
die sich über einen der erhabensten Theile der Weltweisheit verbreitet haben” (Ak. 2: 71).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

God cannot know his own existence through concepts, because God’s existence is
neither entailed by nor contained in his essence.81
That Kant’s fundamental target is the ontotheist explanation of divine necessity is
often overlooked, with the result that Kant’s objection to ontological arguments is
misunderstood.82 In Beweisgrund Kant writes:
That of which the opposite is impossible in itself is absolutely necessary. This is certainly a
correct nominal definition. But if I ask: upon what does the absolute impossibility of the non-
being of a thing depend? then what I am looking for is the real definition; this alone can serve
our purpose. (OPG, Ak. 2: 81)

An object exists necessarily if and only if the non-existence of that object is impos-
sible. Kant is not rejecting this principle, but pointing out that it is not informative. It
is a nominal definition—it provides necessary and sufficient conditions for necessary
existence—but it is not a real definition.83 Kant’s point is that this definition is not
explanatory: it does not tell us what it is for a necessary being to exist necessarily, nor
does it tell us, if there is a necessary being, why that necessary being is necessary (the
ground of its necessary existence). Immediately after this passage, Kant summarizes
his objection to the ontological argument, and then writes:
In the final reflection of this work all of this will be made more convincing, by clearly
explaining the untenability of the mistaken view that absolutely necessary existence could be
explained by means of the law of contradiction. (OPG, Ak. 2: 82)

“The mistaken view” is the ontotheist view that God exists necessarily in virtue of
his essence containing or entailing his existence. This view explains God’s neces-
sary existence via the law of non-contradiction because, in its logicist version,
God’s existence is a logical consequence of his essence; if God did not exist, this
would be a contradiction.84 That Kant’s fundamental target in this work is a view
about what explains God’s necessary existence is further evidenced by the fact that
he goes on to give an account of what it is for God to exist necessarily (his non-
existence would cancel all real possibility) and to argue that there is a ground of
God’s necessary existence: he grounds all real possibility.85 I explore both points in
detail in Chapters 4 and 5.
Likewise, Kant’s discussion of the ontological argument in the CPR does not begin
as a criticism of them as arguments that God exists, but as explanations of why God

81
This is clear in ND (Ak. 1: 394, 396), as well as the Herder metaphysics lectures (Ak. 28: 13, 14).
82
For instance, Proops (2015)’s reconstruction of Kant leaves him without an objection to ontological
arguments as such, but only individual objections against individual arguments. Kant’s claim, however, is
more sweeping: ontological proofs as such are impossible (A592/B620).
83
For the distinction between nominal and real definitions, see JL }106 (Ak. 9: 143–4) and Ch. 9.2,
where it is discussed in more detail.
84
Cf. Baumgarten: “If God were not actual, the principle of [non-]contradiction would be false” (Meta.
}824).
85
Cf. OPG (Ak. 2: 82–3).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

exists necessarily. In the immediately preceding section of the Ideal of Pure Reason,
“The grounds of proof of speculative reason for inferring the existence of a highest
being,” Kant describes reason as assuming the existence of an absolutely necessary
being and then casting about for an explanation of why that being necessarily exists.
He first argues that reason has no grounds for regarding the concept of an unlimited
being—one possessed of every reality—as a necessary being, and, conversely, no
grounds for rejecting limited beings as candidates for necessary existence. His discus-
sion of the ontological argument in section 4 (“On the impossibility of an ontological
proof of the existence of God”), therefore, begins with reason already having formed
the concept of a necessary being, and inferred (illegitimately) its existence.86
For now, I am identifying logicism as the view that:
(NecessityRealDef) For any proposition p, it is necessary that p if and only if there is
a (finite) demonstration that p from identities and real definitions of the constitu-
ents of p.
When combined with the assumption that God necessarily exists, this entails that:

(1) There is a finite demonstration of the proposition God exists from the real
definition of God (which states his essence) and existence.
This means that logicism is committed to a specific form of ontotheism: the relation
between the real definition of God (which states his essence) and the proposition that
God exists is logical entailment. Ontotheism is a more general position than logicism.
For instance, Descartes is an ontotheist (he accepts that God exists in virtue of his
essence) but he is not a logicist.87 In this chapter and the next I am focusing on Kant’s
objection to ontotheism in general; if successful, this would also refute logicism.

1.5. Possibilism
Ontotheists hold that God exists necessarily in virtue of the fact that his existence is
grounded in his essence. They are committed to the following in-virtue-of claim:
(1) (□God exists) in virtue of the fact that (God’s existence is grounded in his
essence).
This is an instance of the general principle that objects have predicates necessarily in
virtue of those properties being logically grounded in their essences. That general
principle entails that:

86
A592–3/B620–1. Kant connects the ontological argument to the issue of what explains necessary
existence in a number of different texts. Cf. the metaphysics lectures: MH (Ak. 28: 131), MV (Ak. 28: 418),
MvS (Ak. 28: 498–500), ML2 (Ak. 28: 556–8, 599), and MK2 (Ak. 28: 724, 783).
87
That is one reason Descartes will be important to my argument; the other is that Descartes’
ontological argument was deeply influential on Leibniz.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

(2) a’s essence grounds a’s being F ⊃ □(Fa).

In other words, any predicate grounded in an object’s essence is a predicate the object
necessarily has. Furthermore, the consequent of this conditional obtains in virtue of
the antecedent; the object in question necessarily has the predicate because (in virtue
of the fact) that predicate is grounded in its essence.
However, this conception of the relation between necessary predication and
essence runs into problems if we make the assumption that necessarily every object
exists, i.e.
(3) □(x)(exists(x)).
Note that this is not the claim that every object de re necessarily exists; it is merely the
claim that, necessarily, the quantifier ‘for all’ ranges only over existing objects. It is an
immediate consequence of the most natural way of defining the predicate ‘exists’ in
terms of the universal quantifier:
(4) exists(x) = ¬(y)(y6¼x).88
Intuitively, (4) says that an object exists just in case it is not the case that no object is
identical to it. If we define ‘exists’ in this fashion, then the sentence inside the scope of
the necessity operator in (3) is a logical truth; since logical truths are necessary, (3) is
true. If we define the predicate ‘exists’ in this fashion, (2) entails:
(5) a’s essence grounds a’s being F ⊃ □(exists(a)).89
In other words, any object with an essence sufficient to ground some of that object’s
predicates is an object that necessarily exists. This entails that either no object other
than God has essential predicates, or that every object with essential predicates
necessarily exists. Neither is an acceptable consequence.
The ontotheist might try to escape this consequence by modifying (2) as follows:
(6) a’s essence grounds a’s being F ⊃ □(exists(a) ⊃ Fa).
But notice that the consequent of this conditional is trivial where ‘exists’ replaces ‘F’:
(6*) a’s essence grounds a’s existence ⊃ □(exists(a) ⊃ exists(a)).

88
I define the predicate ‘exists’ using the universal quantifier rather than the so-called ‘existential’
quantifier (the two are interdefinable). I go on to argue that the ontotheist should understand that
quantifier not as expressing ‘there exists an F’ but as expressing ‘there is an F.’ Thus, introducing ‘9xFx’
at this stage would only serve to confuse the issue.
89
Because □(Fa) entails that □(¬(y)(y6¼a)), which, by (4), is equivalent to □(exists(a)). The inter-
mediate step—from □(Fa) to □(¬(y)(y6¼a))—fails in a ‘free logic’ (Nolt 2014). However, I think that the
ontotheist who adopts a free logic will require a theory structurally isomorphic to the one I argue for in the
main text; for more, see the supplementary article “Free Logic and Ontotheism” on my website (see Notes
on the Text). Thanks to Catharine Diehl, Andrew Stephenson, and Damien Melamedoff for pressing me on
this point.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

The triviality of this conditional undermines the ontotheist position. The ontotheist
view, after all, is that the antecedent of this conditional explains the consequent. If the
consequent of the conditional is trivial, then the antecedent of the conditional is
doing no explanatory work. To accept (6) is to trivialize the doctrine that God
necessarily exists in virtue of his existence being grounded in his essence. If (6) is
taken to be the meaning of (2), then every object necessarily exists in precisely the
same sense that God necessarily exists: necessarily, if it exists, it exists.
In order to counter these problems, the ontotheist needs to retreat to (2) and reject
the principle that every object exists. Recall:
(2) a’s essence grounds a’s being F ⊃ □(Fa).
and the example from section 1:
(7) Caesar’s essence grounds Caesar’s being human
from which it follows that:
(8) □(human(Caesar)).
The ontotheist needs to understand (8) in a way that does not entail that it is
necessary that Caesar exists, that is, to interpret (8) so that it is consistent with:
(9) ◇¬exists(Caesar).
Claim (8) is the claim that in any counterfactual situation (or ‘possible world’) Caesar
is human; claim (9) is the claim that in some counterfactual situation (or ‘possible
world’) Caesar does not exist. What the ontotheist needs in order to make these
consistent is a distinction between an object having properties in a counterfactual
situation and that object existing in that counterfactual situation. This will involve
distinguishing, in each counterfactual situation, the objects that are merely the subject
matter of true propositions in that counterfactual situation, and the objects that exist in
that situation. Accepting (2) means accepting the consequence that every object has its
essential properties in every counterfactual situation, but this does not require embra-
cing the conclusion that every such object exists in every counterfactual situation. In
other words, the ontotheist needs to accept that there are values of a and F such that:
(10) ◇(Fa & ¬exists(a)).
In general, where a is a being that does not necessarily exist, but has an essence that
grounds its possession of property F, it will follow that a has F in every counterfactual
situation, even ones where it does not exist. In other words, the ontotheist must accept:
(11) ◇¬(x)(exists(x)).90

90
Cf. Fine (1994a), 3–4.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

This requires rejecting the most natural definition of the ‘existence’ predicate:
(12) exists(x) =def ¬(y)(y6¼x)
or, equivalently:
(13) exists(x) =def 9y (y=x).91
What my discussion so far brings out is that the ontotheist needs to deny (12) and (13)
and claim that existence is not equivalent to falling within the scope of the universal
quantifier (more precisely, being the value of a variable bound by the universal quanti-
fier). The ontotheist needs to understand ‘exists’ as a predicate that (potentially) applies
to only a subset of the objects that fall within the domain of the universal quantifier.
Consequently, the ontotheist should not think of the quantifier ‘9’—interdefinable in the
standard way with the universal quantifier—as the existential quantifier, but as a broader
quantifier that includes not only existing objects but (potentially) non-existent objects as
well. The ontotheist might think of this quantifier as expressing the natural language
quantifier expression ‘there is’ [es gibt] (that is why I have refrained from using ‘9’ up to
this point). It is important to point out that the ontotheist does not need to claim there
are non-existent objects, but only that such objects are possible; however, for many of the
ontotheists considered here, their other philosophical commitments may push them to
accept non-existent possible objects (e.g., Leibniz’s possible worlds).
It might be objected that the ontotheist can do the same work by distinguishing
between the objects that are actual and the objects that are not, while maintaining
that, necessarily, all objects (both actual and non-actual) exist.92 On such a view,
objects can have properties (e.g., those contained in their essences) in counterfactual
situations in which they are not actual. However, many of the philosophers we are
discussing use ‘existence’ and ‘actuality’ interchangeably. One notable instance is
Kant, because he does not consistently distinguish existence [Dasein] and actuality
[Wirklichkeit].93 Kant may have been simply following standard usage in German
philosophy at the time. In his Vorbereitung zur natürlichen Theologie, from which
Kant sometimes lectured, Eberhard’s list of recent proofs of the existence of God
seems to use ‘Dasein’ and ‘Wirklichkeit’ interchangeably (Ak. 18: 563); Baumgarten
also identifies ‘actualitas’ and ‘existentia’ in Metaphysica }55. Consequently, I will
treat existence and actuality, insofar as they are predicates of objects, as equivalent.94

91
These definitions are equivalent by stipulation: I am defining the quantifier ‘9x’ in terms of the
universal quantifier.
92
E.g., David Lewis famously believed in non-actual objects, but had no truck with the neo-Meinongian
view that there are non-existent objects. See Lewis (1986), 97–8.
93
Cf. ND (Ak. 1: 396), OPG (Ak. 2: 72, 75), and Danz.RT (Ak. 28: 1151, 1256, 1291). See, however, MK3
(Ak. 29: 986) as well as Refl. 6324, where Kant explicitly distinguishes actuality from existence. In Chs. 8.4
and 10.7 I argue that Kant’s Critical view does include a distinction we could call the existence/actuality
distinction, but that distinction will not affect the dialectic here.
94
‘Actuality’ here refers to a predicate of objects (actual(x)). Actuality as a propositional operator
(actually p) is different.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

Since the ontotheist needs to presuppose that there may be non-existent objects, and
‘objects’ here means (possible) beings, this is equivalent to the presupposition that
there may be non-actual/non-existent possible objects/beings.95
This thesis, that there could be non-existent objects, I will call possibilism:
(Possibilism) There could be merely possible objects (beings, entia, Dinge) that
do not exist.
The ontotheist, in order to account for God’s necessary existence, must accept
possibilism, or so I have argued. The negation of possibilism is what I will call
actualism:
(Actualism) Necessarily, there are no non-existent merely possible objects
(beings, entia, Dinge).
Earlier, I pointed out that one path to possibilism is to reject the most natural
definition of existence in terms of the ‘there is’ quantifier:
(14) exists(x) =def ¬(y)(y6¼x) [=def 9y (y=x)].
On the other hand, if we adopt this as the definition of existence, the following
expresses the claim that every object exists:
(15) (x) ¬(y)(y6¼x).
Since this is a logical truth, and logical truths are necessary, this entails that
necessarily every object exists, actualism. So actualism is a direct consequence of
defining the existence predicate as in (14); possibilism entails the negation of that
definition.
Some readers may feel that they began reading a book on eighteenth-century
metaphysics and are now reading a book on contemporary analytic philosophy.
What have possibilism and non-existent possibilia to do with Leibniz, Wolff, and
Baumgarten, much less Kant? Recall that ‘being’ is the highest genus of ontology and
all beings have essences.96 For each being (ens) there is an answer to the question,
what is it for that being to be? For creatures, an answer to the question ‘what is it for
that being to be?’ does not determine whether that being exists. God is the unique
being whose essence grounds his existence. The question of possibilism—could there
be non-actual possibilia?—can be formulated in this terminology as the question:
could there be a being that has an essence but lacks existence? This question, I take it,

95
‘Object’ here means object of quantification. I argued in }3 that objects of quantification (the objects
there are) are all and only beings (Dinge).
96
I am assuming, as Leibniz, Wolff, Baumgarten, and (I think) Descartes do, that every ens has an
essence. This assumption may not be universally shared in the Scholastic tradition, though. Thanks to Kris
McDaniel for drawing my attention to this point.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

is not an anachronistic interjection, but one that these ontotheists could make perfect
sense of. 97

1.6. Real Predicates

In Beweisgrund Kant objects to the ontological argument that “existence is not a
predicate or a determination of a thing” (Ak. 2: 72), and in the CPR he writes that
“being is obviously not a real predicate” (A598/B626). In this section I explore what
these claims mean.
First of all, ‘real predicate’ and ‘determination’ are synonymous, as Kant makes
clear in a parthenthetical remark at A598/B626: “a real predicate (that is, a deter-
mination of a thing).” A logical predicate is any concept that can be a predicate in a
judgment: “anything one wishes can serve as a logical predicate, even the subject can
be predicated of itself” (A598/B626).98 So from the fact that there are existential
judgments, judgments whose predicate is <exists>, it follows immediately that
<exists> is a logical predicate. So there is really one claim, expressed in two syn-
onymous ways: <exists> is not a determination, and <exists> is not a real predicate.
While Kant does not explicitly distinguish ‘real’ from merely ‘logical’ predicates in
Beweisgrund, this distinction is implicit when he claims that, although existence is
not a (real) predicate, it is admissible to use it as a (logical) predicate in ordinary
speech as long as one does not thereby assume that it is a determination of objects:
Nonetheless, the expression ‘existence’ is used as a predicate. And, indeed, one can do this
safely and without troublesome errors, as long as one does not insist on trying to derive
existence from merely possible concepts [ . . . ] (OPG, Ak. 2: 72)

However, it is not at all clear what Kant means by denying that existence is a
“determination.” On this point, his definition in the Critique is unhelpful: “the
determination is a predicate, which goes beyond the concept of the subject and
enlarges it. It must therefore not be contained in the subject concept” (A598/B626).99
On the most literal reading of this passage, a determination of an object is a ‘synthetic
predicate,’ a predicate of the object that is not contained in its concept. More
precisely, since every object falls under indefinitely many concepts, a concept C is a
synthetic predicate of concept C* just in case C is not contained in C* (the judgment

97
Descartes explicitly discusses the essence/existence issue; see Chapter 2.2. For Wolff ’s views see Dt.
Met. }544; for an analysis of Wolff ’s relation to the Scholastic debate about essence and existence see
Honnefelder (1990), 320, and 367–70. See also the supplementary article “Essence and Existence” on my
website (see Notes on the Text). Thanks to Uygar Abaci for pointing out the relevance of the essence-
existence debate to my argument here.
98
Since the predicate of one judgment can be made the subject of another, this means that all concepts
that can be predicates of judgments—that is, all concepts whatsoever (A69/B94)—are logical predicates; see
A94/B129. See Proops (2015), 11 for more evidence that predicates are concepts.
99
For more on the notion of a ‘synthetic predicate’ see JL (Ak. 9: 59).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

All C*s are C is synthetic). However, as other commentators have pointed out, this
interpretation commits Kant to the following inconsistent triad:
(1) Existence is not a determination of any concept, i.e. the predicate <exists> is
not synthetic with respect to any concept.
(2) All existential judgments are synthetic.
(3) If a judgment is synthetic, then its predicate is synthetic with respect to its
subject.100
Since Kant repeatedly asserts (2) in this section of the CPR and (3) follows from the
definition of a synthetic judgment, either the interpretation of determinations as
‘synthetic predicates’ is mistaken, or Kant contradicts himself within the space of a
few paragraphs.
Another possibility is that Kant means ‘determination’ here in Baumgarten’s
technical sense: “what is either posited to be A, or posited not to be A, is DETER-
101
MINED.” However, this refers to any predicate whatsoever; that there are existential
judgments (judgments in which the predicate is <exists>) entails that existence is a
determination in Baumgarten’s sense. In his own copy of Metaphysica Kant identifies
Baumgarten’s definition of ‘determination’ with that of a logical predicate (Refl. 3520,
Ak. 17: 33); since <exists> is a logical predicate, it is a determination in Baumgarten’s
sense.
A more tempting possibility is to deny that <exists> is a predicate of objects at all,
but a predicate of concepts, anticipating the Fregean theory of the existential quan-
tifier as a second-order concept that applies to concepts that have a non-empty
extension. This interpretation finds support in this oft-quoted passage from
Beweisgrund:
But when existence occurs as a predicate in common speech, it is a predicate not so much of the
thing itself as the thought which one has of the thing. For example: existence belongs to the
narwhal [Seeeinhorn] but not to the unicorn [Einhorn]. This simply means: the representation
of a narwhal is an empirical concept; in other words, it is the representation of an existent
thing. [ . . . ] The expression ‘A narwhal is an existent animal’ is not, therefore, entirely correct.
The expression ought to be formulated the other way around to read ‘The predicates, which
I think collectively when I think of a narwhal, attach to a certain existent animal.’ (OPG, Ak. 2:
72–3)102

Kant claims that the judgment a narwhal is an existent animal does not assert that
some predicate is contained in the concept <narwhal>, and does not attribute further
predicates to the objects that fall under <narwhal>; it asserts that there is at least one
object that falls under <narwhal>, i.e. that the concept is instantiated.103 I take this to

100
Cf. Wood (1978), 105 and Shaffer (1969), 125. 101
Meta. }34. 102
Cf. A599/B627.
103
Kant gives the mistaken impression that, on his analysis, ‘narwhals exist’ is equivalent to ‘there is a
narwhal and I have experienced it’. This would make the following judgment false as a matter of meaning:
there are narwhals but no one has ever encountered one. But he drops the misleading suggestion that the
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

be clear evidence that the fundamental (though not the only) role of the predicate
<exists> is to apply to concepts: it applies to a concept if and only if that concept is
instantiated by an object. While this is not yet the complete Fregean theory of the
existential quantifier, it does anticipate it.104 Consequently, I will borrow from the
contemporary symbolism and formalize narwhals exist as:
(4) 9x(narwhal(x))
which is to be read as ‘there is an object in the extension of <narwhal>.’105
This, however, cannot exhaust the content of Kant’s claim that existence is not a
determination or ‘real predicate’ because it is not something ontotheists need to deny
and thus has no force by itself against the ontological argument. As we saw in section
5, ontotheists are committed to possibilism, the view that there could be non-existent
objects. This is entirely compatible with the view that to make existence claims we
need a quantifier expression, a second-order predicate that applies to a concept just
in case it is instantiated; the ontotheist needs merely to add that in existential
judgments like narwhals exist the quantifier is restricted to existing objects.106 So
the ontotheist can fully accept Kant’s analysis of existential judgments as long as he
interprets the quantifier-expression in (4) as implicitly restricted to existing objects,
that is:
(4*) 9x2E (narwhal(x))
where E is the set of all existing objects. This is equivalent (assuming E is non-empty)
to the following claim, using an unrestricted quantifier and an existence predicate for
objects:
(5) 9x(exists(x) & narwhal(x)).107
In other words, the ontotheist can fully accept that existence is a second-order
predicate (a quantifier) as long as it is a restricted quantifier; alternately, that

instance of the concept must be experienced when he analyzes ‘God exists’ as “an existing thing has those
predicates, which we collectively designate with the expression: God” (Ak. 2: 74).
104
For one thing, Kant has not yet developed the Fregean function-object analysis of judgment, and his
neglect of relational predicates leaves him without the resources to develop a theory of polyadic quanti-
fication. Friedman (1992a), 96–135 discusses the limitations of Kant’s logic.
105
Rosenkoetter (2010) objects to what he calls the ‘Frege-anticipation’ thesis that it is incompatible
with Kant’s claim that the assertoric function of judgment corresponds to the category of existence. He
claims that the Frege-anticipation thesis would entail that “Kant would need to hold, in parallel, that all
assertoric judgments can be reduced to q is true” (552). However, it is unclear what Rosenkoetter’s
argument for this claim is, nor is it clear why the defender of the Frege-anticipation thesis cannot hold
Rosenkoetter’s own account of the assertoric function of judgment.
106
One can (as e.g., Quine did) hold that existence is a quantifier without holding that it is a second-
order predicate of concepts. For the purposes of this book I will be identifying the view that existence is a
quantifier with the view that it is second-order.
107
Where the extension of the predicate ‘exists(x)’ is E, the set of existing things used to restrict the
quantifier in (4*).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

existential judgments are made using an unrestricted quantifier (‘there is,’ which
ranges over all objects) and an existence predicate that applies to only some objects.
This interpretation may even be encouraged by Kant’s own text, quoted above,
because he interprets narwhals exist as the judgment that “the predicates, which
I think collectively when I think of a narwhal, attach to a certain existent animal”.
This may give the (false, or so I will argue) impression that Kant himself thinks that
existential judgments are made using an unrestricted quantifier (‘there is’) and an
existence predicate for objects.
This brings out an important, and, I believe, too often neglected point: the real
issue between the ontotheist and Kant over existence is not whether existence is a
quantifier (second-order predicate) but whether it is a restricted or unrestricted
quantifier. The real issue is whether there is an existence predicate for objects that
applies to only a subset of them (equivalently, whether the existence quantifier is a
restriction of the ‘there is’ quantifier); the ontotheist (I have argued) must maintain
that the existence predicate for objects applies only to a subset of them.108
The ‘synthetic predicate’ interpretation, from above, assumes that, when Kant
writes that the determination “enlarges” the subject concept, all he means is that the
determination is not one of the marks analytically contained in the subject concept.
But this is not the only sense in which a predicate might be said to “enlarge” the
subject concept of a judgment. A predicate might also enlarge a concept by enlarging
its content and rendering that contept more determinate by restricting the range of
objects that can fall under it. I propose, then, the following interpretation of Kant’s
technical term ‘determination’:
(Defn.) A concept P determines a concept C if and only if it is possible that there
is an object that instantiates C and P and it is possible that there is an object that
instantiates C but not P.109, 110
(Defn.) A predicate P is a determination if and only if P determines at least one
concept.111
One concept can determine another, in the sense of specifying the nature of the
objects falling under the concept. <Scalene> determines the concept <triangle>, but
<having interior angles that sum to 180 degrees> does not. It does not add any new
specification to the concept <triangle>, even though it is not analytically contained in

108
Forgie (2007) argues that Kant’s claim that existence is a second-order predicate should not be
conflated with Gassendi’s claim that existence is not a property but the precondition for having properties
in the first place. My point is that Kant’s attempt to refute ontotheism requires him to make a version of
Gassendi’s claim: there cannot be objects that do not exist.
109
Cf. Pöl.RT (Ak. 28: 1027), Volck.RT (Ak. 28: 1176), and Danz.RT (Ak. 28: 1258), as well as the texts
from Kant’s metaphysics lectures on ‘determinieren’ cited below.
110
This interpretation is similar to that of Van Cleve (1999), 188 and Hanna (2001), 133.
111
I am assuming that existence, in virtue of being a logical predicate, is a concept.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

that concept.112 Thus, on my definition, all determinations are synthetic predicates,

but not every synthetic predicate of an object is a determination.
This interpretation also preserves the natural reading of the passage: “the
determination is a predicate, which goes beyond the concept of the subject and
enlarges it [the concept]. It must therefore not be contained in the subject concept
[sie muß also nicht in ihm schon enthalten sein]” (A598/B626—my emphasis). On
my reading, the second sentence is a consequence of the first, but is not identical to
it: because a determination enlarges the subject concept in the sense of adding
further content to it, the determination cannot be analytically contained in the
subject concept. On the ‘synthetic predicate’ reading, the second sentence merely
restates the first sentence.113
In addition to respecting the natural reading of this passage, this interpretation has
at least five other advantages. First, it allows us to escape the inconsistent triad of
views; just because <exists> does not determine any subject concept, it does not
follow that existence is not a synthetic predicate. This is because:
(1) For any concept C, necessarily anything that falls under C falls under
<exists>; <exists> is not a determination
is compatible with:
(2) No judgments of the form there exists a C are analytic; existence is not a mark
of any concept.114
The second advantage to this reading is that it allows us to explain why <exists> is not
a determination: it does not ‘enlarge’ or further specify any concept, because it is the
concept that every object falls under. If <exists> were a determination of some
concept C, it would follow that it is possible for there to be objects that fall under
C but not <exists>; it would follow that it is possible for there to be non-existent
objects.
This point connects directly to our earlier discussion of the quantificational theory
of existence. As a predicate of objects, <exists> is a determination just in case there is
some concept C such that:
(3) ◇9x(Cx & ¬exists(x))
from which it follows that:
(4) ◇9x(¬exists(x)).

112
Because for Kant, the judgment that every triangle has internal angles that sum to two right angles is
synthetic, not analytic.
113
Similarly, when Kant defines ‘determination’ in the Mrongovius metaphysics lectures, he says “the
logical predicate can be analytic, but determination is always synthetic” (Ak. 29: 819). He notably does not
claim that all and only synthetic predicates are determinations.
114
The compatibility of these two claims follows from Kant’s acceptance of synthetic a priori judg-
ments, which, by definition, are necessarily true and not analytic. See B4 and Ak. 8: 235.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

LOGICISM AND ONTOTHEISM 

If the object-predicate <exists> is a determination, it follows that it is possible for

there to be a non-existent object. But notice that, conversely, if (4) is true, it follows
that <exists> is a determination; (4) entails that <exists> determines the trivial
concept <x=x>, the concept under which all objects necessarily fall. So existence is
a determination if and only if (4) is true.
The natural way of defining the existence predicate for objects is by using the
quantifier ‘9’ (which is equivalent to the definition in terms of the universal quan-
tifier, given in brackets):
(5) exists(x) =def 9y (y=x) [$ ¬(y)(y6¼x)].
Applying this to ‘exists’ in (4), we get:
(6) ◇9x¬9y(y=x)
which is a logical falsehood. There cannot be an object such that there is no object to
which it is identical. The definition of existence using the quantifier ‘9’ (claim (5))
entails that <exists> is not a determination. If we interpret ‘determination’ as I have
argued we should, Kant’s claim that existence is not a determination, while not
philosophically uncontroversial, is highly plausible. This is the third advantage of my
reading. Kant’s claim that existence is not a determination is equivalent to defining
the object-level existence predicate in the natural way, using the quantifier ‘9’. It is
equivalent to claiming that the quantifier expression ‘there is’ ranges only over
existing objects, that is, that there are no non-existent objects. On such a theory of
existence, it is appropriate to call the quantifier ‘9’ the existential quantifier. This also
shows that, although the fundamental sense of existence for Kant may be given by the
existential quantifier ‘9’, he also has the resources to define an existence predicate for
objects, as in (5). Furthermore, if the existence predicate for objects, <exists>, is
defined in terms of the quantifier ‘9’ then we can translate unproblematically between
them. Henceforth, when I talk about <exists>, I will always have the object-predicate
in mind; when I mean the second-order predicate of concepts (what is now called a
quantifier), I will use ‘9’.
One might well wonder, then, if the fundamental meaning of existence is a
quantifier, why does Kant bother discussing an existence predicate for objects? To
refute the ontotheist, of course! As I argued earlier, the ontotheist can accept that
existence is a quantifier, but then the real issue becomes: is the existential quantifier a
restriction of the wider quantifier ‘there is’? On my interpretation, by denying that
existence as a predicate of objects is a determination, Kant is answering this question
in the negative. This is equivalent to denying that there are (or could be) non-existent
objects, which, as we saw in section 5, would be devastating to the metaphysics of
ontotheism in general, and, I will argue in Chapter 2, the specific ontological argu-
ments given by Kant’s specific opponents (Descartes, Leibniz, etc.). So the fourth
advantage of my reading is that Kant’s claim that <exists> is not a determination, if
true, spells serious trouble for ontotheism.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 LOGICISM AND ONTOTHEISM

Fifthly, and finally, this interpretation of what it is to ‘determine’ a concept is well

supported by Kant’s lectures on metaphysics. For example:
The principle that, of two contradictorily opposed predicates, one pertains to the object is
called the ‘principle of excluded middle among two contradictories.’ This is called determining
[determinieren] for various objects are left indeterminate by the concepts we have of them. For
example, the concept of human [der Begriff des Menschen] is indeterminate. [A human] can be
learned or not, can be a man or a woman, etc. If I then say that one of the two must pertain to
[an object], and I posit one of them, [the object] is determinate; indeterminate means
determinate only in respect of what we think through our concept. For example, the concept
of an angle is indeterminate, for it can be obtuse or acute. (MV, 28: 410)115

To determine [determinieren]116 a concept is to predicate one of two contradictorily

opposed concepts, each of which is a possible way that the original concept can be
specified; e.g., <acute> determines <angle> because it is possible for angles to be acute,
and possible for them to be obtuse. We can say that <acute> and <obtuse> possibly
divide the extension of <angle>. To say that <exists> does not determine any concept
means therefore that <existence> and <non-existence> do not possibly divide the
extension of any concept. The extension of no concept divides into existing and
non-existing instances of that concept because there are no non-existent objects.117
If Kant is right that existence is not a determination, and if I am right that
ontotheism—the doctrine that God exists in virtue of his essence—entails that
existence is a determination, then Kant has refuted ontotheism as such. The gener-
ality of Kant’s objection entails that there can be no being, not even God, that exists in
virtue of its essence. Since logicism, combined with the traditional doctrine that God
necessarily exists, entails that God exists in virtue of his essence, this also refutes
logicism. If existence is not a determination, then there is at least one necessary truth,
that God exists, that is not a logically necessary consequence of the essences of
possible beings. In the next chapter I reconstruct in detail Kant’s arguments that
existence is not a determination.

115
Cf. MH (Ak. 28: 14, 19, 24–5, 843, 845), MM (Ak. 29: 818), MvS (Ak. 28: 491), and ML2 (Ak. 28:
551–2).
116
These passages typically use the Latinate phrase ‘determinieren’ but we have seen that in Beweis-
grund ‘Determination’ is also Kant’s term for what he will later refer to as a ‘Bestimmung’ or ‘real predicate’
(Ak. 2: 72).
117
Crusius’s technical notion of Determination in Ent. }23 is quite similar; it is likely that Kant is using
the Crusian notion of determination, rather than Baumgarten’s.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

2
Is Existence a Real Predicate?

2.1. Introduction
In the first half of this chapter I reconstruct in detail the ontological arguments of
Descartes (section 2), Leibniz (section 3), and Baumgarten (section 4), to show that each
of them is committed to possibilism, the view that there could be objects that do not exist.
This confirms my argument, in section 1.5, that ontotheology as such is committed
to possibilism. In the second half of the chapter I reconstruct Kant’s arguments in
Beweisgrund and in the CPR that existence is not a determination, which, I argued
in Chapter 1.6, means that possibilism is false (there cannot be non-existent objects).
Although Wolff does give an ontological argument in Part II of his Theologia
naturalis (1737), it introduces little that is not already found in Leibniz’s arguments,
so I will largely ignore Wolff in this chapter.1 Descartes is a difficult figure in this
context. First of all, he is not a logicist; on the contrary, he believes in brute necessary
connections, knowable to us through clear and distinct perception, but not grounded
in any relation of logical entailment. However, like the logicists, he is an ontotheist;
he believes that God exists necessarily in virtue of the ‘involvement’ of existence in
God’s essence. Furthermore, Descartes in many ways set the pattern for early modern
ontological arguments. Consequently, I begin with Descartes’ version of the onto-
logical argument and argue that it too is committed to possibilism.

2.2. Descartes’ Ontological Argument

Descartes’ ontological argument in the Fifth Meditation is the most famous version
of the argument in modern philosophy. It deeply influenced Leibniz’s thinking about
the ontological argument, as well as that of Wolff, Baumgarten, and Kant.2 Descartes

1
TN was translated into German by Gottlieb Friedrich Sagen under the title Natürliche Gottesgelahrheit
nach beweisender Lehrart abgefasst (NG); both versions are included in Wolff ’s Werke (see Note on
Sources). Wolff first introduces the concept of compossibility (}1), defines the ens perfectissimum as the
being in which all compossible determinations are present to the highest degree (}6), argues that the ens
perfectissimum is possible because its concept contains no negations (because all of its determinations are
unlimited, hence positive) and thus cannot contain a contradiction (}13), claims that existence is a
determination in things (}20), and concludes that the ens perfectissimum has existence to an unlimited
degree, hence it exists (}21).
2
Cf. Wolff ’s summary of the recent history of the ontological argument in NG }13.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

is widely credited with resuscitating the ‘Anselmian’ argument, although I think that
Anselm’s argument is different enough from the Cartesian argument that I forgo
discussion of it here.3 Although the Cartesian argument is originally given in the
body of the Fifth Meditation,4 I think the three-line, syllogistic version of the
argument in the First Replies is clearer, so I will start there:
[1] That which we clearly and distinctly understand to belong to the true and immutable
nature, or essence or form of something, can truly be asserted of that thing. [2] But once we
have made a sufficiently careful investigation of what God is, we clearly and distinctly
understand that existence belongs to his true and immutable nature. [3] Hence we can now
truly assert of God that he does exist.5

Premise [1], Descartes’ argument for it, and the very notion of a clear and distinct
perception on which Descartes bases much of his argument in Meditations, are
highly problematic, as was forcefully pointed out by Descartes’ contemporaries in
the Objections, and as has been pointed out again and again by scholars since. I will
not focus on those problems with the argument, though, because they are not
problems with the ontological argument as such.
As the second set of Objections points out, since only possible beings have true and
immutable natures, Descartes must assume that God is possible: “it does not follow
from this that God in fact exists, but merely that he would have to exist if his nature is
possible, or non-contradictory” (CSM II: 91/AT II: 127).6 But Descartes had already
anticipated, and answered, this objection in his response to Caterus in the First
Replies: “possible existence is contained in the concept or idea of everything we
clearly and distinctly understand.”7 The possibility of God is guaranteed by our clear
and distinct understanding of his true and immutable nature. So expanding upon the
original argument and filling in its enthymematic premises, we can reconstruct it as
follows:
Descartes’ Ontological Argument:
(1) If I clearly and distinctly understand that p then p.
(2) If I clearly and distinctly understand a true and immutable nature, then the
object whose nature that is, is possible.
(3) There is a true and immutable nature of God and I clearly and distinctly
understand it.

3
Cf. Barnes (1972).
4
Cf. First Objections and Replies (Objection: 99–100; Reply: 113–21); Second Objections and Replies
(Objection: 127; Reply: 150–2, 163–7); and Fifth Objections and Replies (Objection: 322–6; Reply: 382).
References give the page numbers in vol. 7 of AT; this pagination is also given in the margins of CSM. (See
‘Note on Sources and Translations’.)
5
CSM II: 83/AT VII: 115–16. I have inserted the premise numbers and the line spacing; Descartes
presents the argument in continuous prose.
6
Presumably this is because there are no true and immutable natures of impossibilities (non-things,
Undinge).
7
CSM II: 83/AT VII: 116.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

(4) ∴ God is possible. [From (2) and (3).]

(5) I clearly and distinctly understand that every perfection is contained in God’s
true and immutable nature.
(6) ∴ Every perfection is contained in God’s true and immutable nature. [From
(1) and (5).]
(7) Any property contained in the true and immutable nature of a possible object
is a property had by that object.
(8) God has every perfection. [From (6) and (7).]
(9) Existence is a perfection.
(10) ∴ God exists.8 [From (8) and (9).]
Lines (1)–(4) establish the possibility of God; lines (5)–(10) spell out precisely how
Descartes goes about showing that existence is contained in God’s true and immut-
able nature and deriving the conclusion that God exists.
In the Replies, Descartes’ interlocutors raise a number of objections to this
argument, but few have been as influential as Gassendi, who writes:
[ . . . ] existence is not a perfection either in God or in anything else; it is that without which no
perfections can be present. For surely, what does not exist has no perfections or imperfections,
and what does exist and has several perfections does not have existence as one of its individual
perfections; rather its existence is that in virtue of which the thing itself and the perfections are
existent, and that without which we cannot say that the perfections are existent, and that
without which we cannot say that the thing possesses the perfections, or that the perfections are
possessed by it. Hence we do not say that existence ‘exists in a thing’ in the way that perfections
do; and if a thing lacks existence, we do not say it is imperfect, or deprived of a perfection, but
say instead that it is nothing at all.9

Gassendi is claiming that existence, far from being one perfection among others, is
the condition without which a thing has no perfections, indeed no properties.10 Any
bearer of properties must exist; if a thing does not exist, “we do not say that it is
imperfect . . . but say instead that it is nothing at all,” that is, that there is no such
thing. Gassendi is assuming an actualist theory of existence: every object (being)
exists. This means that (7), sometimes known as the ‘Cartesian predication principle,’
has to be understood as the conditional claim that if some property is contained in an

8
In some presentations of the argument Descartes claims that it is necessary existence that is one of the
divine perfections (Second Replies); in other texts he identifies existence as the perfection in question (e.g.,
the Fifth Meditation itself, as well as the First and Fifth Replies). In the second set of Replies he switches in
the space of a few paragraphs from claiming that it is existence that is one of the divine perfections (AT VII:
151) to claiming that it is necessary existence (AT VII: 152) to deciding in favor of the latter in his
‘geometric’ presentation of the argument (AT VII: 166). I have focused on the former presentation of the
argument (in terms of existence simpliciter) in order to facilitate comparison of Descartes with the logicists.
9
CSM II: 224/AT VII: 323.
10
Forgie (2007) cautions us against assimilating Gassendi’s objection to the ontological argument
(existence is not a property) to Kant’s (existence is not a determination). On this point I agree; but
I dissent from his reading of Kant. See Ch. 1.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

essence and there is an object with that essence, then the object has the property.11
The Cartesian predication principle cannot (if we are assuming actualism) be the
claim that if some property is contained in an essence then the object of that essence
contains the property, because that presupposes the essence has an object, i.e. some
existing object instantiates that essence. Instead, it has to be understood as:
(7*) (x)[(E is the essence of x & F is contained in E) ⊃ Fx)], for any essence E and
any property F.
But recall the work (7) is supposed to do in the argument. It is supposed to license the
inference from the claim that every perfection is contained in the true and immutable
nature of God (6) to the conclusion that God has every perfection (8). In order to do
that, though, Descartes would have to show for some x and some E that E is the true
and immutable nature of x, i.e. he would have to show of some x that x is God. In
other words, he would have to first show that God exists in order to show that God
exists! Descartes’ ontological argument is pointless, according to Gassendi, because it
requires its conclusion as one of its premises.
The logical form of Descartes’ argument, as Gassendi understands it, is:
Gassendi’s Reading of Descartes’ Ontological Argument:
(1*) (I clear and distinctly understand that p) ⊃ p.
(2*) I clearly and distinctly understand an essence E ⊃ ◇9x(E is the essence of x).
(3*) There is an essence (true and immutable nature) of God and I clearly and
distinctly understand it. Call this essence ‘G.’
(4*) ◇9x(G is the essence of x). [From (2*) and (3*).]
(5*) I clearly and distinctly understand that every perfection is contained in G.
(6*) ∴ Every perfection is contained in G. [From (1) and (5).]
(7*) (x)[(E is the essence of x & F is contained in E) ⊃ Fx)], for any essence E and
any property F.
(8*) ∴ For any perfection F, (x)(G is the essence of x ⊃ Fx)). [From (6*) and (7*).]
(9*) Existence is a perfection.
(10*) ∴ 9x(G is the essence of x).
Thus, the argument is invalid on Gassendi’s reading because (10*) does not follow
from (8*) and (9*). Descartes never proves that there is a God (an entity that has the
divine essence, my clear and distinct perception of which was the origin of the
argument), only that it is possible that there is one. The most Descartes is warranted
in concluding is the conditional claim that:
(10**) (x)(G is the essence of x ⊃ exists(x)).

11
Caterus points out in the First Objections that the most Descartes is entitled to is this conditional
claim (CSM II: 72/AT VII: 99). Compare this to Gassendi’s remark that: “all that this means is that if
anything is a man, it must resemble other things to we which we apply the same label ‘man’ ” (CSM II: 222;
AT VII: 320). Cf. A594–5/B622–3.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

If we assume that bound variables only take existing objects as values, this is trivial;
it means something like: if anything existing is God (has the divine essence), then it
exists. As another insightful critic of Descartes’ ontological argument, Caterus, put it
in his Reply: “you cannot infer that the existence of God is anything actual unless you
suppose that the supreme being actually exists; for then it will actually contain all
perfections, including the perfection of real existence.”12 Note, though, that (10**) is
trivial only if we are defining the predicate ‘exists’ in the usual way:
(Ex) exists (x) =def ¬(y) (y6¼x) = 9y(y=x).
This, as discussed earlier, is tantamount to assuming actualism. Gassendi, and
Caterus, are correct that the conclusion of this argument (10**), is trivial if we
assume actualism.
Descartes’ response to Gassendi is instructive:
Here I do not see what sort of thing you want existence to be, nor why it cannot be said to be a
property just like omnipotence—provided, of course, that we take ‘property’ to stand for any
attribute, or for whatever can be predicated of a thing; and this is exactly how it should be taken
in this context.13

Descartes is reiterating that God has the properties contained in his essence, regard-
less of whether God exists or not (because at this point in the argument, we have not
yet proved that God exists). In other words, Descartes is thinking that he has already
proved that there is a God whose true and immutable nature he understands,
otherwise he would be unable to use the Cartesian predication principle (7*) to
prove anything other than a conditional conclusion. But this means Descartes is
thinking of the logical form of the initial steps of the argument as:
(1{) (I clear and distinctly understand that p) ⊃ p.
(2{) I clearly and distinctly understand an essence E ⊃ 9x(E is the essence of x &
possible(x)).14
(3{) There is a divine essence & I clearly and distinctly understand it. Call this
essence G.
(4{) ∴ 9x(G is the essence of x & possible (x)). Let g be such an object. [From (2{)
and (3{).]15
At this point, Descartes is licensed in claiming that there is a God, because he has
shown there is an object with the divine essence. The rest of the argument shows that
God, the being with the divine essence, exists:

12 13
CSM II: 72/AT VII: 99. CSM II: 263/AT VII: 383.
14
The second disjunct of the consequent is redundant because I have stipulated that quantifiers range
only over beings, and beings as such are possible (}1.3). However, I retain it so as to make explicit the
difference between how the actualist (2*) and possibilist (2{) deal with possibility: the actualist uses a
modal operator, while the possibilist uses a wider quantifier.
15
For ease of exposition, I am ignoring the issue of whether g is the unique such object.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

(5{) I clearly and distinctly understand that every perfection is contained in G.

(6{) ∴ For every perfection F, F is contained in G. [From (1{) and (5{).]
(7{) (x)[(E is the essence of x & F is contained in E) ⊃ Fx)], for any essence E and
any property F.
(8{) For any perfection F, Fg. [From (4{), (6{), and (7{).]
(9{) Existence is a perfection.
(10{) ∴ Existence(g).
Descartes is thinking of the argument as showing that there is a God, and thus
introducing God as a subject for non-conditional predication, prior to proving that
God exists. This only makes sense if Descartes rejects the claim that:
(Actualism) □(x)(exists(x)).
But Descartes has to reject actualism anyway, because it is in tension with (2{).
Assuming an actualist theory of existence, (2{) would entail that every idea I clearly
and distinctly understand is the idea of an actually existing object. This would have
the absurd consequence that every mathematical idea is the idea of an actually
existing object. From the mere fact that I can clearly and distinctly understand the
idea of a chiliagon it would follow that there is an actually existing chiliagon. To a
contemporary reader, this might appear unproblematic; chiliagons are abstract
objects, so it might seem unproblematic that if we have a clear and distinct idea of
them, they exist. However, mathematical objects are not, according to Descartes,
‘abstract’ in the contemporary sense; mathematical objects are bodies (res extensa), so
an actualist ontology would entail that, for every geometrical idea we clearly and
distinctly understand, there is a body with precisely that shape. I take it this is a
consequence Descartes would want to avoid.
In some texts, Descartes appears to endorse explicitly this commitment to non-
existent possibilia.16 However, in other passages his attitude is more ambivalent; for
instance, in Principles of Philosophy he endorses a merely ‘rational’ distinction
between essence and existence:
we do understand the essence of a thing in one way when we consider it in abstraction from
whether it exists or not, and in a different way when we consider it as existing; but the thing
cannot be outside our thoughts without its existence [ . . . ] Accordingly I say that . . . there is a
lesser distinction between [them] [ . . . ] I call it a rational distinction.17

16
AT VII: 64–6/CSM II: 45. Cf. Kenny (1968) and Griffin (2013), 9–33. More recently, some scholars
have questioned whether Descartes accepts non-existent possibilia; see esp. Nolan (1997) and (2005), and
Cunning (2008).
17
CSM III: 349/AT IV: 349. Thanks to Nolan (2005) for drawing my attention to this passage. Note,
though, that Descartes goes on immediately to write that, “if by essence we understand a thing as it is
objectively in the intellect, and by existence the same thing insofar as it is outside the intellect, it is manifest
that the two are really distinct” (CSM III: 281). This complicates matters, however; it is not clear this is
compatible with a merely rational distinction between essence and existence. In Principles, }26 Descartes
claims that the existence of a substance is one of its attributes, and since there is merely a rational
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

This doctrine means that every thing that has an essence also exists; the distinction
between essence and existence is a distinction between two different ways of con-
ceptualizing one and the same thing: through its primary attribute (essence) or as the
fully concrete thing it is (existence).18 If this is the case, then there cannot be non-
existent possibilia with true and immutable essences; such an ontology would require
a deeper distinction between essence and existence than is allowed by the ‘rational’
distinction doctrine.
My central interpretive claim is that Descartes’ ontological argument commits him
to a possibilist ontology; without it, the argument is invalid, as Gassendi and Caterus
pointed out. In Kantian terms this means he is committed to regarding existence as a
real predicate, a predicate that some objects have and some lack. That some of his
other views may be incompatible with this ontology—for instance, the merely
rational distinction between essence and existence—shows only that in one part of
his philosophy he is committed to a thesis that he rejects in another part. This is not
an uncommon situation, even for the greatest thinkers, like Descartes.

2.3. Leibniz’s Ontological Argument

Leibniz developed his own ontological argument, and his views about existence, in
attempting to overcome what he saw as the defects in Descartes’ argument. His most
fundamental objection to the Cartesian argument—repeated in numerous texts—is
that Descartes fails to prove that God is possible. Descartes’ argument proves, at
most, that God’s existence follows from his essence, his possibility. To conclude that
God exists we must first discharge the assumption that God is possible, that there is a
consistent essence of God.19 Descartes’ method of doing this is irreparably flawed,
according to Leibniz, by its reliance on the method of clear and distinct perception.
However, Leibniz’s theory of conceptual analysis and adequate ideas in “Medita-
tions” provides a replacement for the rule of clear and distinct perception that does
not suffer from the same problems as its predecessor. If we adequately understand a
concept, i.e. can give a real definition of it, then the concept is logically consistent and
therefore possibly instantiated. If our adequate understanding of one concept reveals
another concept contained in it, then the latter concept pertains necessarily to the

distinction among the attributes of a substance, it follows that the distinction between the essence
(principal attribute) and existence of a substance is merely rational; however, it is unclear what he
means in saying that existence is an attribute of a (contingent) substance. See Cunning (2008) for further
discussion.
18
Cf. the discussion of the distinction between essence and existence in Question XXXI of Suárez’s
Metaphysical Disputations (Suárez 1983); for critical discussion, see Wippel (1982a), (1982b), and Witt
(2010).
19
Leibniz makes this same point—that the Cartesian argument establishes only that if God is possible
then he exists—in a number of texts: “Meditations on Knowledge, Truth, and Ideas” (AG 25), DM }23,
Leibniz’s correspondence with Elizabeth (AG 238), and a wide range of letters and unpublished notes from
throughout his career (L 165 f., 168, 211, 231, 286).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

former concept. Clear and distinct perception, for Leibniz, consists in ascertaining
the logical relations among concepts and their internal logical consistency.20
To return to the original objection—that Descartes fails to prove that a most real
being is possible—Leibniz’s logicist modal metaphysics is designed to dispense with
this objection. As we saw in chapter 1.3, complex concepts are composed of simple,
logically atomic concepts. The operations by which complex concepts are formed
are conjunction, negation, and limitation. The primitive concepts, therefore, are
logically simple (they are not conjunctions of other concepts), purely positive (they
are not negations of other concepts), and they are concepts of unlimited perfections.
Leibniz takes the unlimited perfections to be logically prior to limited perfections, i.e.
limited perfections are ‘understood’ through unlimited ones.21 This means that all
concepts of finite things are logical constructs—through conjunction, negation, and
limitation—of infinite and unlimited perfections. A finite mind possesses limited
versions of, for instance, the unlimited perfections of understanding and will.22
The unlimited perfections are the logically atomic constituents of all other con-
cepts. Leibniz’s definition of logical entailment in terms of conceptual analysis means
that an atomic concept, one with no internal logical structure, cannot stand in
entailment relations with another atomic concept. The unlimited perfections, being
logically atomic, are logically independent. Since the unlimited perfections are
logically independent, they are logically compatible.23 Therefore, the concept of
a being possessed of all possible unlimited perfections, <ens perfectissimum>, is a
logically consistent concept. Consequently, given the logicist analysis of possibility in
terms of logical self-consistency, an ens perfectissimum is possible.24 This is Leibniz’s
point in Monadology }45: “nothing can prevent the possibility of what is without limits,
without negations, and consequently without contradiction” (AG 218).
Before continuing, I want to flag two features of Leibniz’s ontological argument
(and the ontological arguments of the ‘Leibnizians’, Wolff and Baumgarten) that I am
going to ignore in what follows. The first is the uniqueness of the ens perfectissimum.
All three logicists held the doctrine of the identity of indiscernibles, and since the
concept of an <ens perfectissimum> completely determines the intrinsic properties of
its object, it follows that there cannot be two distinct objects instantiating it. I will
take it for granted that, if there is at least one ens perfectissimum, there is only one.
Secondly, I am going to ignore the difference between perfection and reality and the

20
It should be noted, however, that Leibniz’s theory is subject to a problem analogous to one that afflicts
Descartes: how do we know that a candidate analysis of a concept is the correct one? Furthermore,
Leibniz’s epistemology rests on a logicist metaphysics, which is problematic in its own right, as I will
argue at length in Chs. 3–4.
21
L 167/DSR 101. Cf. the correspondence with Elizabeth (AG 240).
22
Cf. Mon. }48. For an extensive discussion of Leibniz’s views on limited and unlimited perfections, see
Adams (1994), 115–19.
23
See DSR 69, though, for a slightly different picture of the relation among the divine attributes.
24
Adams analyzes in great detail Leibniz’s argument for the compatibility of the unlimited perfections
in (1994), 142–8.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

concomitant difference between conceiving of God as the ens perfectissimum and as

the ens realissimum. In Beweisgrund Kant claims that “ ‘perfection’ [Vollkommenheit]
always presupposes a relation to a being endowed with cognition and desire” (Ak. 2:
90) and substitutes the morally neutral term ‘reality’ in its place. While there are
reasons to dispute whether Kant is right about the Leibnizian concept of perfection,25
I am going to sidestep the issue and just treat ‘perfection’ as standing in for the more
morally neutral metaphysical concept of ‘reality,’ which we can think of as a gradable
degree of being.26 Ignoring these complications is appropriate to my goal: not to
reconstruct Kant’s objections to particular ontological arguments (e.g., that the
identity of indiscernibles might be false, or that perfection is distinct from reality)
but to ontological arguments as such.
The modal independence of the unlimited perfections allows Leibniz to formulate
his own version of the ontological argument in “Meditations.” If we make explicit its
suppressed premises, that argument is:
Leibniz’s “Meditations” Argument:
(1) Any logically consistent concept is the concept of a possible thing.
(2) The concept <ens perfectissimum>—the concept of a being that possesses
every unlimited perfection—is logically consistent.
(3) ∴ The ens perfectissimum is possible. [From (1) and (2).]
(4) Existence is a perfection.27
(5) ∴ Existence is contained in the concept <ens perfectissimum>. [From (2) and
(4).]
(6) Whatever is contained in the concept of a possible thing can be predicated of
that thing.
(7) ∴ The ens perfectissimum exists. [From (5) and (6).]
Note, though, that premise (1) is ambiguous in the same way that premise (2) in
Descartes’ argument was. It can be read as having either of two distinct logical forms:
(1*) C is a logically consistent concept ⊃ ◇9xCx.
(1{) C is a logically consistent concept ⊃ 9x(possible(x) & Cx).
(1*) would allow Leibniz to prove that possibly there is an ens perfectissimum or, at
most, the conditional conclusion that if there is one then it exists; (1{) would allow

25
Leibniz’s view is that the concepts of cognition (perception of perfection) and desire (appetition of
perfection) are to be understood through the concept of perfection, not the other way around, so Kant’s
claim is false if it is understood as a claim of conceptual dependence; see DM }1–6.
26
For a particularly vivid Kantian illustration of the idea that there are degrees of being, see Refl. 4244,
Ak. 17: 477–8.
27
There are a number of texts in which Leibniz accepts the premise that existence is a perfection, e.g.,
“Meditations” (“the most perfect being includes all perfections, among which is existence”, AG 25); the
correspondence with Elizabeth (AG 237); NE 437–8; and L 167, 231. However, there are other texts in
which Leibniz accepts that existence is “involved in” God’s essence, but without straightforwardly
identifying existence as one of the perfections.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

him to discharge the antecedent of that conditional (that there is an ens perfectissimum)
and prove that there is an existent ens perfectissimum, but at the price of committing
him to a possibilist ontology of non-actual possibilia. (1{), plus Leibniz’s doctrine
that there are logically consistent concepts of non-actual possible worlds and the
substances that compose them, entails that there are non-actual possible objects.
These are familiar points from the discussion of Descartes’ argument. Interest-
ingly, though, while Leibniz offers a Cartesian version of the ontological argument
(along the lines of (1)–(7) above) in several texts,28 in other texts he rejects this
argument for the very reason that it proves only a conditional conclusion. In a note
from 1678 he writes:
Spinoza reasons thus, following Descartes: It is the same to say that something is contained in
the nature or concept of some thing, as to say that very [predication] is true about that thing (as
it is contained in the concept of a Triangle, or follows from its essence, that its three angles are
equal to two right angles). But necessary existence is contained in the same way in the concept
of God. Therefore it is true about God to say that necessary existence is in him, or that he exists.
To this reasoning, and others like it, it can be objected that all those propositions are
conditional, for to say that three angles equal to two right angles are involved in the nature
or concept of a triangle is to say only that if a triangle should exist, it would have this property.
So in the same way, even if it be granted that necessary existence belongs to the concept of God,
still all that will be inferred from that is that if God should exist, then he would have this
property (of necessary existence), or that if God should exist, he would [exist] necessarily.29

For our purposes, we can ignore the reference to Spinoza and focus on Leibniz’s
critique of the Cartesian argument (which, he claims, Spinoza takes over into his own
philosophy).30 Leibniz’s claim here is interesting for a number of reasons. First, it is
precisely the objection that Gassendi and Caterus raised. Secondly, as we saw
previously, Descartes was not bothered by this objection, because, as he writes in
the Fifth Meditation, “everything which I clearly and distinctly perceive to belong to
that thing really does belong to it.”31 That Leibniz does not adopt the Cartesian
response shows that, in this text at least, he is hesitant to embrace the ontology of
non-actual possibilia to which this would commit him. It is not clear, though, that
this is Leibniz’s settled view on the matter.
This means that Leibniz understood that the ontological argument, as conceived
by Descartes and, in some texts, himself, is committed to a possibilist ontology. In the
rest of this section I want to explore Leibniz’s work on the ontological argument with
this question in mind: does Leibniz succeed in developing a theory of existence to
support the ontological argument, while avoiding this commitment to possibilism?

28
E.g., in several notes from 1676 (DSR 47, 102).
29
A II.i.393. Cf. the discussions of this passage in Adams (1994), 161–2 and Griffin (2013), 40.
30
In the Ethics, a draft of which Spinoza showed Leibniz when they met in 1676 in Amsterdam, Leibniz
presumably has in mind Ip7 and Ip11Dem. For recent work on the complex Leibniz–Spinoza dialectic, see
Newlands (2010), Lin (2012), and Griffin (2013), 58–82; for a comprehensive study, see Laerke (2008).
31
AT VII: 64–6/CSM II: 45.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

Or are Leibniz’s ontological arguments inescapably committed to an ontology of

non-actual possibilia?
The problem that exercised Leibniz during his most intensive period of work on the
ontological argument, his stay in Paris in the 1670s, is how to give an account of what a
‘necessary being’ is and why such a being is possible.32 Abstracting for a moment from
the details of these various arguments, the general idea behind Descartes’ original
argument was that God is a necessary being because existence is ‘involved’ in his
essence, though Descartes may not have provided a satisfactory account of that
‘involvement.’ Leibniz’s main task in amending the Cartesian argument—aside from
finding a way around appealing to the rule of clear and distinct perception—was to
clarify what it means for existence to be ‘involved’ in God’s essence. The “Meditations”
expresses the most straightforward way in which this can be understood: existence is
one of the perfections that make up God’s essence as the ens perfectissimum.
That Leibniz continued to develop more sophisticated models of why an ens
perfectissimum would be an ens necessarium suggests that he found something deeply
wrong with this model, and it is not hard to determine what that might be. The
“Meditations” picture of how God’s essence ‘involves’ existence (existence is simply
one of the perfections) leads to insuperable difficulties, as Gaunilo33 and others had
pointed out. If existence is a ‘first-order’ perfection like others, i.e. the kind of
property that can be a constituent of an essence, this very quickly leads to a
multiplication of necessary entities. The perfection existence must be compatible with
the concept <sandwich>; otherwise, it would be a necessary truth that no sandwiches
exist (which, fortunately, is not true). So the concept <existent sandwich> is a consist-
ent concept. It follows that a sandwich exists, by parity of reasoning with the onto-
logical argument. So there is an ontological argument for the existence of at least one
sandwich. The straightforward explanation of how existence is contained in God’s
essence—it is just one more perfection like all of the others—leads to absurdity.34
An alternate explanation would be that necessary existence is the infinite unlimited
perfection possessed by God, rather than the limited, contingent existence possessed
by finite things; on such a view, necessary existence is a primitive concept, and
contingent existence is a logically complex concept involving a limitation of neces-
sary existence. In some passages Descartes himself appears to favor this model for
understanding how existence is involved in the true and immutable nature of God.35
However, as Adams (1994) points out, this view is subject to an analogue of the
Gaunilo problem. If necessary existence is one of God’s perfections, then it is a
logically atomic property, containing no negations. Consequently, it is compatible
with any other property. Take the essence E of any finite thing and conjoin necessary
existence to it, forming <E + necessary existence>. Since necessary existence is

32
See Griffin (2013) and Adams (1994), 157–76.
33
Gaunilo’s objection to Anselm’s argument can be found in Plantinga (1965).
34 35
Cf. Adams (1994), 150–1. E.g., AT VII: 152, 156.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

logically atomic, the resulting concept is logically consistent, hence there is some-
thing that necessarily exists and has essence E. So now we have a profusion of
ontological arguments, just as we had with Gaunilo. Simply making necessary
existence a perfection in its own right will not help Leibniz understand the involve-
ment of existence in the divine essence.36
For these reasons, during his period of intense work on the ontological argument
in the 1670s Leibniz tried to develop an account of the ‘involvement’ of existence in
the essence of the ens perfectissimum, without assuming that it is a perfection itself.
“For perfections seem to be qualities,” he writes, “as existence is not.”37 He developed
several such accounts, but I will discuss only one of them here,38 for it helps to bring
out the desiderata on any theory of existence that might repair the ontological
argument. Leibniz, while denying that existence is a perfection itself, retains the
connection between existence and the perfections that constitute a thing’s essence.
His idea is that existing is not possessing a certain degree of reality, but a comparative
degree of reality: being more perfect than the alternative possibilities (what Leibniz
calls “mutually incompatible things”).39 The alternative possibilities are not merely
individuals; if they were, Leibniz would be claiming that all of the most perfect
individuals exist, something he consistently denies. God’s choice is, in the first
instance, a choice of which possible world to create (the most perfect one) and so
the “mutually incompatible things” are possible worlds. His idea is that, for a finite
being, to exist is to be part of the most perfect possible world.40 Since God is not part
of a world, but still exists, we can generalize this notion of existence as follows: to exist
is to be part of the most perfect possible maximal state of affairs, where a state of
affairs is maximal just in case any state compatible with it is part of it. This definition
of existence will apply univocally to God and his creatures.41

36
Cf. Adams’s objection to the idea that necessary existence is a primitive perfection (Adams (1994),
151). A possible Leibnizian response to this problem would be to claim that although <existent sandwich>
(or <necessarily existent sandwich>) are consistent concepts they are not essences; the ontological argument
only works for essences, as I have reconstructed it. This is a promising Leibnizian response, but to be
dialectically successful it would need to be backed up with an account of which concepts are essences;
without such an account, it reduces merely to the (unobjectionable but trivial) claim that no sandwich can
essentially exist because no sandwich can have existence in its essence. See Refl. 3706 (Ak. 17: 240–2) for
some evidence that Kant appreciated this point. Thanks to Colin Marshall for pressing me on this point.
37
A II.i.313. There are texts, however, in which Leibniz asserts that existence is a perfection: L 177,
L 231, and NE 358.
38
E.g., I forgo discussion of Leibniz’s idea, discussed in Adams (1994), 151–6, of arguing that the ens
perfectissimum would exist necessarily because it would be conceptually and hence ontologically inde-
pendent of every other being. Given that the ultimate purpose of this book is to understand Kant, ignoring
this Leibnizian strategy is appropriate because Kant would think of any such argument for the necessary
existence of the ens perfectissimum as a cosmological, rather than ontological argument, because it derives
existence from a version of the principle of sufficient reason.
39
A VI.iv, 1354. See the discussion of this passage in Adams (1994), 165 and Griffin (2013), 38.
40
DSR 21, 67, and Adams (1994), 165–7. Cf. “On the Ultimate Origination of Things” (AG 149–55), as
well as Blumenfeld (1973) and Look (2005).
41
That Leibniz has a ‘univocal’ conception of existence is denied by Nachtomy (2012). I agree with
Adams (1994), 170, though, that this would be highly problematic for Leibniz.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

However, this view faces a very serious problem. Whether or not a given series of
finite beings (for instance, the actual ones) is part of the most perfect maximal state of
affairs is a fact determined by their essences. God understands that they are so
compatible and, on the orthodox theological picture, creates them; his creation
does not make them compatible with the most perfect world.42 But this means the
envisaged theory of existence entails that actually existing creatures exist in virtue of
their essences, not in virtue of God’s creative activity, his will. This is unacceptable
because Leibniz, following theological orthodoxy, holds that God is the only being
who exists in virtue of his essence. One of Leibniz’s consistent objections to Spinoza
is that Spinoza denies God’s will, and thus his goodness and wisdom, any role in
explaining the existence of finite creatures (modes of the one Spinozistic substance);
for Spinoza, finite modes are a necessary consequence of the eternal nature of
God, the one substance. The theory of existence as ‘compatibility with the most
perfect possible maximal state of affairs’ has the same unacceptable Spinozistic
consequence.
For our purposes, though, the most important point is that on this conception,
existence is not a ‘first-order’ property that can be contained in an essence (namely,
God’s), it is the ‘higher-order’ property of having perfections (properties) that make
an object part of the most perfect possible maximal state of affairs.43 Adams (1994)
stresses the importance of this ‘higher-order’ conception of existence and interprets
Leibniz as anticipating, in his Paris notes, what is usually taken (in my view,
correctly) as a Kantian insight: that existence is a predicate that applies in the
first instance to concepts themselves. Existence is the predicate that applies to a
concept if and only if it is instantiated by an object. As we saw above, there are
texts from this period that suggest that Leibniz was at least hesitant to adopt a view
on which existence is a property possessed by some possible objects but not by
others. However, there are really three theses about existence that need to be
distinguished here:
(Higher-order) Existence is higher-order, either by being a predicate of concepts
(the predicate of being instantiated) or a higher-order property of objects (e.g., of
being comparatively more perfect than other objects).
(Quantifier) Existence is fundamentally a predicate of concepts (e.g., the predicate
of being instantiated), not of objects. In contemporary terms, it is a quantifier.44
(Actualism) There are no non-existent objects.

42
For more on Leibniz’s view that possibility depends upon God’s understanding, while existence
depends upon his will, see Newlands (2013). The grounding of possibility and actuality in divine faculties is
also an important theme in Kant’s modal theory; see Ch. 4.
43
By ‘higher-order’ property I will mean a property of having certain specified properties; e.g., redness
is a first-order property while the property ‘having at least one color property’ is a higher-order property.
44
One can (as e.g., Quine did) hold that existence is a quantifier without holding that it is a second-
order predicate of concepts. For the purposes of this book I will be identifying the view that existence is a
quantifier with the view that it is second-order.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

As I argued in Chapter 1, Kant holds the second two doctrines; since (Quantifier)
entails (Higher-order), he is also committed to the first. Adams convincingly shows
that Leibniz must hold that existence is higher-order to avoid the problems that arise
when existence is treated like any other predicate (e.g., the Gaunilo problem), but this
does not by itself show that he took it to be a predicate of concepts (much less that he
endorsed actualism). While there is some textual evidence that Leibniz experimented
with thinking of existence as a second-order predicate (a quantifier), and, what is
more, the ‘widest’ quantifier there is (actualism), this does not show it was ever his
settled view.45
Looking back over Leibniz’s various attempts to repair the ontological argument,
we can see him trying to develop a theory of existence that meets certain desiderata,
among which are:
(Univocal) There is a univocal sense of ‘exists’ in which both God and finite
creatures exist. There may be more specific senses of ‘exist’ (e.g., necessary
existence) that apply only to one or the other, but there must be a common
sense of ‘exists.’
(Difference) In this univocal sense of ‘exists,’ God exists essentially, but none of
his creatures do. It follows from God’s essence that he exists, and God is the only
being of whom this is true.
(Higher-order) Existence is higher-order, either by being a predicate of concepts
(e.g., the property of being instantiated) or a higher-order property of objects (e.g.,
of being comparatively more perfect than other objects).
However, Leibniz was unable to develop a theory of existence that met these
desiderata and, as a result, was unable to repair the ontological argument satisfac-
torily. In the next section I will argue that, on one reading, Baumgarten succeeded
where Leibniz did not.

2.4. Baumgarten’s Ontological Argument

The natural next step would be to consider Christian Wolff ’s ontological argument
and whether it commits him to possibilism. However, as I noted in section 1, Wolff ’s
own ontological argument in Theologia naturalis is an uninteresting derivative of
Leibniz’s “Meditations” argument. Furthermore, as Kant himself points out, Wolff ’s
definition of existence as the “completion of possibility” [complementum possibilita-
tis] is too indeterminate to ascertain what, precisely, his theory was.46 I therefore pass

45
Cf. Adams (1994), 158–64 and Mates (1986), 170–88.
46
“The Wolffian definition of existence, that it is a completion [Ergänzung] of possibility, is obviously
very indeterminate. If one does not already know what over and above possibility can be thought in a thing
[über Möglichkeit in einem Dinge kann gedacht werden], one will not learn it from this definition” (OPG,
Ak. 2: 76). For more on Wolff ’s view of existence, see Ont. }}173–4. Honnefelder (1990), 363–71 concurs
that Wolff adds little to what is found in Leibniz and Baumgarten.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

over Wolff and consider Baumgarten, whose theory of existence is more developed
than Wolff ’s and is the target of constant criticism by Kant.47
Baumgarten gives an ontological argument for the existence of God in Metaphy-
sica }}803–9 that is largely derivative of Leibniz. This may explain why Baumgarten
has not received the attention of scholars of the ontological argument that, I think, he
deserves.48 In addition to this relatively uninteresting variant of the Leibnizian
argument, though, Baumgarten also developed an innovative theory of existence as
complete determinacy:
Apart from essence, (}53) the possible is either determinate in respect of all affections which
are compossible in it, or not (}34, 10). In the first case, it is ACTUAL [actuale]; in the second case
it is called PRIVATIVE (merely possible) NON-BEING (nihil, cf. }7).49
EXISTENCE [EXISTENTIA] (actus, cf. }210, actualitas) is the totality [complexus] of all affections that
are compossible in a being, i.e. the complement of the essence, or internal possibility, insofar as
these are considered only as a totality of determinations (}40).50

The parentheses are Baumgarten’s own references to previous definitions and prin-
ciples within Metaphysica. As we have seen, ‘determination’ is a perfectly general
term for Baumgarten: it refers to any predicate that can be attributed to a being
(}34).51 Every possible being (ens) is fully determinate with respect to its essential
properties and attributes. In addition to essential properties and attributes, beings
also have accidental properties, which Baumgarten divides into intrinsic accidents
(modes) and relational accidents (}37). Each being is either fully determinate in
respect of its intrinsic accidents (modes), or it is not. In the former case it is actual/
exists; in the latter, it is merely possible, non-actual/non-existent. Following Baum-
garten in Metaphysica }55 (see previous quotation), I will henceforth identify exist-
ence and actuality. For a being to exist is for it to possess a fully determinate set of
modes (intrinsic accidents) in addition to its essential properties and attributes.
Baumgarten is committed, therefore, to at least the following principle:
(Ex-Determinacy) For all x, x exists if and only if, for any intrinsic predicate A,
either x is A or x is ~A.52

47
Baumgarten’s theory of existence is (implicitly or explicitly) criticized by Kant in Refl. 5760, 5783,
5784, 5786, 6245, 6255, 6290, and 6322; in the metaphysics lectures, MV (Ak. 28: 413), MvS (Ak. 28: 503),
ML2 (Ak. 28: 554), MD (Ak. 28: 630), MK2 (Ak. 28: 722–4, 779), and Volck.RT (Ak. 28: 1156), as well as the
more familiar discussions in the CPR and OPG (see the next chapter). It is even alluded to in the CJ (Ak. 5:
475–6).
48
Symptomatic of this is Proops (2015), 13–15, who does not consider the obvious rejoinders Baum-
garten could make to (Proops’s reconstruction of) Kant’s objection.
49
Meta. }54. 50
Meta. }55.
51
Thus it does not have the specifically Kantian meaning of “a predicate which goes beyond the
concept of the subject and enlarges it” (A598/B626).
52
The second disjunct means that x has the negative property ~A; I discuss the difference between
having the negative property ~A, and merely lacking A, and the related distinction between negations of
predicates and negations of propositions, subsequently.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

By itself, this is a very minimal claim. It says only that there are no existents that are
indeterminate with respect to one or more intrinsic predicates. Since Baumgarten
identifies existence with complete intrinsic determination, henceforth I will drop the
explicit restriction to intrinsic predicates; when I talk about accidental determin-
ations in Baumgarten, I mean only accidental intrinsic determinations (modes).
At various places in Metaphysica Baumgarten hints at an argument for God’s
existence, based on this theory of existence as complete determination. For instance:
[ . . . ] God is also determined in regard to the rest of his internal perfections as much as
anything whatsoever can be determined in regard to an internal perfection. Therefore, God is
actual (}54).53
A non-actual God would be a being enjoying all realities, and yet missing some reality.
In regard to all internal perfections, he would be determined as greatly as a being can be
internally determined, and yet, in regard to some of these perfections he would not be so
determined (}54).54

For the sake of readability, I have eliminated many of Baumgarten’s in-text citations to
other parts of the Metaphysica, except for the final references in both passages to }54,
quoted above, where he introduces his complete determination theory of existence.
Very roughly, Baumgarten’s idea in the first passage is that God is essentially com-
pletely determinate with respect to every intrinsic predicate; if A is an intrinsic
predicate, God either essentially has A (e.g., if A is an infinite unlimited perfection
or a conjunction thereof) or God essentially lacks A (e.g., if A contains or entails any
limitation of perfection). God’s essence has no ‘room’ to be ‘filled out’ with intrinsic
accidents. In the second passage, Baumgarten puts the point negatively: if God did not
exist, he would be not completely determinate (by the definition of existence), but
since God is essentially completely determinate, this is absurd. I will focus on the direct
version of the argument, which we can provisionally reconstruct as follows:
Baumgarten’s Complete Determination Argument, Preliminary Reconstruction:
(1) The concept <ens perfectissimum> is logically consistent (contains no contra-
dictions). [This was proven earlier, on essentially Leibnizian grounds—see
Meta. }}806–9.]
(2) Anything that is logically consistent (contains no contradictions) is possible.
(3) ∴ The ens perfectissimum is possible. [From (1) and (2).]
(4) Every possible being has all of the properties contained in, and entailed by, its
essence (essential properties and attributes).
(5) The essence of the ens perfectissimum entails that it is completely determinate
with respect to every intrinsic predicate.
(6) Complete intrinsic determinacy is existence.
(7) ∴ The essence of the ens perfectissimum entails that it exists.

53
Meta. }819. 54
Meta. }823.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

(8) ∴ The ens perfectissimum exists. [From (4) and (7).]

(9) Anything whose essence entails its existence is a necessary being.
(10) ∴ The ens perfectissimum is a necessary being. [From (8) and (9).]
One potential source of skepticism about this argument is that, from a Kantian
perspective, the definition of existence on which it rests (premise (6)) fails to
distinguish between concepts and objects. Is this a specification of which concept
<existence> is, or does it specify necessary and sufficient conditions on existent
objects, or both? Consequently, there are two ways of understanding Baumgarten’s
definition of existence/actuality and two ways of understanding this argument. On the
first reading—which I will call the ‘intensional’ reading—Baumgarten is distinguishing
between two different ways of conceptualizing one and the same set of objects: we can
think of those objects as merely possible, or we can think of them as actually existing. On
the ‘extensional’ reading, by contrast, the distinction between the possible and the actual
is not merely a conceptual (or ‘rational’) one. On this reading, Baumgarten is distin-
guishing ‘possible’ and ‘actual’ in respect of their extensions; he is distinguishing between
two kinds of objects, those that are merely possible and those that actually exist.
If the intensional reading is correct, then the complete determination argument
is quite weak. Consider, again, the problem that marred many earlier ontological
arguments. Premise (2) can be understood in either of two ways:
(2*) Concept C contains no contradictions ⊃ ◇9xCx.
(2{) Concept C contains no contradictions ⊃ 9x(Cx & possible(x)).
For what should now be familiar reasons, the argument is invalid if we assume only
(2*), while it is committed to a possibilist ontology if we assume (2{). However, the
intensional reading of Baumgarten’s theory of existence is incompatible with a possibi-
list ontology, for it allows only a conceptual, rather than extensional, distinction
between possible and existent objects. This leaves Baumgarten with (2*), which will
allow him to prove only the conditional conclusion that if there is a God, he exists.
One way for Baumgarten to overcome this defect in his argument would be to
appeal to a principle like:
(DI) Concept C is a fully determinate concept of an object ⊃ 9xCx.
This principle is fully compatible with the intensional reading, and hence with
actualism, and would allow Baumgarten to prove almost trivially that there exists
an ens perfectissimum. However, this principle is subject to the fatal objection that, as
Kant never tired of pointing out when lecturing from the Metaphysica, there are in
principle completely determinate concepts that are not instantiated, for instance,
completely determinate concepts of non-actual possible worlds.55 I take this to be a

55
MV (Ak. 28: 413), MvS (Ak. 28: 503), ML2 (Ak. 28: 554), MD (Ak. 28: 630), MK2 (Ak. 28: 722–4, 779),
and Volck.RT (Ak. 28: 1156); cf. Refl. 5760, 5783, 5784, 5786, 6245, 6255, 6290, and 6322.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

devastating objection to DI and to Baumgarten’s complete determination argument

for God if the ‘intensional’ reconstruction is correct (i.e. if there is only a conceptual
distinction between possible and actual objects).
The obvious weakness of Baumgarten’s argument on the intensional reading
suggests that another reading is possible. Baumgarten, after all, was no fool; Kant
even refers to him as “the excellent analyst Baumgarten” (A22/B36). On the exten-
sional reading of Baumgarten’s theory of existence, the distinction between the actual
and the possible is an extensional distinction between two sets of objects, one of
which (the actual) is a subset of the other (the possible). On this way of understand-
ing Baumgarten’s theory of existence, there is a domain of possible objects and each
object in the domain has an essence. For objects other than God, these essences are
incompletely determinate; for some intrinsic predicate A, they do not contain (or
entail) either A or ~A. Merely possible objects (possibilia) have only the properties
contained in or entailed by their essences, and the essences of finite beings are
incompletely determinate. Consequently, finite possibilia are incompletely determin-
ate. They have their essential properties and their attributes, but no accidents.
Existing finite beings have their essential properties and their attributes, but are
also equipped with a fully determinate set of accidents. The extensional reading,
therefore, attributes to Baumgarten the familiar principle:
(Possibilism) ◇9x(¬exists(x)).
In fact, it attributes to Baumgarten the view that there are non-existent objects
(beings, entia).
The extensional reading allows us to reconstruct Baumgarten’s argument in such a
way that it is not subject to the obvious Kantian rejoinder that there are completely
determinate but uninstantiated concepts. To do so, it must distinguish between
concepts in general and essences: concepts that contain only the essential properties
of a being. It must appeal, not to:
(2{) Concept C contains no contradictions ⊃ 9x(Cx & possible(x))
from above, but to a principle of the form:
(2{{) ◇9xCx & C is an essence ⊃ 9x(Cx & possible(x)).
While not uncontroversial, this principle is a theorem in many modal systems.56
Intuitively, it says that if it is possible for there to be some being with a certain
essence, then there is such a being, although it may not actually exist. It entails that
there could not be ‘more’ possible beings that there are, although ‘more’ of the merely
possible beings could have the predicate of existence.

56
It is a consequence of the Barcan Formula: (x)□Ax ⊃ □(x)Ax. For an influential recent defense of the
Barcan Formula and its converse, see Williamson (2013).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

With these pieces in position, we can now give an extensional reconstruction of

Baumgarten’s argument. To bring out its complete logical structure, I have expanded
the preliminary reconstruction from earlier and made explicit some enthymematic
premises (letting EP abbreviate the concept <ens perfectissimum>):
The Complete Determination Argument (Extensional reconstruction):
(1) The concept EP is logically consistent.
(2) Concept C is logically consistent ⊃ ◇9xCx. (Note: this principle is the
unproblematic one from above, which is not committed to possibilism. So
far, this argument has not made use of any possibilist ontology.)
(3) ∴ ◇9xEPx. [From (1) and (2).]
(4) The concept EP is an essence; anything that falls under the concept EP is
essentially an ens perfectissimum.
(5) ◊9xCx & C is an essence ⊃ 9x(Cx & possible(x)). [Note: this is (2{{) from
above.]
(6) ∴ There is a being whose essence is EP and this being is possible. [From (4)
and (5).] Call this being g.
(7) For any object x and any concepts E and F, if E is the essence of x,
necessarily Fx if and only if the proposition Ex entails the proposition Fx.
[Note: this is just the logicist theory of necessary predicates; see Metaphysica
}}34–8.]
(8) For any intrinsic predicate A, if Ag is logically compatible with EPg, EPg
entails Ag.
(9) For any intrinsic predicate A, if Ag is not logically compatible with EPg, EPg
entails (~A)g.
(10) ∴ For any intrinsic predicate A, EPg entails Ag or EPg entails (~A)g. [From
(9) and (10).]
(11) ∴ For every intrinsic predicate A, Eg entails Ag or (~A)g. [From (10); if Ex
entails either of the disjuncts, it entails the disjunction.]
(12) ∴ Eg entails, for any intrinsic predicate A, Ag or (~A)g. [From (11).]
(13) For all x, x exists =def for any intrinsic predicate A, Ax or (~A)x. [Note: this
is Baumgarten’s theory of existence as complete determinacy.]
(14) ∴ Eg entails g exists. [From (12) and (13).]
(15) ∴ Necessarily g (the ens perfectissimum) exists. [From (7) and (14).]
Intuitively, this argument shows that the ens perfectissimum essentially exists
because the ens perfectissimum has a completely determinate essence, and being
completely determinate is equivalent to existence, according to Baumgarten. The
ens perfectissimum has no ‘gaps’ in its essence to be filled in with accidents.57

57
Proops (2015) is thus incorrect when he claims that, “according to Baumgarten, existence is
something external to the concept of a thing” (14). Existence is external to the essences of finite things
(it is a mode); it is contained in the essence of God, the unique ens necessarium. Proops’s reconstruction of
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

By contrast, finite creatures have incompletely determinate essences. When they

are actualized or brought into existence this incompletely determinate set of essential
properties is supplemented with a set of accidents, making them completely deter-
minate. Finite creatures do not exist necessarily and there are no sound ontological
arguments for their existence. To see this, observe that a crucial premise of the
complete determination argument—(8)—does not hold where x is a finite object and
E is its essence:
(8*) For any intrinsic predicate A, if Ax is logically compatible with Ex, Ex
entails Ax.
This is false where x is a finite creature and E is its essence, because for some intrinsic
predicates A, Ex is logically compatible with Ax but does not entail it; for some A, Ex
entails neither Ax nor (~A)x.
The validity of this ontological argument shows that Baumgarten’s theory of
existence satisfies the three Leibnizian desiderata on a theory of existence:
(Univocal) There is a univocal sense of ‘exists’ in which both God and finite
creatures exist. There may be more specific senses of ‘exist’ (e.g., necessary
existence) that apply only to one or the other, but there must be a common
sense of ‘exists.’
(Difference) In this univocal sense of ‘exists,’ God exists essentially, but none of
his creatures do. It follows from God’s essence that he exists, and God is the only
being of whom this is true.
(Higher-order) Existence is higher-order, either by being a predicate of concepts
(e.g., the property of being instantiated) or a higher-order property of objects (e.g.,
of being comparatively more perfect than other objects).
It supplies a univocal theory of existence because there is a common sense in which
God and finite creatures exist: they are fully determinate. I explained in the previous
paragraph why it satisfies the difference requirement. By taking existence to be a
‘higher-order’ property of objects (the property of having a fully determinate set of
accidents), it avoids the problems that plagued Descartes and Leibniz when they took
existence to be one perfection among others that can partly constitute an essence
(Gaunilo’s objection and variants thereof). For instance, there is no corresponding
ontological argument for the existence of a ‘completely determinate island’ because a
completely determinate concept of an island is not an essence; no completely

Kant’s objection to Baumgarten—existence cannot be contained in the concept of the ens realissimum
because it is not contained in the concept of finite things—is ineffective. Further, Proops is too hasty in
claiming that Baumgarten’s ontological argument is subject to Caterus’s objection that the ontological
argument proves only a connection between concepts in thought, not the existence of a thing outside
thought. Kant’s own response to the Caterus objection could be made by Baumgarten himself; see Refl.
3706, Ak. 17: 240–1. Although I have left out the details of Wolff ’s theory, I do not see why Wolff could not
make the same response as well.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

determinate island essentially has the completely determinate set of accidents con-
tained in that concept. It is no surprise, then, that Kant devotes so much time and
effort to refuting Baumgarten’s theory of existence as complete determination; the
ontological argument based on this theory is, of all of the ontological arguments we
have canvassed so far, the strongest.
This argument relies on the extensional reading of Baumgarten’s theory of exist-
ence. We might then wonder whether this is the correct reading. But recall that the
distinction between the extensional and the intensional reading of Baumgarten’s
theory arose when we imposed a sharp distinction between concepts and objects, or,
equivalently, between distinguishing concepts in virtue of their intensions (concep-
tual content) and in virtue of their extensions (the objects that fall under them). Since
Baumgarten does not rigorously maintain this distinction (nor, arguably, does
anyone before Kant), there is no reason to think that his theory is determinately
either extensional or intensional. It can be read in either way.
In fact, there is evidence that Kant was aware of this ambiguity in Baumgarten’s
theory. For instance, in Beweisgrund, he writes:

[ . . . ] the proposition that a possible thing, regarded as such, is indeterminate with respect to
many of its predicates, could, if taken literally, lead to serious error. For such indeterminacy is
forbidden by the law of excluded middle, which maintains that there is no intermediate
between two predicates which contradict one another. It is for example impossible that a
man should not have a certain stature, position in time, age, location in space, and so forth.
Our proposition [that existence is complete determination] must rather be taken in the
following sense: the predicates which are thought together in a thing in no way determine
the many other predicates of that thing. Thus, for example, that which is collected together in
the concept of a human being as such specifies nothing with respect to the special character-
istics of age, place, and so forth. But then this kind of indeterminacy is to be found as much in
an existing thing as it is in a merely possible thing. (Ak. 2: 76).

Kant’s point in the first half of this passage is that the distinction between the
incompletely and the completely determinate cannot be a distinction between two
kinds of objects, because incompletely determinate objects would violate the prin-
ciple of excluded middle. Instead, Kant argues, the distinction is between objects and
concepts: objects are fully determinate but concepts of them can be incompletely
determinate. I read this passage as Kant remarking on exactly the ambiguity in
Baumgarten’s theory that I have noted: read extensionally, it runs afoul of the
principle of excluded middle, and read intensionally it provides no basis for a
sound ontological argument.
One might think that the extensional version of Baumgarten’s theory of existence
is decisively refuted by Kant’s point: if an object x is not determinately A and not
determinately ~A, this violates the principle of excluded middle. But how to apply the
principle of excluded middle to predicates is a controversial matter in eighteenth-
century logic, and at least one prominent philosopher of this period denies that the
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

principle of excluded middle for propositions (p∨¬p, for any p) entails the corres-
ponding principle for predicates (Ax ∨ (~A)x, for any x and any A)—Kant himself!
As I have argued elsewhere,58 the Critical Kant does not hold that the principle of
excluded middle as applied to predicates—what Kant calls, tellingly, the “principle of
complete determination”—is a logical principle at all.
In this book, I do not have the space to explain why, in the Critical period, Kant
denied that the principle of complete determination is a logical principle. I want
merely to explain why it is coherent to deny that the principle of excluded middle
entails the principle of complete determination. The principle of excluded middle (in
the case of simple subject-object predications) is:
(PEM) For any predicate A and object x, Ax ∨¬(Ax)
while the principle of complete determination (in the case of such predications) is:
(PCD) For any predicate A and object x, Ax ∨ (~A)x.
The difference is in their second disjunct. The second disjunct of the PEM is what
Kant calls a ‘negative judgment.’ Applied to the judgment Ax it says that it is not true
that x falls within the extension of the predicate A. The second disjunct of the PCD is
what Kant calls an ‘infinite judgment.’ It says that x falls within the extension of the
predicate ~A. The negation in a negative judgment applies to a whole judgment,
while the negation in an infinite judgment applies to a predicate; I have used two
different negation strokes and positioned the parentheses to bring out these differ-
ences in scope. We might assume that the negative judgment ¬(Ax) entails the
infinite judgment (~A)x. We might, in other words, assume that if x is not in the
extension of A, then x is in the extension of ~A. But that is equivalent to assuming
that the extension of ~A is the complement of the extension of A, which is equivalent
to assuming that the principle of complete determination is true. So the PEM alone
does entail the PCD. Kant denies that, as a matter of logic, the extension of
a predicate A and its negation ~A must be exhaustive; no logical laws are
violated if an object x belongs neither to the extension of A nor to the extension of
~A. This means that the following represents a logical possibility, according to the
Critical Kant:
(1) ¬Ax & ¬((~A)x).
In this case, object x is indeterminate with respect to predicate A.
In its extensional version, Baumgarten’s theory of existence entails that where x is a
merely possible object and A is an intrinsic accidental predicate, x is indeterminate
with respect to A. I am not going to argue that Baumgarten would analyze this as (1)
or that he would follow Kant in distinguishing between infinite and negative

58
Stang (2012).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

judgments, and deny that the principle of complete determination follows from the
principle of excluded middle. Whether or not the historical individual Alexander
Gottlieb Baumgarten ever held the extensional version of his theory of existence, and
whether he would defend it by distinguishing PCD from PEM as I have done above,
the philosophical fact remains: such a theory and such an argument are positions
within logical space, and, what is more, may be stronger than any ontological
argument given by Leibniz or Descartes (or Wolff). We have already seen that
Kant offers objection both to the extensional and to the intensional versions of
Baumgarten’s theory of existence. Given that Kant’s ambition is to prove that no
ontological proof whatsoever is possible, this is entirely the correct procedure.
However, this entails that he must be able to refute whatever the best version of
Baumgarten’s theory is, including the version that adopts Kant’s own strategy for
maintaining PEM while abandoning PCD.59 In the remainder of this chapter I will
reconstruct in some detail Kant’s arguments that existence is not a ‘real predicate’
with the following question in mind: do they carry any force against the most
defensible version of his theory of existence that Baumgarten could have given
(whether or not he actually did)? If the answer to that question is no, the most we
will be able to claim on Kant’s behalf is that he refuted several historical ontological
arguments, but that he did not show (as he claimed to have) that ontological proofs as
such are impossible.

2.5. The Argument in Beweisgrund: Leibniz

Having argued in }1.5 and }1.6 that ontotheism as such is committed to the claim
that <exists>60 is a real predicate, and in the first half of this chapter that the
ontological arguments actually given by Descartes, Leibniz, and Baumgarten assume
that it is, I will now reconstruct Kant’s arguments that existence is not a real
predicate, starting with Beweisgrund.
After a brief introduction, the first ‘Reflection’ of the first Division of Beweis-
grund 61 begins with Kant’s argument that existence is not a real predicate, which
I here divide to clarify its internal structure:
[A] This proposition [that existence is not a predicate or a determination of a
thing] seems strange and absurd, but it is indubitably certain. Take any
subject you please, for example, Julius Caesar. Draw up a list of all the

59
Note that my point holds even if the argument of Stang (2012) is wrong and Kant does not deny that
PCD is a logical principle; the availability of that view in logical space means that Kant must refute it to
refute the in-principle possibility of ontological proofs.
60
In what follows I will sometimes drop the angled brackets, for ease of comprehension.
61
Beweisgrund consists of three Divisions (Abteilungen), which are themselves divided into a series of
‘Reflections,’ which consist in one or more numbered sections. For ease of reference I will refer to parts of
Beweisgrund by the Division number, followed by the Reflection number, followed by the section number;
e.g., II.3.2 will refer to the second Division, third Reflection, second section.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

predicates which may be thought to belong to him, not excepting even those
of space and time. You will quickly see that, with all of these determinations,
he can either exist, or not exist. The Being who gave existence to the world
and to our hero within that world, could know every single one of these
predicates without exception, and yet still be able to regard him as a merely
possible thing which, in the absence of that Being’s decision to create him,
would not exist.
[B] Who can deny that millions of things which do not actually exist are merely
possible from the point of view of all the predicates they would contain if they
were to exist? Or who can deny that in the representation which the Supreme
Being has of them there is not a single determination missing, although
existence is not among them, for the Supreme Being cognizes them only as
possible things? It cannot happen, therefore, if they were to exist, they would
contain an extra predicate; for, in the case of the possibility of a thing in its
complete determination, no predicate at all can be missing.
[C] And if it had pleased God to create a different series of things, to create a
different world, that would have existed with all the determinations, and no
additional ones, which he cognizes in it [die er an ihr doch erkennt], although
that world is merely possible. (Ak. 2: 72)
As I read this passage, [A] contains Kant’s primary argument that existence is not a
determination. He first claims that existence is not contained in the complete concept
of an object, the concept that contains every (real) predicate that applies to the object
(“draw up a list of all the predicates which may be thought to belong to him”).62 He
concludes from this that existence is not a (real) predicate, a determination. [B]
might be thought to add something, because Kant there explicitly considers merely
possible objects, but in my reconstruction of [A] below I will show that this thought is
already implicit in [A]. In [C] Kant shifts gears slightly, and rephrases the argument
in [A] in terms of possible worlds, rather than possible individuals (e.g., Caesar). In
the next two sections Kant expands upon and explicates this claim (e.g., his discus-
sion of absolute and relative positing) but does not give any substantial further
argument for it. If we want to understand Kant’s argument that existence is not a
determination we must examine [A] closely.63
That Kant focuses on complete concepts of finite substances (e.g., Caesar) is
puzzling. He ultimately needs this argument to refute the ontotheist view that the
existence of the unique infinite substance, God, follows logically from its essence.
On the ontotheist picture, essences of finite substances are compatible with
existence, but do not contain or entail existence; this is why finite substances exist

62
This may not refer to the predicates that actually belong to Caesar, but to all of the predicates that
could possibly apply to Caesar. I consider that possibility below.
63
The last sentences of the paragraph quoted above expand slightly upon this argument, but do not
contain additional arguments.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

only contingently, while God is the unique necessary being. So complete concepts of
finite substances are, strictly speaking, irrelevant to the ontotheist view of the
necessary existence of God. Ontotheists are not as such committed to affirming, or
denying, that the complete concept of a finite substance determines whether it exists
or not; they are committed, at most, to denying that the essence of a finite substance
determines whether it exists or not.64 This passage is especially puzzling in light of
my interpretation; if the point of Kant’s objection is to deny that there are non-actual
possible objects, how are complete concepts of actual objects at all relevant to his
argument?
One explanation for this puzzling fact would be that Kant identifies an object’s
essence and its complete concept. If this were Kant’s view, he could argue that
existence is never contained in a complete concept, and then validly conclude that
it is never contained in an essence. However, to interpret Kant as identifying essences
and complete concepts would be to attribute to him a form of essentialism even
stronger than Leibniz’s, for Leibniz, who holds that no individual could exist with a
different complete individual concept, distinguishes between essences and complete
individual concepts. Since it is questionable whether Kant accepted even that weaker
Leibnizian doctrine, it is unlikely that he identifies essences of finite substances with
their complete concepts.
Alternatively, it might also be thought that Kant is simply arguing from a stronger
claim to a weaker one: since a complete concept contains all the predicates of an
object, and an essence contains a subset of them, if its complete concept does not
contain existence, then neither does its essence. But the ontotheists in question
(Descartes, Leibniz, Wolff, Baumgarten) already agree that the existence of a finite
substance does not follow from its essence! On this issue, Kant and the ontotheist are
in full agreement, so it cannot be the point of his argument. The ultimate aim of
Kant’s argument is to show that not even God’s existence follows from his essence. In
order to derive that conclusion, Kant would need to argue first that God’s complete
concept does not contain (or entail) existence, which he notably does not do, focusing
instead on the complete concepts God has of finite substances.
Finally, it might be thought that Kant is arguing that Leibniz’s view that the
complete concept of any individual (e.g., Julius Caesar) contains all of his (real)
predicates entails that existence is not a real predicate (not a determination). Kant
could be reasoning that the predicates contained in a complete individual concept
(CIC) are the real predicates (determinations) of the individual and that these are the
predicates God thinks of the object as having when he thinks of that individual as a
denizen of some possible world, as the member of a candidate maximal plan for

64
Leibniz holds that, in some sense, existence is ‘contained’ or ‘involved’ in the complete individual
concepts of actually existing substances: an infinite analysis of such a concept and an infinite comparison of
those concepts with concepts of other possible worlds God could have created would reveal why these
individual substances, and not others, are actualized. See Ch. 1.2.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

creation.65 If existence is a real predicate, then it is contained in God’s CIC of Julius

Caesar as a possible being. So God already thinks of him as existing in thinking of
him as possible, which entails, absurdly, that God cannot think of any individual as
merely possible. Spelled out fully, this reconstruction of Kant’s reasoning would go as
follows:
(1) All of the real predicates Julius Caesar has if he exists are contained in God’s
CIC of him. [“The Being who gave existence to the world and to our hero
within that world, could know every single one of these predicates without
exception [ . . . ].”]
(2) If existence is a real predicate, then if Julius Caesar exists he has the real
predicate of existence.66
(3) ∴ If existence is a real predicate, then existence is contained in God’s CIC of
Julius Caesar. [From (1) and (2).]
(4) If God’s CIC of Julius Caesar contains existence, then God’s CIC of Julius
Caesar represents him as existing.
(5) ∴ If existence is a real predicate, then God’s CIC of Julius Caesar represents
him as existing. [From (3) and (4).]
(6) But God’s CIC of Julius Caesar does not represent him as existing; it represents
him as possible, as existing only if God wills to create him. [“[ . . . ] and yet still
be able to regard him as a merely possible thing which, in the absence of that
Being’s decision to create him, would not exist.”]
(7) ∴ Existence is not a real predicate. [From (5) and (6) by modus tollens.]67
This is a very weak argument, which has no dialectical force against Leibniz or any
other ontotheist. First of all, why would Leibniz admit (1)? Why cannot Leibniz hold
that the CIC of an individual contains all of the real predicates it would have if it were
to exist except existence itself (and predicates that logically entail existence)? On such
a modified Leibnizian view, God’s CICs of possible individuals contain all of their
predicates except the predicate they obtain in virtue of being created by God:
existence. That Leibniz himself fails to make this qualification to his doctrine of
CICs is dialectically irrelevant; Kant would have succeeded at best in showing that

65
What I mean by calling possible worlds ‘maximal plans for creation’ is, first of all, that Leibnizian
possible worlds have their being in the content of God’s thoughts; unlike Lewis worlds, they are not
concrete entities. Secondly, they are maximal: if w is a world and p is logically consistent with w (w does not
entail ¬p) then w contains p (w entails p). See L 661–2. Though my interpretation of Kant’s views on
existence and his arguments for those views are diametrically opposed to his Meinongian reading, I have
benefited greatly from reading the work of Tobias Rosefeldt and from discussing these issues with him. See
Rosefeldt (2008) and (2011).
66
Strictly speaking, this should read: if <exists> is a real predicate, etc. For the sake of readability,
though, I leave out the angled brackets here.
67
The premises of this argument are unaffected if we substitute for ‘Julius Caesar’ the name of some
non-actual individual substance, hence the [B] portion of the text does not really add anything not already
found in [A] (see previous discussion).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

Leibniz himself is not a particularly effective advocate for his own ontological
argument. Even the ontotheist can admit that, while existence is a real predicate, it
is a real predicate with a very different status than the other real predicates, some-
thing of which Leibniz was, in fact, quite aware.68
Even more problematic, though, is claim (6). Why must Leibniz think that God’s
CIC of Julius Caesar does not represent him as existing? God’s CIC of any individual
substance represents what properties that substance would have if God were to create
it. Presumably, existence is one of those properties. But how then, one might ask, can
God also think of the individual substance as being merely possible, as being a
candidate for creation, rather than already actually created? Compare this to an
ordinary case: I can think of how my friend Joe would look if he were bald while, at
the same time, thinking that he is not bald. My thought can represent the non-actual
but possible situation of Joe’s being bald while also representing Joe as being actually
not bald. Likewise, God’s concept of an individual can represent that individual as
existing in that it represents the individual as it would be if it were to exist but does not
represent that individual as actually existing. The error here is thinking that from:
(a) God’s concept of X represents X as F and God is infallible
it follows that:
(b) X actually is F.
This does not follow. For infallibility does not preclude God from representing
possibilities that he does not thereby represent as actual. On the contrary, God’s
omniscience requires that he does represent non-actual possibilities and represents
them as non-actual possibilities!
It is possible that Kant had this anti-Leibnizian argument in mind. If so, he failed
to make any advance against ontotheism or the possibilist ontology. However, as
I have reconstructed the argument, he is arguing against an opponent who is
committed to the Leibnizian view that God’s concepts of individual substances qua
possible are fully determinate, i.e. that God thinks of each possible substance as
belonging to at most one fully determinate possible world. This assumption is
encoded in the argument in premise (1). If God’s CIC of Julius Caesar does not
restrict him to the actual world, then his CIC must include ‘world-indexed’ infor-
mation: it must include the information that JC has one set of real predicates in one
possible world, and a distinct set in another world.69 For instance, if Julius Caesar is
bald in world w but not bald in world w*, God’s CIC of Caesar does not represent
Caesar as bald simpliciter; it represents him as bald in w and not-bald in w*.70 So the

68 69
See Ch. 2.3 and Adams (1994), 157–76. Cf. Adams (1994), 71–4.
70
Recall Kant’s precise words: “draw up a list of all the predicates which may be thought to belong to
him [alle seine erdenkliche Prädicate], not excepting even those of space and time.” This could mean: think
of all of the complete sets of predicates that Caesar could have in different worlds.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

argument as stated applies only against someone who accepts Leibniz’s doctrine that
an individual exists in at most one possible world. But this is not a view that Kant has
any reason to expect would be shared by all ontotheists, much less by all philosophers
who think that existence is a real predicate. It would have the effect of making
his argument apply only against Leibniz himself. However, a modified version of
Kant’s argument can be given that does not encode this Leibnizian assumption:
(1*) For any world w in which Caesar would exist and any real predicate
F Caesar would have in w, God’s CIC of Julius Caesar contains F in w.
(2*) If existence is a real predicate, then, for any world w in which Caesar would
exist, Caesar would have the real predicate exists in w.
(3*) If existence is a real predicate, then for any world w in which Caesar would
exist, God’s CIC of Caesar contains exists in w. [From (1*) and (2*).]
(4*) If God’s CIC of Julius Caesar contains exists in w, then God’s CIC of Julius
Caesar represents him as existing in w.
(5*) ∴ If existence is a real predicate and w is a world where Caesar would
exist, then God’s CIC of Julius Caesar represents him as existing in w. [From
(3*) and (4*).]
(6*) God’s CIC of Julius Caesar does not represent him as existing in w, for any
world w; it represents him as possible, as existing only if God wills to create
him.
(7*) ∴ Existence is not a real predicate. [From (5*) and (6*) by modus tollens.]
But, as with the previous argument, premise (6*) is false. By representing JC as
existing in w, God is not thereby representing JC as existing simpliciter; he is
representing JC as being such that he would exist if he (God) were to actualize w.
But, it will be pointed out, one of the worlds at which God’s CIC represents Caesar
as existing is the actual world. So we can replace (6*) and (7*) with the following:
(6{) The actual world @ is a world where Caesar would exist.
(7{) ∴ If existence is a real predicate then God’s CIC of JC represents him as
existing in @. [From (5*) and (6{).]
(8{) If God’s CIC of Julius Caesar represents him as existing in @ then God’s
CIC of JC represents him as actually existing.
(9{) But God’s CIC of Julius Caesar does not represent him as actually existing; it
represents him as possible, as existing only if God wills to create him.
(10{) ∴ Existence is not a real predicate. [From (7{)–(9{), by modus tollens.]
However, this argument is invalid. Its appearance of validity rests on conflating two
subtly different senses of ‘actual’, the first in (8{), and the second in (9{). It is worth
going into detail about them, because the distinction will be important for under-
standing Kant’s critique of Baumgarten as well. In the first sense—call it the ‘rigid’
sense of ‘actual’ (actualR)—‘the actual world’ is just a name for this world, the world
that God in fact created. Earlier, I introduced the name ‘@.’ In any counterfactual
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

context, ‘@’ just refers to this world. In the second sense—call it the non-rigid sense
(actualNR)—‘actual’ refers to whatever is created by God. This is the sense in which, if
God had created a different world, things that are actual would not have been actual
(which is false for actualR). I am not sure I can give a satisfactory analysis of the
meaning of actualityNR, but perhaps it will help to understand these notions if we note
that they are materially equivalent to one another and to p itself, since this is the actual
world (and the truth of p is evaluated with respect to the world in which it is asserted):
(*) ActuallyR(p) $ actuallyNR(p) $ p
However, this equivalence only holds contingently, because the two operators behave
differently in modal contexts. In particular, actualityR(p) does not change in truth
value in different possible worlds:
(ActualR) ActuallyR(p) is true in w if and only if p is true in @.
No matter which world God created, it would actuallyR be the case that, for instance,
Caesar exists. Consequently, actuallyR(p) is not necessarily equivalent in truth value
to p itself. If God had not created Caesar, Caesar exists would be false, but actuallyR
Caesar exists would be true. By contrast, the truth value of actuallyNR(p) changes in
different possible worlds:
(ActualNR) For any world w, actuallyNR (p) is true in w if and only if p is
true in w.
Consequently, in any possible world, p is true if and only if actuallyNR(p). If God had
created a different possible world, some propositions that are actuallyR the case
would not be actuallyNR the case. When we consider the possibility of God creating
some world other than @, we are considering the possibility that @ might not have
been actualNR.
To return to the argument, we can now see that (8{) is true only if it involves the
rigid notion of actuality:
(8{R) If God’s CIC of Julius Caesar represents him as existing in @ then God’s
CIC of Caesar represents him as actuallyR existing.
while (9{) is true only if it involves the non-rigid notion of actuality:
(9{NR) But God’s CIC of Julius Caesar does not represent him as actuallyNR
existing; it represents him as merely possible, as existing only if God wills to create
him.
Intuitively, (9{NR) says that God’s CIC of Julius Caesar does not prejudge whether
God creates @ or not; it does not represent him as actually existing in the non-rigid
sense. Since it is consistent for God to represent Julius Caesar as actuallyR existing
(existing in @) without representing him as actuallyNR existing, there is no conflict
here and (10{) does not follow by modus tollens.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

2.6. The Argument in Beweisgrund: Baumgarten

A better explanation of Kant’s focus on completely determinate concepts of possi-
bilities is that his unnamed target is Baumgarten. But this raises the question of
why Kant would direct his objections at Baumgarten specifically, rather than to all
ontological arguments as such. In section 4 I argued that there are reasons to think
that Baumgarten’s ‘complete determination’ argument is the strongest ontological
argument fielded by any of Kant’s ontotheist opponents (or at least, among those
known to Kant).71 Furthermore, there is good evidence that Kant thought of
Baumgarten in these terms, for Kant returns repeatedly, in published and unpub-
lished writings, to the complete determination theory of existence, in order to refute
it. One possible explanation of this is that he agrees that Baumgarten’s complete
determination argument is ‘the argument to beat.’
If we read the ‘Julius Caesar’ passage quoted previously with Baumgarten in mind,
a natural Kantian objection emerges: the completely determinate concept of Julius
Caesar does not represent him as existing. One can think of an object as completely
determinate as one wishes without thereby thinking of it as an existing object. This is
the reason he introduces God and God’s thoughts about possibilia into the argument:
our finite minds cannot form a genuinely completely determinate concept.72 The
argument is that God can have a completely determinate thought about some
possible object, without representing it as an existing object. If existence were
equivalent to complete determination, then God’s thought of that object would
represent it as existing. Therefore, existence is not equivalent to complete determin-
ation, contra Baumgarten. This is why Kant emphasizes that, although God has fully
determinate concepts of non-actual possible individuals (he “could know every single
one of these predicates without exception, and yet still be able to regard him as a
merely possible thing”), he nonetheless does not thereby represent those individuals
as existing.
While this argument may initially appear damning for Baumgarten’s position, in
fact Baumgarten can easily reply. Baumgarten needs to explain how God can have a
thought that represents its object as being completely determinate without repre-
senting it as existing.73 The expression ‘a thought that represents the object as being
so-and-so’ could mean either a thought that represents the object as actuallyNR being
so-and-so, or one that represents the object as it would be, if it were so-and-so. To use
an example from earlier, I can think, of Joe, that he is actuallyNR bald, or I can think

71
It may be that one of the versions of the ontological argument Leibniz worked on but never published
is stronger than Baumgarten’s, though the critical discussion in Adams (1994) and Griffin (2013) does not
make that seem very likely.
72
Without talking about God, Kant would have to formulate his argument in an awkwardly counter-
factual form: per impossibile, were I to form a fully determinate concept, etc.
73
In this section, I always mean ‘actually’ in the non-rigid sense. See previous section for more on the
rigid and non-rigid senses of ‘actually.’
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

of Joe as he would be if he were bald. Both thoughts could be described as the

thought ‘of Joe as bald,’ but this description conceals an important difference: the
first thought is a thought about (non-rigid) actuality, and the second is a thought
about possibility. The thought of Joe as he would be if he were bald is not merely
the thought that Joe is possibly bald, or that Joe could be bald. It is a thought
about how Joe would be in respect of appearance and hairline if that possibility
were (non-rigidly) actual.
Similarly, God can think about Caesar either as actuallyNR being fully determinate,
or God can think about Caesar as he would be if he were fully determinate. In fact, he
can do the latter in infinitely many ways, because there are infinitely many ways
Caesar could be completely determinate because there are infinitely many fully
determinate sets of accidents he could have (contra Leibniz). Baumgarten can
accommodate the claim that God can have existentially neutral thoughts that
represent their objects as completely determinate by interpreting this to mean
that God can have thoughts about merely possible and hence indeterminate
objects that represent those objects as they would be if they were completely
determinate, i.e. if they were actualNR. Nothing in the contents of these thoughts
requires God to think that these objects are actuallyNR completely determinate.
Given that God is omniscient, he can only think of actualR things as actuallyNR
being completely determinate; if he were to think of a merely possible object as
actuallyNR being completely determinate, he would be misrepresenting it. Exist-
entially neutral but completely determinate thoughts about possibilia pose no
problem for Baumgarten. This takes the wind out of the sails of Kant’s objection
to Baumgarten.
Over the course of this chapter and the previous one we have seen three Kantian
objections to Baumgarten’s ‘complete determination’ ontological argument:
1. There are completely determinate but uninstantiated concepts.
2. By denying the principle of complete determination, the extensional version of
Baumgarten’s theory violates the principle of excluded middle.
3. God can have completely determinate concepts of non-actual individuals
without representing them as existing.
In section 4 I explained why the first objection carries little or no weight
against Baumgarten. I argued further that, contra the second objection, Kant
gives us no reason why Baumgarten cannot follow Kant himself in denying that
the principle of excluded middle entails the principle of complete determination.
I have just argued that the third objection fails as well. I conclude, therefore, that
in Beweisgrund Kant fails to establish that Baumgarten’s complete determination
theory of existence is false and thereby fails to refute his ‘complete determin-
ation’ version of the ontological argument. This means that, at least in that early
text, Kant fails to refute Baumgarten’s version of the doctrine that existence is a
real predicate.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

2.7. The Argument in the Critique of Pure Reason

In the CPR Kant also argues that existence is not a real predicate, in the famous section
“On the impossibility of an ontological proof of the existence of God.” Within that
section, the arguments that have received the lion’s share of scholarly attention are the
‘hundred dollars’ argument and the other arguments in that paragraph (A599/B627)
and the next paragraph on the following page (e.g., the ‘missing reality’ argument).
I think that those arguments add little that was not already present in Beweisgrund. For
reasons of space, therefore, I forgo further discussion of them here.74
However, that section also contains an apparently distinct objection, which I will
call the ‘no contradiction’ objection, that some readers may think is a stronger
objection, at least to logicist ontological arguments. I will argue, on the contrary,
that the independence of this objection from the claim that existence is not a real
predicate is merely apparent, and that it does not carry any additional force against
the logicist.
Kant states the non-contradiction objection in a relatively straightforward passage:
If I cancel the predicate in an identical judgment
and retain the subject, then a contradiction
results; hence I say that the former necessarily pertains to the latter. But if I cancel the subject
together with the
predicate, then no contradiction results; for there is no longer anything that
could be contradicted. To posit a triangle and cancel its three angles is contradictory; but to
cancel the triangle together with its angles is not a contradiction. It is exactly the same with the
concept of an absolutely necessary being. If you cancel its existence, you cancel the thing itself
with all its predicates; where then is the contradiction supposed to come from? [ . . . ] But if you
say, God is not, then neither omnipotence nor any other of his predicates is given; for they are
all canceled together with the subject, and in this thought not the slightest contradiction shows
itself. (A594–5/B622–3)

Kant’s point is that if we take an analytic judgment like all triangles have three angles
and negate it, we get:
(1) Some triangles do not have three sides
which, given the definition of <triangle>, is equivalent to:
(2) Some (three-sided plane figures) do not have three sides
and this is clearly a violation of the PNC. Thus, we have a very natural account of why
analytic judgments are necessary: their negations violate the PNC. However, if we
take an existential judgment like God exists and negate it, we never get a violation of

74
For readers interested in a reconstruction of those arguments, see the supplementary article on the
arguments of the CPR on my website (see Notes on the Text). Notice that both the ‘hundred dollars’
argument and at least one of the arguments in the subsequent paragraph on A600/B628 (“when I think a
thing, through whichever and however many predicates I like (even in its thoroughgoing determination)”)
are both arguments about completely determinate concepts. So it is prima facie plausible that they restate
the basic argument of Beweisgrund, reconstructed above.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

the PNC. So it is never logically impossible that something fail to exist, not even God.
Although Kant does not make this objection in Beweisgrund it is implicitly contained
in his observation that the proposition that nothing whatsoever exists does not entail a
contradiction;75 it is not a point wholly original to the CPR.
This Kantian point can seem especially strong if we interpret it as the logical point
that a negated existential statement of the form:
(3) ¬9xFx
does not entail a contradiction, and, what is more, can be added to any consistent set
of propositions that lack existential import without producing an inconsistency. If we
take Kant’s target to be the logicist view that God exists with logical necessity then
this might appear to be a knock-down objection: no claim of the form 9xFx is
logically necessary because its negation (3) does not entail a contradiction. Since
this objection does not prima facie depend on controversial claims about whether
existence is a determination or real predicate it might seem to be Kant’s most
straightforward and best objection to the ontological argument.
Kant claims that judgments of the form there are no Fs do not entail contradic-
tions. Without going into detail about the nature of the entailment relation, I take it
that it is uncontroversial that p entails q if and only if there is a proof of q from p
using only logical principles, and everything here depends upon what counts as a
logical principle. While Kant does not yet, in 1763, have his developed Critical theory
of the formality of ‘pure general logic’, his claim in Beweisgrund that the proposition
nothing exists is not a violation of the PNC (the highest principle of logic) is an
anticipation of one important aspect of it.76 For if nothing exists does not violate the
PNC, and the PNC is the highest principle of logic (the principle from which all other
logical principles flow, either directly or indirectly), then logic alone does not
guarantee that anything exists, that is, that there are any objects in the extensions
of any concepts. In contemporary logic, this means that empty domains are not
excluded on purely logical grounds. The same result holds in Kant’s Critical logic.
(Pure general) logic, writes Kant in the CPR, abstracts entirely from the relation of
cognition to objects;77 this means not only that logic abstracts from the differences
among objects, but logic abstracts from whether concepts have objects at all, that is,
whether they have null extensions. So logic, for both the pre-Critical and
Critical Kant, does not guarantee that there are any objects; no existentially
committal propositions can be among the principles of (pure general) logic properly
so called.

75 76
Ak. 2: 78. Ak. 2: 78.
77
A55/B79. I am not claiming that pure general logic does not contain a certain concept of an object;
I am claiming that pure general logic does not assume that there are objects in the sense under discussion
here: objects of quantification (see next section).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

This might be thought devastating to the logicist, for each of the ontological
arguments I reconstructed in the previous chapter crucially relies on what is now
called a ‘comprehension’ principle, a principle that allows us to infer that a certain
concept has a non-empty extension:
Descartes: I clearly and distinctly understand essence E ⊃ 9x(E is the essence of x & possible(x))
Leibniz: C is a logically consistent concept ⊃ 9x(possible(x) & Cx)78
Baumgarten: C is a logically consistent concept & C is an essence ⊃ 9x(possible(x) & Cx)

These are crucial components of their ontological arguments because they allow one
to infer from a claim about the essence or concept of God (left-hand) to the claim
that there is an object that falls under that concept (right-hand). However, from a
Kantian point of view, these are not logical principles. They concern the conditions
under which a certain concept has a non-empty extension (e.g., when it is logically
consistent) and logic abstracts entirely from whether concepts have non-empty
extensions.
But the logicists need not agree with Kant about what logic is. As other scholars
have argued,79 Kant’s thesis that logic is formal (that it abstracts entirely from
objects) is an innovation in the rationalist logical tradition; there is no reason to
expect that Descartes, Leibniz, or Baumgarten would share it. The logicists can
plausibly claim that certain comprehension principles (like those above), or certain
principles about the domain of merely possible being, are logical principles, and
propositions that are incompatible with them are logically impossible. The logicists,
after all, conceive of logic as describing the metaphysical space of possible beings and
their essences. They can plausibly claim that it is a logical principle that there is a
domain of possible beings and the widest quantifier ‘there is’ (9x) ranges over (takes
as the admissible values of its bound variables elements in the set of) possible beings,
including the merely possible ones; consequently, ~9x(x=x) is logically impossible
and there may be other ‘existentially’ quantified propositions whose negations are
logically impossible (notably, 9xGod(x)). The logicist are not bound to agree with
Kant that logic abstracts entirely from objects.
But notice that the crucial logicist move is to claim that logic does not abstract
from objects but considers all possible objects as such and their principles (which are
logical principles). This view rests on the assumption that there is a distinction
between objects in general (possible objects) and existent objects; in other words, it
rests on the assumption that existence is a determination, a particular way objects
can be. For instance, the comprehension principles listed above become highly
implausible if we interpret the quantifier expression 9x as expressing existence; in

78
I suspect that the true Leibnizian comprehension principle is something like: C is a complete concept
⊃ 9x(possible(x) & Cx), but that did not play an explicit role in my reconstruction of Leibniz’s ontological
arguments.
79
McFarlane (2002) and Lapointe (2012).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

Leibniz’s case they would entail that every consistent concept is instantiated by an
existent object.
We are led here to a conclusion very similar to the one I drew in the previous
chapter when we considered whether the claim that existence is not a real predicate
means merely that existence is a quantifier expression that applies in the first instance
to concepts (and at best derivatively to objects): these alternate interpretations of
Kant’s objection deliver decisive objections to ontotheism only if they are supple-
mented with the further claim that there are no non-existent objects (on my inter-
pretation, the meaning of the claim that existence is not a determination). That claim
is Kant’s fundamental objection to ontotheism. In fact, the very organization of the
corresponding section in the CPR supports this. Kant begins with the “no contradic-
tion” objection at A594–5/B622–3 and, aware that this alone will not quiet the resolute
ontotheist, states rather huffily: “I would have hoped to annihilate this over-subtle
argumentation without any digressions through a precise determination of the concept
of existence, if I had not found the illusion consisting in the confusion of a logical
predicate with a real one (i.e. the determination of a thing) nearly precludes all
instruction” (A598/B622). Kant only then makes his famous claim that “being is
obviously not a real predicate.” I interpret the structure of this section, and Kant’s
huffiness, to mean that the decisive objection, the objection that will finally eliminate
ontological arguments, is the claim that existence is not a real predicate. Kant’s other
objections (that existence is in the first place predicated of concepts, not of objects; that
no negative existential is logically contradictory) are parasitic on that claim.

2.8. Absolute Positing

Having discussed in so much detail Kant’s negative thesis about existence (it is not a
real predicate), it is worth briefly discussing his positive theory of existence. Since
Kant’s theory of existence in the CPR is tied up with distinctively Critical doctrines
(most importantly, the distinction between concepts and intuitions), I will explore
the positive theory of existence in Beweisgrund, specifically, where Kant makes a
distinction between two kinds of positing:
The concept of positing [Position] or setting [Setzung] is perfectly simple: it is identical with the
concept of being in general. Now something can be thought as posited merely relatively, or to
express the matter better, it can be thought merely as the relation (respectus logicus) of
something as a characteristic mark of a thing. In this case, being, that is to say, the positing
of this relation, is nothing other than the copula in a judgment. If what is considered is not
merely the relation but the thing posited in and for itself, then this being is the same as
existence. (OPG, Ak. 2: 73)

Kant here distinguishes two forms of positing, which I take to be two ways in which
being can be asserted: relative positing (in which the relation of two predicates is
asserted in a judgment) and absolute positing (in which an object is posited as an
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 IS EXISTENCE A REAL PREDICATE ?

instance of a predicate). Relative or ‘copulative’ positing asserts that whatever falls

under one predicate falls under another, without asserting that any object instantiates
either predicate. To take Kant’s example a few paragraphs later:
If I say ‘God is omnipotent’ all that is being thought is the relation between God and
omnipotence, for the latter is a characteristic mark of the former. Nothing further is being
posited here. Whether God is, that is to say, whether God is posited absolutely or exists, is not
contained in the original assertion. (OPG, Ak. 2: 74)

The relata in this case are the predicates <God> and <omnipotent>; the former
relatum cannot be God himself because the judgment God is omnipotent can be
made whether or not the concept <God> is instantiated, i.e. whether or not there is a
God to stand in such a relation. In fact, Kant goes on to say, we can make judgments
involving concepts that cannot possibly be instantiated, like <the God of Spinoza>.
In absolute positing, by contrast, we assert that there is an object instantiating a
concept (a set of predicates):
If I say ‘God is an existent thing’ it looks as if I am expressing the relation of a predicate to a
subject. But there is an impropriety in this expression. Strictly speaking, the matter ought to be
formulated like this: ‘Something existent is God.’ In other words, there belongs to an existent
thing those predicates which, taken together, we designate by means of the expression ‘God.’
These predicates are posited relative to the subject, whereas the thing itself, together with all its
predicates, is posited absolutely. (Ak. 2: 74)

When we posit absolutely we assert that some concept, e.g., <God>, is instantiated by
an object, and in so doing we also copulatively assert that all of the predicates contained
in the concept pertain to that object as well: anything that instantiates <God> also
instantiates <omnipotent>, <omniscient>, etc. Every absolute positing brings with it a
relative positing (“these predicates are posited relative to the subject [concept], whereas
the thing itself, together with all its predicates, is posited absolutely”).80
This means we need to distinguish and account for all of the following cognitive
acts:
(i) Absolute positing: there is an object that instantiates the concept F.
(ii) Judging: anything that falls under concept F falls under concept G.
Absolute positing, as Kant characterizes it and as I have defined it, is general: it says
that some object or other instantiates a concept. But we also need to be able to judge
of particular objects that they instantiate concepts (subsume them under concepts).
So we need to add to our inventory of cognitive acts:
(iii) Subsumption: object x instantiates concept F.

80
Cf. Abaci (2008).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

IS EXISTENCE A REAL PREDICATE ? 

Otherwise, we would be able to judge (to use another Kantian example) there are
narwhals and narwhals have horns but not this object is a narwhal (in the presence of
a narwhal). But the ability to perform the cognitive act I have called ‘subsumption’
involves the ability to think of particular objects (e.g., using the demonstrative ‘this
object’), which I will call:
(iv) Acquaintance: being cognitively in contact with an object x so that it can
figure in an act of subsumption.
Kant’s 1760 works do not contain a developed theory of these different cognitive acts
and their relations; for that we must wait until the Critique of Pure Reason in 1781.
For instance, one question we might have is: can we understand an act of absolute
positing (there is an F) without understanding what acquaintance with such an object
(with an x that is an F) would be? I will explore Kant’s Critical answers to these
questions in the second half of this study. But before we continue we can use the
notion of ‘absolute positing’ to answer a question that some readers may have: what
does ‘object’ mean? The concept of object I have been working with is (in Kantian
terms) the concept of an object of absolute positing.81 My interpretation of the
doctrine that existence is a real predicate—there are no objects that do not exist—
can be restated in these terms as: there are no objects of absolute positing that do not
exist. To posit an object and to assert that it exists are one and the same act. Hence
Kant’s claim that: “if what is considered is not merely this relation [between predi-
cates] but the thing posited in and for itself, then this being is the same as existence”
(OPG, Ak. 2: 73).82

81
A forerunner of the quantificational notion of ‘object.’ See the discussion of Rosenkoetter and Forgie
and the ‘Frege-anticipation’ thesis in Ch. 1.6.
82
Cf. A593–601/B621–9.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

3
Real Conflict, Real Grounds,
Real Possibility

3.1. Introduction
In the previous chapter we examined Kant’s reasons for rejecting the logicist explan-
ation of divine necessity. If no existential proposition can be logically necessary, as
Kant argues, there is at least one proposition—that God exists—that is necessarily
true, but not logically necessary. Although this entails that logicism is, strictly
speaking, false, it leaves much of logicism intact. For instance, the existence of God
might be the only exception to the logicist theory of necessity; all non-existential
necessarily true propositions might, nonetheless, be logically necessary. That would
leave the logicists without a perfectly general account of possibility and necessity, but
it might leave them with an elegant account of modality for finite beings. To a
contemporary audience less concerned with vindicating the traditional doctrine of
divine necessity, this may not seem like much of a problem at all, and Kant’s attack
on logicism as a whole might appear quite weak. In this chapter I examine the second
strand of Kant’s rejection of logicism: his rejection of the logicist doctrine that logical
contradiction is the only source of metaphysical incompatibility between predicates.
The logicist view of possibility is:

(LogicismP) It is possible that p if and only if p does not entail a contradiction.

In the case of predicates of one and the same subject, this entails that:
(LogicismPRED) For predicates P and Q, it is possible that there is an object x with
P and Q if and only if the proposition 9x(Px & Qx) does not entail a contradiction.

I will refer to the relation of being possibly co-instantiated as being metaphysically

compatible and the relation of being logically possibly co-instantiated as being
logically compatible; logicism entails that these relations are co-extensive. Sometimes
I will refer to the former relation simply as compatibility.
In the pre-Critical works of the 1760s Kant claims that there are logically com-
patible predicates that are not metaphysically compatible, or what I will call really
incompatible predicates. For instance, he writes in Beweisgrund that: “the impene-
trability of bodies, extension, and such like, cannot be attributes of that which has
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

understanding and will” (OPG, Ak. 2: 85). However, Kant does not offer a direct
argument for this claim; elsewhere he appeals to an “immediate judgment of the
understanding” that “forces one to admit” that understanding and will “together with
the greatest possible reality can co-exist in one and the same being” (OPG, Ak. 2: 88).
Presumably, another “immediate judgment of the understanding” is supposed to
provide our access to the incompatibility of thought and extensions. But these bald
assertions of real compatibility and incompatibility are not likely to satisfy his logicist
opponent, for whom there must be some hidden logical incompatibility between the
concepts of thought and extension (and a lack of such logical incompatibility
between the concepts of understanding and will).
In this chapter I reconstruct Kant’s argument that there is real incompatibility,
focusing on Beweisgrund and Kant’s 1763 essay, Attempt to Introduce the Concept of
Negative Magnitudes into Philosophy.1 In Negative Magnitudes he distinguishes
between the ‘logical conflict’ (when two predicates are logically incompatible) and
the ‘real conflict’ of predicates. Two predicates stand in ‘real conflict’ just in case they
are both realities (they are not mere absences of positive determination) and, when
they are co-instantiated by a single object, each of them causes the effect the other
would have on its own to be canceled. Kant’s primary example is that of two forces
of equal magnitude, but opposite direction, acting on the same body. Each force
cancels the motion that would be caused by the other force, if that force were the only
force acting on the body. But realities stand in real conflict only if they are possibly
co-instantiated by the same object; realities in real conflict must be metaphysically
compatible. So even if Kant were to prove that there is real conflict between
predicates, this would not show that there is real incompatibility between predicates.
Far from entailing metaphysical incompatibility, real conflict between predicates
entails that they are compatible.2
In this chapter I explore how Kant uses the phenomenon of real conflict to show
that there are really incompatible predicates. The key idea is that real conflict is a
causal relation; predicates stand in real conflict when they mutually causally cancel
one another’s effects. Cause, in eighteenth-century German metaphysics, is usually
taken to be one species of the more general notion of ground, and I begin, in section 2,
by exploring the logicist theory of grounding, and in particular, of causation. In
section 3 I explain one of the key elements of Kant’s pre-Critical metaphysics: the
distinction between logical grounds and ‘real’ grounds and his argument that causes
are real grounds of their effects. In section 4 I show how this distinction between
logical and real grounds is the missing premise in Kant’s argument that logicism
cannot allow for real conflict between predicates, and in section 5 I reconstruct his
argument that, if there is real conflict between predicates, then there are really

1
The topic is also broached in Beweisgrund, but not discussed as thoroughly as in NG, so it is on that
work I will focus here; see Ak. 2: 86.
2
As Abaci (2014) correctly points out.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

incompatible predicates (though not the predicates that are in real conflict—see
previous discussion). I conclude with a brief discussion of ‘real’ possibility.
Before continuing, I want to make a note about Kant’s notion of a ‘predicate’
[Prädicat], which, from a contemporary perspective, plays two distinct conceptual
roles in Kant’s pre-Critical thought: (a) a metaphysical or ‘worldly’ role, in which a
predicate is something in the world (e.g., a force in a substance), rather than the
content of a representation, akin to a property in contemporary usage;3 and (b) a
‘semantic’ role, in which predicates are concepts and can be combined in judgments.4
There are at least two ways to interpret Kant on this point: (i) he is conflating two
kinds of things that are (correctly) distinguished in contemporary philosophy,
properties and concepts; or (ii) he is making the ambitious and interesting claim
that the contents of our thoughts are also worldly items, i.e. that we think ‘with’
properties, the very items that metaphysically constitute and characterize objects.5
While I find (ii) intriguing, I do not have the space here to articulate or develop this
idea.6 It would be premature, though, to simply adopt (i) and convict Kant of a
conflation. Rather than try to resolve this problem, which would require answering
some difficult philosophical questions (What exactly are concepts? Why should we
assume that properties do not make up the contents of thoughts?), I will simply talk
about predicates as Kant does: as ‘amphibious’ entities that are both worldly—they
include physical forces and can be causally efficacious—and semantic—they have an
internal logical structure and can be combined in judgments. In some contexts the
semantic role will be primary; in others, the worldly role. In both contexts I will refer
to the relation between an object and a predicate that object has as instantiation.

3.2. Logical Grounds

‘Ground’ is a very general notion in the pre-Kantian German rationalist tradition.
The definition that perhaps best captures its wide variety of meanings is that given by

3
NM, Ak. 2: 171, 172, 176, 193, and 200. Things Kant describes as predicates include: forces, feelings of
displeasure, and tendencies. These are not mere representational contents in our (or the divine) mind: they
are real determinations in things.
4
See OPG, Ak. 2: 76, 77, 78, 80 (where he identifies the predicates fiery and body as concepts), 82, 83,
and 156, and Prize, Ak. 2: 284. It might be objected that ‘predicate’ in logical contexts just means the second
term of the judgment, but any concept can be the predicate (including the subject itself). See Ch. 1.3 and
A69/B94.
5
One way to resolve this difference in Kant’s uses of the term ‘predicate’ would be to claim that
predicates-as-properties are semantic-representational items in the mind of God; to say that some object
has predicate/property F is just to say that it is subsumed under God’s concept of F. However, I will argue in
the next chapter that this cannot be right: God grounds all possible predicates but does not do so through
his intellect; his intellectual representation of predicates presupposes their real possibility, and so I do not
think we can analyze predicates as contents of divine thoughts. Yong (2014) defends such an ‘intellectual’
grounding of possibility.
6
The idea that the space of thinkable content and the space of possibility are one and the same might be
found in Hegel’s Science of Logic; for an interpretation of Hegel, and Kant’s Beweisgrund, along these lines
see Yong (forthcoming), ch. 4.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

Christian Wolff: “the ground is that through which one can understand why some-
thing is.”7 Kant argues that this definition is tacitly circular, because “why a thing is”
means the ground of that thing.8 Nonetheless, this definition, unlike more specific
definitions given by individual thinkers, has the generality (indeed, possible empti-
ness) appropriate to this concept. A ground is something that explains why a thing is,
or why it has the predicates it does.9 ‘Thing’ here should be understood very generally
as any possible being whatsoever: there are grounds of the truth of propositions, of
predicates had by objects, alterations in substances, of the existence of substances, etc.
A standard distinction in eighteenth-century theories of grounds is between what
I will call ‘epistemic grounds’ and ‘explanatory grounds’ (what Kant calls ‘ante-
cedently determining grounds’).10 Wolff, Crusius, and Kant all draw this distinction,
but use different terminology which, in some cases, is then borrowed by other
philosophers to mean something different, which creates ample opportunities for
confusion;11 I have introduced my own terms to avoid that confusion. The distinc-
tion concerns the grounding relation and what is grounded, not the ground itself; one
and the same thing can be an epistemic ground and an explanatory ground. An
epistemic ground provides justification for believing some proposition to be true;
when the proposition is believed on the basis of that reason, this belief is justified.
Most of these thinkers focus on the sub-class of epistemic grounds that are grounds
of knowledge: when a belief is formed on the basis of these grounds, the belief counts
as knowledge (assuming some background conditions are met).12 For instance,
smoke is an epistemic ground for knowing that there is fire. I will follow them and
temporarily leave out of consideration epistemic grounds that confer an epistemic
status short of knowledge.13 An explanatory ground is simply something that explains
why something is (or is the case) where the explanandum (the consequence, that
which is grounded) is not the epistemic status of a belief. An explanatory ground of
fire, for instance, is something that explains why there is fire (e.g., the cause of the fire),

7
Dt.Met. }29; cf. Meta. }14 and Crusius’s discussions of Wolff ’s and Leibniz’s definitions of ‘ground’ in
the footnotes to De Usu }1.
8
ND, Ak. 1: 393; cf. MH, Ak. 28: 11.
9
Since there are also grounds of the existence of objects, and existence is not a real predicate, we cannot
restrict ‘predicate’ here to real predicates. See Chapter 1. However, in this chapter I will be concerned
almost exclusively with real predicates; henceforth, unless otherwise noted, ‘predicate’ refers only to real
predicates (determinations).
10
This distinction is not exclusive; it is possible for the explanatory ground also to be an epistemic
ground.
11
ND, Ak. 1: 392–3. The most notable instance of this is that, as Kant himself points out, his distinction
between ‘real’ and ‘logical’ grounds is very different from Crusius’s distinction between ‘real’ and ‘ideal’
grounds (Ak. 28: 12, 37); in fact, Crusius himself borrows the ‘real’ and ‘ideal’ distinction from Leibniz and
Baumgarten; see Weg }139, 142; Ent. }}34–35; De Usu }37; Th. }66; and Meta. }212. For critical discussion
see Watkins (2005), 74–8, 162–5.
12
In fact, it is common in this period to refer to what I am calling ‘epistemic grounds’ as
Erkenntnisgründe.
13
Until Chapter 9, where Kant’s theory of epistemic grounds that are not Erkenntnisgründe will play an
important role in reconstructing his Critical attitude to his pre-Critical theistic proof.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

rather than why I (or anyone else) would be justified in believing there is a fire. There
are different kinds of explanatory grounds, corresponding to different kinds of
explanation, including logical grounds, causal grounds, and essential grounds. The
very same thing can be both the ground of knowledge and an explanatory ground; for
instance, the existence of fire both causally grounds smoke and is an epistemic ground
of knowing that there is smoke.14 It is important to note that the notion of explanation
here is not an epistemic one: that x is an explanatory ground of y does not entail that x
is a ground to believe that y exists (is an epistemic ground of belief that y exists) and
it does not entail that x is an explanation of y that we are in a position to grasp or to
understand. It does not, in other words, entail that we could explain y by citing x. The
explanatory (ground-consequence) structure of reality is a metaphysical structure
that potentially outstrips human capacities for grasping these explanations (finding
grounds for given consequences).
In this chapter, though, my interest will be mainly in what I have called
‘explanatory grounds’,15 so, unless otherwise noted, ‘ground’ means explanatory
ground, or in Kant’s terminology, an antecedently determining ground. Various
kind of explanatory grounds are distinguished by different thinkers in this period,
including (efficient) causal grounds, logical grounds, as well as what Crusius calls
‘real-existential grounds.’16 The first two I will discuss here, focusing on causal
grounds; I discuss the third (real-existential grounds) briefly in Chapter 7.
I defined logicism in Chapter 1 as set of views about modality shared (I argued) by
Leibniz, Wolff, and Baumgarten. However, it is not clear that these three thinkers
shared a common set of views about grounds in general; Wolff and Baumgarten
define ground in very similar ways, but by itself this is not decisive.17 They differed
over crucial issues in the metaphysics of causation (e.g., the nature of pre-established
harmony), and this may indicate a deeper disagreement over the nature of grounds
as such.18
Fortunately, we have abundant evidence, from his lecture transcripts, of how Kant
himself read Baumgarten’s general theory of grounds, and there are reasons to think
he would have extended this interpretation to Wolff (and perhaps even Leibniz,
though I will not consider Leibniz here). Baumgarten defines ground in Metaphysica
}14 as “that from which it is knowable why something is,” and, in his own copy of

14
This is what Crusius calls ‘real-ideal grounds a priori’: a real ground because it is explanatory, and an
ideal ground a priori because it is a ground of knowing not merely what is the case (a posteriori) but why
(a priori). See Weg }142 and Ent. }35; for a similar idea in Baumgarten, see Meta. }311.
15
Crusius and Baumgarten: real grounds; Kant: antecedently determining grounds.
16
For Crusius’s notion of ‘real-existential grounds’ see Ent. }}37, 79.
17
Wolff: “the ground [Grund] is that by which one can understand why something is” (Dt.Met. }29).
Baumgarten: “a ground [ratio] is that from which it is knowable why something is” (Meta. }14).
18
Wolff upheld pre-established harmony only between souls and bodies (as a solution to the mind–
body problem), while Baumgarten, in this respect the more orthodox Leibnizian, extended it to all
substances. See Watkins (2005), 45–50, 74–8, and the translator’s Introduction to Baumgarten (2004),
xiii–xiv.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

Metaphysica, Kant adds the marginal comment: “a ground is either analytic (logical)
or synthetic (real ground)” (Refl. 3504, Ak. 17: 28), a comment which he expands
upon at length in the Herder transcripts of his metaphysics lectures from the early
1760s (roughly simultaneous with the composition of Negative Magnitudes and
Beweisgrund): “all grounds are either logical, in which the consequence, which is
[partly—NS] identical to [the ground] as a predicate, is posited through the law of
identity; or real, in which the consequence is not posited by the law of identity and is
not [partly—NS] identical with the ground” (MH, Ak. 28: 11).19 The distinction Kant
draws here between logical and real grounds is a distinction in the ‘connection’
(nexus) between ground and consequence, the relation by which the ground posits
the consequence (MH, Ak. 28: 12). A logical ground posits its consequence by the
‘law of identity’ (which I read as: the principle of non-contradiction) because the
consequence is a predicate contained (immediately or mediately) in the concept of
the ground; this means that if the ground were present, and the consequence were
not, a contradiction would result. In the case of a real ground, the consequence is not
contained (even mediately) as a predicate in the concept of the ground, and the
reason the ground ‘posits’ the consequence cannot be explained logically, that is,
through the principle of non-contradiction. In such a case, the presence of the
ground without the consequence does not entail a contradiction. As Kant says a
few pages later: “insight into the nexus of a real ground with its real consequence
cannot be had through the law of identity” (MH, Ak. 28: 24).20 This is not merely
the epistemic point that we lack the cognitive resources to analyze the concept of
the ground and find the consequence. Kant’s point is a metaphysical one: the fact
that X grounds Y is not a fact about concept containment or logical entailment. As
he puts it in an unpublished Reflexion: “logical grounds are not distinguished from
real grounds merely by the limits of my cognition [Erkenntnis], but in themselves”
(Refl. 3719, Ak. 17: 266).21
Kant draws this distinction but does not make explicit its critical relevance to
Baumgarten. Since Wolff ’s definition of ground is sufficiently similar to Baumgarten’s,22
and I am not aware of anything in the work of either thinker that would make a
relevant difference here, I am going to treat them equivalently. Leibniz’s views on
grounding and causation are sufficiently complicated that I will leave them aside
for now. It is not immediately clear from the text whether the distinction between
logical and real grounds is supposed to be a criticism of Wolff and Baumgarten. It
is not clear whether, for instance, Kant is claiming that Baumgarten (and Wolff)
held that all grounds are logical grounds and not real grounds, or whether he is
pointing out that Baumgarten (and by extension Wolff) overlooked this distinction
entirely. It is clear enough, though, that they do hold that essential grounds are

19
Cf. MH, Ak. 28: 12, 13, 18, and 25. Kant repeatedly return to this distinction in his metaphysics lectures:
MV (Ak. 28: 402–4), MvS (Ak. 28: 486), ML2 (Ak. 28: 549), MD (Ak. 28: 625), and MK3 (Ak. 28: 807–10).
20 21 22
Cf. MV, Ak. 28: 403. Cf. Refl. 3756, Ak. 17: 284. See previous discussion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

logical grounds. The essence of a being contains various predicates (its essential
predicates) and the reason the essence ‘posits’ those predicates is logical: if the
being lacked those predicates, a contradiction would result. The more difficult
question, however, is whether they extend this view to (efficient) causal grounds: is
the nexus between a cause and its effect a logical relation, for Baumgarten or Wolff?
While it might seem obvious that this is the view of Wolff and Baumgarten (and
even Leibniz), the textual evidence is not clear-cut.23
However, the transcripts from Kant’s metaphysics lectures in the 1760s (the
‘Herder’ metaphysics) contain an interesting argument that Baumgarten (and by
extension Wolff) is committed to the view that all grounds as such are logical
grounds. The jumping-off point is the standard logicist distinction between ‘in se’
necessity and hypothetical necessity.24 In se necessity is what I have been referring to
as necessity simpliciter since Chapter 1: that, the negation of which entails a contra-
diction. Something is hypothetically necessary when it is necessary given the assump-
tion of something else (paradigmatically, its ground). If X is the ground of Y, and
Y is not necessary in se, Y is hypothetically necessary, namely, on the hypothesis
of X. Kant claims, though, that there is a connection between in se and hypothetical
necessity: “every hypothetical necessity can be transformed into an absolute [in se—
NS] necessity, if the hypothesis is added to the concept of the thing, e.g., all fallen
humans sin” (Ak. 28: 18).25 The idea of Kant’s example is that humans sin is not
necessary in se (its negation is not contradictory), but only necessary on the hypoth-
esis that humanity is fallen; it becomes in se necessary if we add that hypothesis to the
concept of the subject: all fallen humans sin. Translating this point back to the case of
grounding: if X is the ground of Y and Y is not necessary in se, it is necessary on the
hypothesis that X exists. This means, by Kant’s conversion rule, that the conditional
if X exists then Y exists is itself in se necessary (“every hypothetical necessity can be
transformed into an absolute necessity”). Recall the logicist theory of in se necessity:
it is in se necessary that p if and only if ¬p entails a contradiction. But this means that
the nexus between a ground and its consequence, and, more specifically, between an

23
Eric Watkins follows Kant in reading Wolff and Baumgarten as positing a logical relation between
ground and consequence. He points out that Wolff derives the PSR, the principle that every being has a
ground, from the PNC, a logical principle, and reasons that the PSR must be a logical principle as well
(Watkins 2005, 119). But even if the PSR follows from the PNC, this does not mean that the ground–
consequence relation is a logical one. Watkins also points to the ubiquitous use of ‘posit’ [setzen, poni], an
originally logical term, to describe the relation of ground to consequence (Watkins 2005, 119 and n. 17 on
that page); but Kant himself talks of the ‘absolute’ positing of an object for a concept, while his point is
precisely that the object is not ‘contained in’ the concept (Ak. 2: 74). Hogan (2005) moves from Kant’s very
early defense of real inter-substantial causation (in Living Forces and Nova Dilucidatio) to his upholding of
real causal relation in the 1760s without making explicit that ‘real’ means something different in the two
contexts: in the first context it means non-ideal (i.e. not merely an epistemic ground of knowing), and in
the second context it means non-logical. While I agree with Watkins and Hogan that Kant is reading
Baumgarten and Wolff (and perhaps Leibniz) correctly, I am not sure the textual case is so clear-cut.
24
See Meta. }}102–5, Dt.Met. }}574–5, and AzDM }}195, 197. See also Leibniz, Th. }235.
25
MV, Ak. 28: 426.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

(efficient) cause and its effect, is a logical relation and can be explained through the
principle of contradiction.
It is not completely clear that Wolff and Baumgarten would accept this argument;
in particular, it is not clear that they accept that if B is necessary on the hypothesis of
A, then if A exists then B exists is necessary in se. However, I am going to follow Kant
in assuming that they do, and thus that they are committed to the principle that all
grounds as such are what Kant calls logical grounds; I will attempt to reconstruct a
plausible theory of grounding on this basis. However, a problem immediately arises
for such a view: if A is the causal ground of B (taking A and B to be events, for the
sake of simplicity), the logicists cannot claim that the proposition A occurs but B does
not26 is impossible in se, or impossible ‘in its own nature’, for truths impossible or
necessary ‘in their own nature’ do not depend upon God’s will; God’s will makes
things actual, but not possible or impossible ‘in themselves.’ So even God could not
make it the case that the cause occurs without the effect occurring. This is not only
implausible, but theologically unacceptable.27 To take an example, if my striking a
match causes it to ignite, the logicists would be committed to claiming that even God
could not cause that match-striking to occur without the match igniting. The logicists
need the relation between a causal ground and its effect to be weaker than strict
logical entailment, without giving up on the more general idea that all grounding can
be assimilated to logical grounding (assuming Kant’s argument from the previous
paragraph is correct).
The logicists can do this relatively easily by distinguishing, in the case of causal
grounds, between the ground, the consequence, and the conditions. The basic idea is
that there are some additional conceptual conditions that, in combination with the
ground, are logically sufficient for the consequence.28 So, in the case of causal
grounding, Kant’s ‘conversion rule’ from hypothetical to necessity in se should be
reformulated: if Y is necessary on the hypothesis of its cause, X, then (assuming that
causes and effects are events) there is some conjunction S of propositions about the
background conditions such that if if X occurs and S is true then Y occurs is necessary
in se. In the case of causal grounds, the natural candidates to include in the conditions
are the facts about which natural laws obtain and background information about the
substances involved (perhaps even the PSR itself). The proposition that I strike the
match, plus propositions about the laws that govern the world (e.g., if I strike a
match, the match is dry, there is sufficient oxygen, it is not in a Faraday cage, etc. then
the match ignites), the background information (the match is dry, there is sufficient
oxygen, it is not in a Faraday cage, etc.), and perhaps even the principle of sufficient
reason itself, will suffice to entail the proposition that the match ignites. The logicists

26
I am assuming that the existence of an event is equivalent to its occurring.
27
Unless one individuates events very finely, i.e. builds the conditions into the essence of the event.
28
Cf. Longuenesse (1998), 95–7, which offers textual evidence that this was in fact Wolff ’s view; see
Log. }}505, 513, 516, and 523.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

thus need to distinguish, in the case of causal grounding, between the ground, which
explains why the effect occurs, and the conditions, with which the ground must be
supplemented in order to entail the effect. To make such a view work, they would
need to draw this distinction in a principled way, and there are significant philo-
sophical difficulties in doing so, but I am not going to go into those details here.29
If Leibniz and Baumgarten do hold the ‘logicist’ view of (specifically, causal)
grounding that Kant attributes to them, this would be a promising strategy for
them to resolve an obvious problem about the (in se) contingency of causal connec-
tions. What is more, there is some indirect evidence that this is their strategy. In
response to Kant’s claim in the CPR that in a synthetic judgment the predicate is not
contained in the subject (A7–10/B11–14)—a doctrine anticipated by the pre-Critical
distinction between logical and real grounds—the rearguard Wolffians Eberhard and
Maaß objected that, although the predicate may not be immediately contained in the
subject, it may be ‘mediately’ or ‘virtually’ contained when the subject is supple-
mented with some third concept or conceptually articulated ‘law of the understand-
ing.’30 Applying this to the cause–effect case, while the concept of the effect is not
directly contained in the concept of the cause, it may be contained in the concept of
the cause when that concept is supplemented with some additional conceptual
conditions, like the background information and laws, or perhaps the PSR itself.
While Maaß and Eberhard are somewhat less than crystal-clear on how this is
supposed to work, I think my distinction between the ground and the conditions
(articulated above) retains the core of their idea. It retains the original (at least
according to Kant) logicist idea that the relation between cause and effect is itself
explained by the PNC while allowing that the modal strength of the connection
between cause and effect is something less than strict logical (in se) necessity (because
without the condition the entailment does not go through).

3.3. Real Grounds

In Negative Magnitudes Kant presents a short but powerful argument that a cause
cannot be the logical ground of its effect:
As for this real ground and its relation to its consequence my question presents itself in this
simple form: how am I to understand the fact that, because something is, something else is
[weil Etwas ist, etwas anders sei]? A logical consequence is only really posited because it is
identical with the ground. Human beings can err: the ground of this fallibility is to be found
in the finitude of man’s nature, for if I analyze the concept of a finite mind [Geist], I see that
fallibility is to be found in it. In other words, I recognize that fallibility is identical with

29
They are parallel, in many respects, to the difficulties attendant to Nelson Goodman’s ‘cotenability’
theory of counterfactuals; see Goodman (1954), ch.1.
30
Maaß (1789), 223; see also Eberhard (1789). For critical discussion of the Eberhard–Kant dialectic on
the analytic/synthetic distinction, see the Introduction to Allison (1973), as well as Hogan (2013).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

what is contained in the concept of a mind [Geist]. The will of God alone contains the
real ground of the existence of the world. The will of God is something. The world which
exists is something completely different. Nonetheless, the one is posited by the other.
(NM, Ak. 2: 202)
Kant’s point is that, while human finitude is a logical ground of fallibility, God’s will
is not a logical ground of the existence of the world. The relation of God’s will to the
world is an instance of causation; Kant’s larger point is that causes in general are not
logical grounds of their effects. At first, this argument may appear to beg the question
against the logicist theory of grounds: “you may subject the concept of divine willing
to as much analysis as you please: you will never encounter in that concept an
existent world as something which is contained within the concept of God’s willing,
or as something posited by that concept through identity” (Ak. 2: 202). Kant’s
opponent could simply claim that God is the logical ground of the existence of the
world in the sense that logical analysis of the concept of God’s will (perhaps
supplemented with the principle of sufficient reason, or the principle that God
creates the most perfect possible world) reveals that God creates this world.
The key to seeing how this argument does not beg the question is the term
‘identity.’ Kant’s point earlier in the paragraph is that if concept A is the logical
ground of concept B, it follows that if an object falls under A, then that very object
falls under B. Logical grounds explain why an object that falls under one concept also
falls under another concept. As he puts it in the Herder metaphysics lectures: in the
case of logical grounds, “the consequence, which is identical to [the ground] as a
predicate, is posited through the law of identity” (Ak. 28: 11). The consequence is
(partly) identical [einerlei] to the ground in that the consequence (the predicate) is part
of the ground; the ‘law of identity’ means that if some object falls under concept A and
if B is part of (a mark of) A, then the object instantiates B. In the case he discusses,
humanity is a logical ground of fallibility because fallibility is partly identical to
humanity—fallibility is a constituent mark of the concept of humanity—and so if
any object instantiates humanity that very same object thereby instantiates fallibility.
This model of logical grounds appears to provide no way of understanding how the
positing of one object under a concept entails the existence of a distinct object under
another concept. “The will of God is something,” he writes; “The world which exists is
something completely different. Nonetheless, the one is posited by the other.”
It might be objected, though, that there are pairs of concepts such that, if one
object falls under the first concept, it follows that a distinct object falls under the
second concept. David instantiates the concept <father>; this entails that there is a
distinct object instantiating the concept <child of David>. So it appears, contra Kant,
that logical grounds can explain how the existence of one object falling under a concept
(<father>) entails the existence of another object falling under another concept (<child>).
But further reflection on the father/child example supports Kant’s conclusion.
David’s falling under the concept <father> entails that he has a child, but it does not
explain why this other person exists and is his child. Intuitively, the explanation works
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

the other way around: because this person exists and is David’s child, David is a father.
Nor does the fact that this other person exists and is a child of David’s explain why he
(David) exists. This leads to two conclusions. First, the father/child example does not
give us a model for thinking about how one object falling under a concept explains
the existence of a distinct object. Secondly, it gives us reasons to think that logical
entailment can never issue in such an explanation. Causal explanation is asymmetric
(at least in the case of creation), but logical entailment is not. David is a father if and
only if there is a child of David’s, but, intuitively, the relation of explanation works in
only one direction: David is a father in virtue of the fact that there is a child of David’s
(and this is because of an act of procreation between David and the child’s mother).
The analogue of the father/child strategy in the example of God’s creation of the
world would be to claim that it is built into the concept of God’s will that God creates
a world. Therefore, God’s will is a logical ground of the existence of the world. But
that will not work, for reasons analogous to the father/child case. If God creates
the world then it may well be (if we follow e.g., Leibniz) that logical analysis of the
complete concept of God’s will reveals the predicate <creates the world>. But this only
shows that the proposition that the complete concept of God’s will is instantiated
entails that the world exists; it does not show that the existence of God’s will explains
why the world exists.31
That Kant has anticipated this move is clear from the way his argument continues:
Nor am I willing to be fobbed off by the words ‘cause’ and ‘effect,’ ‘force’ and ‘action.’ For if
I already regard something as a cause of something else, or if I attach the concept of force to it,
then I am already thinking of the cause as containing the relation of the real ground to its
consequence, and then it is easy to understand that the consequence is posited in accordance
with the rule of identity. For example, the existence of the world can be understood with
complete distinctness in terms of the omnipotent will of God. But here ‘power’ signifies
something in God, in virtue of which other things are posited. But this word already designates
the relation of a real ground to its consequence, but it is this relation which I wish to have
explained. (NM, Ak. 2: 203).

The move Kant has anticipated by his opponent is to build the causal connection
between God and the world directly into the concept of God’s will, so that logical
analysis will reveal that one of the marks of that concept is <creates the world>. The
objection Kant here levels against this move is that by building the causal connection
into the content of the concept of the ground (God’s will), his opponent has
abandoned the original project of giving an account of what the causal connection
consists in. The original logicist idea was that A causes B in virtue of the logical
connection between A and B. If the logicist now says that <causes B> is built into the
concept of A, and that is why A is a ground of B, then he has failed to give any

31
The logicists might claim that only an infinite analysis would uncover the concept of the world within
the concept of God’s will, in order to defend the contingency of God’s creative act; see Ch. 1.2 for more on
the infinite analysis view of contingency.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

account of the content of that logical mark, and has thus failed to actually account for
the causal connection. Thus, Kant concludes, the causal connection between cause
and effect cannot be assimilated to a logical relation of entailment or of concept
containment.

3.4. Real Conflict

This principle—that causal grounds are real grounds of their effects—is the crucial
premise in Kant’s argument that logicists cannot admit the existence of ‘real conflict’;
without it, the argument does not go through. His preferred example of real conflict
is that of two equal but opposite forces acting on the same body:
The motive force of a body in one direction and an equal tendency of the same body in the
opposite direction do not contradict each other; as predicates, they are simultaneously possible in
one body. The consequence of such an opposition is rest, which is something (repraesentibile). It
is, nonetheless, a true opposition. For that which is posited by the one tendency, construed as
existing on its own, is canceled by the other tendency, and the two tendencies are true predicates
of one and the same thing, to which they belong simultaneously. The consequence of the
opposition is also nothing, but nothing in another sense to that in which it occurs in a
contradiction (nihil privativum, repraesentabile). We shall, in the future, call this nothing:
zero=0. Its meaning is the same as that of negation (negatio), lack, absence. (NM, Ak. 2: 171–2)32

A body, X, is acted upon by two forces, A and B, which are equal in magnitude, but
opposite in direction. The features of this example that make it a case of real conflict
are: (i) it is possible for one thing to be acted upon by both A and B; (ii) the two forces
are positive determinations in themselves, not mere absences of positive determin-
ation; (iii) force A, by itself, has an effect, motion a, and force B, by itself, has an
effect, motion b; and (iv) if one body is acted upon by both A and B, rather than the
combined effect a+b, the effect is rest, a lack of motion. The rest of body X is what
Kant calls a ‘deprivation,’ an absence of overall positive determination that is the
result of a real conflict of forces. Forces A and B cancel one another’s effects.33
On its own, though, the phenomenon of real conflict has little or no dialectical
force against the logicist. On the logicist picture, grounds are logically intelligible
grounds: the concept of force A contains the concept of motion a, and the concept of
force B contains the concept of motion b. This may make it seem that the logicist
cannot escape the conclusion that logically compatible realities cannot mutually
cancel one another’s effects; if motion a is a logical consequence of the concept of
force A, it might seem that nothing logically compatible with A (e.g., B) can cancel
this logical entailment.34 However, as we saw in the previous section, logicist theories

32
See Weg }160 for a Crusian anticipation of the Kantian concept of real conflict.
33
For the distinction between ‘deprivation’ (Beraubung, privatio) and ‘lack’ (Mangel, defectus, absentia)
see NM (Ak. 2: 177–8); cf. OPG, Ak. 2: 87.
34
In contemporary terms, logical entailment is said to be ‘monotonic.’
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

of causal grounds can distinguish between the cause and the conditions that make
possible the grounding relationship. The logicist can claim that the concept of the
force plus the conditions entails the effect. For instance, in the case where A is the sole
force acting on the body, this will be included in the conditions, and in the case
where the body is also acted on by B, this will be included.35 Thus, the concept of
force A can contain an infinite series of conditionals of the form: if a body is acted
upon by force A and force f, then the body will experience motion y. One of these
conditionals will be: if a body is acted upon by force A and by no other force, the
body will undergo motion a. Another will be: if a body is acted upon by force A and
by force B and no other forces, it will be at rest. Likewise for the concept of
force B and motion b. The ground will be a set of forces mentioned in the
antecedent of one and only one of these conditionals (e.g., A and B); the propos-
ition that the ground exists, together with the propositions that make up the
condition (e.g., that no other forces act on the body), will entail that the body is
not moved, in Kant’s example. By means of this strategy, the logicist can satisfy all
four criteria given above:
(i) A and B are both positive qualities, i.e. neither is a mere lack of positive
determination.
(ii) It is possible for something to be acted on by both A and B.
(iii) If a body were acted on by A alone, it would undergo motion a, and if a body
were subject to force B alone, it would undergo motion b.
(iv) If a body were subject to force A and force B, the result would be a lack of
motion.
By itself, then, Kant’s argument that logicism is incompatible with the existence of
real conflict is quite weak.
The key to understanding how Kant, steeped as he was in the Leibnizian tradition,
could estimate this argument so highly is the suppressed premise that causal grounds
are real non-logical grounds.36 Real conflict per se does not pose a special problem
for logicism; real conflict, by definition, involves (efficient) causation and, according
to the argument reconstructed in the earlier part of this section, the relation between
a cause and its effect cannot be modeled on logical relations of entailment or
conceptual containment. Logicism cannot account for real conflict because it cannot
account for real causal connections. Conflict itself does not pose any additional
problems for the logicist.

35
What I have in mind is that a causal judgment of the form G causes E is true just in case the
proposition G+C occurs entails E occurs where C is a condition specified by the context of utterance of the
original judgment.
36
He gives essentially the same argument in the Amphiboly section of the CPR (A264–5/B320–1) and
in the Progress essay (Ak. 20: 282–3).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

3.5. The Argument for Real Incompatibility

The discussion in this chapter so far has been at best obliquely related to modal
issues. Kant’s argument that causal grounds are real non-logical grounds spells
trouble for the logicist theory of causation specifically, and grounding more generally,
but does not by itself conflict with the logicist view that possibility and logical
possibility are co-extensive. However, I will show that this is the crucial premise
in Kant’s argument that there are metaphysically incompatible predicates that
are nonetheless logically compatible, which I have been calling really incompatible
predicates.37
It is unclear how Kant argues that there is real incompatibility between some pairs
of predicates. As we saw in the introduction to this chapter, he claims there is a real
incompatibility between thought and extension, but provides no direct argument for
this. However, I think a substantive Kantian argument for real incompatibility can be
teased out of his discussion of the rationalist idea of God as the ens realissimum. In
Beweisgrund he writes:
The data of all possibility must be found in the necessary being either as determinations of it, or
as consequences which are given through the necessary being as the ultimate real ground. It is
thus apparent that all reality is, in one way or another, embraced by the ultimate real ground.
[ . . . ] But this is not to be understood to mean that all possible reality is included among its
determinations. (OPG, Ak. 2: 85)

This is a reference to the argument, which we will analyze extensively in the next two
chapters, that God is the ground of real possibility. For now, this passage is important
for its negative claim: that God, the most real of all possible beings, is not the
rationalist ens realissimum that possesses all unlimited realities. Kant goes on to
explain why:
This is a conceptual confusion which has been unusually prevalent until now. All realities are
attributed indiscriminately as predicates to God, or the necessary being. That all these
predicates can by no means co-exist together as determinations in a single subject is not
noticed. The impenetrability of bodies, extension and such like, cannot be attributes of that
which has understanding and will. (OPG, Ak. 2: 85)

First of all, the ens realissimum is not possible because thought and extension, which
Kant is here assuming are realities, are not metaphysically compatible. So far, though,
he has given no argument that not all logical possibilities are metaphysically possible.
He goes on:
Nor does it help if one seeks to evade the issue by maintaining that the quality in question is
not regarded as true reality. The thrust of a body or the force of cohesion are, without doubt,
something truly positive. Similarly, in the sensations of the mind, pain is never merely a
deprivation. (OPG, Ak. 2: 85–6)

37
This is equivalent to what Chignell (2009a) calls ‘subject-canceling real repugnance.’
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

This is a reference to the logicist view that all realities, all positive determinations, are
logical complexes of limitations of the unlimited realities possessed by God. This
view requires the logicists to hold that predicates like <pain> are deprivations, that is,
being in pain is a matter of lacking some further reality possessed by God. In context,
Kant’s point is that there are pairs of really incompatible realities (e.g., thought and
extension) and the logicists are wrong in assuming that, in such cases, one of the pairs
is not a true reality (e.g., extension) but a complex of limitations of God’s unlimited
realities. But this does not constitute an argument against logicism.
Kant then introduces real conflict:

A confusion has seemingly justified such an idea. It is said: reality and reality never contradict
each other, for both of them are true affirmations; as a consequence, they do not conflict with
each other in the subject either. Now, although I concede that there is no logical contradiction
here, the real repugnancy is not thereby canceled. (OPG, Ak. 2: 86)

He goes on to define what real conflict is, and illustrates it through his preferred
example of a body acted on by two equal but opposite forces. These are familiar
points by now, but the interesting issue in this context is the implications of real
conflict for the idea of God as the ens realissimum:

Now, in the most real being of all there cannot be any real opposition or positive conflict
among its own determinations, for the consequence would be a deprivation or a lack, and that
would contradict its supreme reality. Since a conflict such as this would be bound to occur if all
realities existed in the most real being as determinations, it follows that they cannot all exist in
it as determinations. Consequently, since they are all given through it, they will either belong to
its determinations or to its consequences. (OPG, Ak. 2: 86)

Now we see the beginning of an argument from the phenomenon of real conflict to
the claim that not all logically compatible predicates are metaphysically compatible:
(1) For some realities A and B, there is real conflict between A and B.
This is something Kant takes himself to have already established; later we will see
reasons to question it.
(2) If there were an ens realissimum, a being possessed of every unlimited reality,
some of those realities would stand in real conflict. The effects they would
normally ground will therefore be canceled. The effects they would normally
ground are realities, so the cancellation of them involves a lack of reality.
Kant does not give specific examples of which realities would stand in real conflict,
but based on the theory of real conflict in Negative Magnitudes it is not hard to
determine which ones he has in mind. Although I focused on the real conflict of
physical forces, Kant’s other main example of real conflict is moral: the will to do evil
and the will to do good are distinct positive determinations that stand in real conflict.
If there were an ens realissimum it would possess both of these realities to the highest
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

possible degree and would thus be in a state of permanent equilibrium, unable to act.
This contradicts its nature as the most real possible being.38 It follows that:
(3) The concept <ens realissimum> is not possibly instantiated.
But recall:

(4) The co-instantiation of A and B is a real ground of their effect, a lack of

overall positive determination in the subject that instantiates them.
(5) The relation between a real ground and its effect is not a logical relation of
entailment or concept-inclusion.
This is a consequence of Kant’s theory of ‘real grounds’—the relation between a cause
and its effect cannot be modeled on the logical relations of inclusion (or entailment)
between concepts (or propositions). But notice that:

(6) ∴ If the concept <ens realissimum possessing both A and B> were logically
inconsistent then the real conflict between A and B and their effect, a lack of
overall positive determination, would be a logical relation.
If the concept <ens realissimum possessing both A and B> were logically inconsistent
then the logicist would have an explanation of the ‘canceling’ that occurs between A,
B and the nature of the ens realissimum; if one and the same object possesses A and B,
then as a matter of logic, this being is not the most real possible being (because <ens
realissimum possessing both A and B> is logically inconsistent), so it lacks some
positive determination. If this concept is logically inconsistent, then the relation
between the co-instantiation of A and B and their effect, not being the ens realissi-
mum, is a logical relation, contra (4). But since, by hypothesis, A and B are among the
unlimited realities:
(7) <ens realissimum> = <ens realissimum with A and B>.
(8) ∴ The concept <ens realissimum> is logically consistent but not possibly
instantiated.
(9) ∴ There are really incompatible predicates, those contained in the concept
<ens realissimum>.

38
Strictly speaking, this contradicts the nature of the ens realissimum (ER) only if we assume that (i) the
ER is the most real being in the sense of having the most overall reality of any possible being, and (ii) that
some possible being would have more overall reality in the envisaged situation in which two of the
unlimited realities of the ER stood in real conflict. Premise (i) is plausible enough, though it is an additional
assumption, because possessing every unlimited reality does not strictly entail possession of the highest
overall degree of reality (as Kant notes in Danz.RT, Ak. 28: 1251). Premise (ii) is the more substantive
assumption here; Kant’s reasoning, I take it, is that a being that had A and not B (or B and not A) and all of
the other unlimited realities—call it ER+—would have more overall reality than ER because, in the example
given in the main text, it would have the unlimited will to do good and that will would have its ordinary
effects. The ER, by contrast, would be in a perpetual state of equilibrium, unable to act. So, while I think
Kant’s argument is ultimately successful on this point, it does appear to rely on some premises he fails to
make explicit.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

The basic idea of this argument is that the fact the relation between the grounds
and their effects (cancellation of realities) cannot be understood logically—the
fact that this is real conflict—is used to show that there is a logically consistent
concept (that of a being possessing every possible reality) that is not possibly
instantiated.
However, this argument relies on Kant’s assumption that there are primitive
realities (the realities possessed by the ens realissimum) that stand in real conflict.
Kant’s own examples of such pairs of conflicting realities include: pleasure and
pain, the will to do good and the will to do evil, etc. But the logicist will simply deny
that one half of each of these pairs is a reality at all; the logicist will claim, for instance,
that the will to do evil is a mere privation of the will to do good, and pain is a mere
privation of pleasure. By relying on his own views about which properties are realities
rather than mere privations (e.g., pain), Kant begs the question against the logicist.
However, a structurally identical argument can be made using less controversial
examples of realities, like Kant’s favorite example of real conflict, the real conflict
between two equal but opposite forces acting upon the same body. While the logicist
need not admit that forces are logically primitive realities (see section 4), it is not
plausible for the logicist to claim that either of these forces is a mere privation of
reality. Kant’s argument, as I reconstruct it, requires only that they are realities, not
that they are primitive:

(1) There is real conflict between A and B.

(2) The concept <body acted upon by forces A and B and no other forces and
moved in direction a> is not possibly instantiated.

I am assuming for the sake of this argument that, in cases of real conflict, it is
necessary that if no other factors are present the deprivation (lack of positive
determination due to real conflict) results. This does not mean that God could not
intervene to move the body; in that case, there would be an additional force acting,
God’s will, so it would not falsify the conditional.

(3) The co-presence of A and B is a causal ground of the lack of motion in

direction a (and of lack of motion in direction b).
By definition, real conflict involves causation; premise (3) is just the application of
Kant’s general principle that causal grounds are real grounds.
(4) A causal ground is not a logical ground of its effects; a causal ground does not
logically entail its effect.

This is just the definition of real ground.

(5) If the concept <body acted upon by forces A and B and no other forces and
moved in direction a> is logically inconsistent, then the co-presence of A and B is a
logical ground of their effect, lack of motion in direction a.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY 

If this concept is inconsistent, then the proposition x is acted on by forces A and B and
no other forces logically entails the proposition x is not moved in direction a; by parity
of reasoning, it also entails x is not moved in direction b. If this concept were logically
inconsistent (due, perhaps, to the highly complex logical structure of the concepts of
A and B), then there would be a purely logical explanation of why the body does not
move in direction a, contra (3) and (4).
(6) ∴ This concept is logically consistent.
This follows from (3)–(5) by modus tollens.
(7) ∴ There is at least one concept that is logically consistent but not possibly
instantiated. [From (2) and (6).]
This directly contradicts the logicist view of possibility. While Kant never explicitly
makes this argument, it is a very natural extension of the argument in Beweisgrund
that the logicist ens realissimum, the being possessed of every fundamental reality, is
impossible. In fact, that argument can be naturally extended to any example of real
conflict to show that the concept of the being possessed of the conflicting realities and
one of the consequences that is ‘canceled’ by the real conflict, is a case of a logically
consistent but metaphysically impossibly instantiated concept.39

3.6. Real Possibility

In the previous chapter we saw that Kant argued for an important anti-logicist claim:
there is at least one necessary proposition, that God exists, which is not logically
necessary. However, Kant’s argument that there are really incompatible predicates
shows that there is a much wider class of counter-examples to the logicist equation of
possibility and logical consistency. For any pair of really incompatible predicates
P and Q, the proposition 9x(Px & Qx) will be logically consistent but not possibly
true; the existence of really incompatible predicates generates a wide class of counter-
examples to the logicist thesis. This means that possibility diverges quite widely from
logical possibility, contra the logicists.
A note about terminology. Although Kant does not use the Critical term ‘real
possibility’ in the pre-Critical works to distinguish possibility from logical possibility,
it is appropriate to use that Critical term.40 In the works of the 1760s Kant argues for
various non-logical real conditions on the possibility of a thing: a lack of real

39
Both this argument, and the argument concerning the ens realissimum, are spellings-out of an
argument also found in Abaci (2014), 9–10 (cf. the response by Chignell, who basically accepts the
argument: 2014, 58). However, Abaci informs me (private correspondence) that the argument derives
from my anonymous report on the original version of his paper. So the argument Abaci and Chignell
discuss is originally my argument, spelled out in full detail for the first time here.
40
As Abaci (2014) correctly notes. However, Kant does use the term ‘reale Möglichkeit’ in Refl. 4196
(Ak. 17: 452), which Erich Adickes dates to 1769–70.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL CONFLICT , REAL GROUNDS , REAL POSSIBILITY

incompatibility among logically compatible predicates and, as I will explain in

Chapter 4, an existing real ground of the possibility of its predicates. Consequently,
I will borrow Kant’s Critical term to describe the notion of possibility that, in the pre-
Critical works, he distinguishes from logical possibility in terms of its real elements
(real compatibility, real grounds).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

4
Grounding Possibility

4.1. Introduction
In the previous chapters we critically examined Kant’s reasons for rejecting several
central logicist doctrines about modality. So far, though, our account of Kant’s
modal metaphysics has been largely negative: God exists necessarily, but not
because his existence is grounded in his essence; not all logically compatible
predicates are really compatible; and hence not all logical possibilities are really
possible. In this chapter and the next I explore Kant’s positive theory of modality in
the pre-Critical period.
In this chapter I explore Kant’s idea of a real ground of possibility and his
conception of God as the unique real ground of all real possibility. In the next
chapter I critically reconstruct Kant’s argument that there is such a unique real
ground of all real possibility. This order of investigation—trying to understand
what the conclusion of the argument means before examining the argument
itself—might strike some readers as odd. However, it is justified by the unclarity
that surrounds Kant’s notion of a real ground of real possibility, a notion which is
never adequately explained in Beweisgrund itself. In fact, I will argue, not only are all
the extant interpretations of this notion in the secondary literature explicitly rejected
by Kant, he even asserts that we cannot understand how God grounds all real
possibilities. So I have chosen to devote an initial chapter to discussing a notion
that is both one of the most important, and most confusing, in Kant’s pre-Critical
modal metaphysics.
In section 2 I explore two conditions Kant imposes on real possibility: the formal
condition that the predicates of a possible thing must be logically consistent, and the
‘material’ or ‘real’ requirement that the predicates of a possible thing must themselves
be really possible. Predicates are really possible only if they have an existing ground.
Consequently, if nothing were to exist, nothing would be (really) possible. I point out
that Kant does not explicitly state what relation must obtain between a ground and
the possibility it grounds, although he does set down certain criteria, which I explain.
In this section I limit myself to what I take to be interpretively uncontroversial
aspects of Kant’s view in Beweisgrund, most importantly: real possibilities require
real grounds; God is the real ground of all real possibilities; and there is a distinction
between ‘fundamental’ predicates (instantiated by God himself) and ‘derivative’
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

predicates that are possible in virtue of being ‘consequences’ of God. In section 3

I explore some proposals in the secondary literature for how to understand the
Kantian notion of a ground of possibility and I argue that each of them is explicitly
rejected in Beweisgrund. In section 4 I point out that the mystery surrounding this
notion in Beweisgrund is no accident: Kant himself claims that God’s essence grounds
all real possibilities but that we cannot understand why God’s essence grounds real
possibilities. I interpret this to mean: we cannot understand what the relation
between God’s fundamental predicates and the derivative predicates is, and, hence,
we cannot understand why the obtaining of this relation makes those derivative
predicates really possible. But this is not the end of the story; in the next chapter I will
argue that Kant’s argument that there is a unique ground of all real possibility
implicitly relies on thinking of the grounding relation in causal terms and, conse-
quently, thinking of God as grounding real possibilities through his power. We are
bound to think of possibility as grounded in God’s power even though, according to
Kant, we know that this is not so; our cognitive constitution requires us to think of
God in a certain way, even though this does not accurately reflect his reality an sich.
The full resolution of this tension, I will argue in the final chapter, only comes in 1790
in the Critique of Judgment and Kant’s theory of the intuitive intellect. In the text of
the Beweisgrund Kant focuses mainly on the possibility of predicates. As I argued in
Chapter 3, predicates in the pre-Critical period are ‘amphibious’ entities: they have
both a worldly role (they include forces, and can be causally efficacious) and a
semantic/representational role (they have an internal logical structure, they can be
combined in judgments, and they have a representational content). In these next two
chapters I will be mainly focused on their ‘worldly’ role, but in section 5 I will also
consider their role as thought-contents. Beweisgrund consists of three Divisions
(Abteilungen), which are themselves divided into a series of ‘Reflections,’ which
consist in one or more numbered sections. For ease of reference I will refer to parts
of Beweisgrund by the Division number, followed by the Reflection number, followed
by the section number; e.g., II.3.ii will refer to the second Division, third Reflection,
second section.

4.2. Two Requirements on Possibility

In the first section of I.2, “On a necessary distinction in the concept of possibility,”
Kant writes:
Likewise, in every possibility that which is thought must be distinguished from the agreement
of what is thought with the principle of contradiction. A triangle with a right angle is in itself
possible. The triangle, as well as the right angle, are the data or the material [element] of this
possibility, while the agreement of the one with the other according to the principle of
contradiction is the formal [element] of possibility. I will also call the latter the logical
[element] of possibility, for the comparison of the predicates with their subject according to
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

the rule of truth is nothing other than a logical relation; the something, or what stands in this
agreement, will sometimes be called the real [element] of possibility. (OPG, Ak. 2: 77–8)1

A predicate (like <triangular>) has both a logical form (the logical relation holding
among its constituents) and a matter (the more basic predicates standing in that
logical relation). For a predicate to be possibly instantiated it must have a logically
non-contradictory form. This establishes its logical possibility. For real possibility, it
is required that any constituent predicate “is itself something and can be thought”
(Ak. 2: 77). This is what Kant calls the ‘real’ or ‘material’ requirement on possibility.
I take it to be the requirement that the predicates that compose a more complex
predicate (the ‘data’ or ‘matter’ of possibility) must themselves be possibly
instantiated.
It is important not to confuse this requirement with the principle, discussed in the
previous chapter, that some logically compatible predicates are really incompatible.
The material requirement on possibility is the requirement that the predicates that
compose complex predicates must themselves be really possible; it is not a principle
about the compatibility of these predicates. Kant is not rejecting the ‘real compati-
bility’ principle in the opening paragraphs of II.1; he is simply not committing
himself to it. He mentions Crusius’s view that there are non-logical relations of
real incompatibility between predicates (“ . . . Herr Crusius, who does not locate this
conflict [of predicates] merely in internal contradiction . . . ”) without either explicitly
accepting or rejecting it.2
Later he writes:
You know that a fiery body, a cunning man, and similar things, are possible. If I require
nothing more than inner possibility you will not think it necessary that a body or a fire, etc.
must exist as the data of their possibility, for they are at least thinkable, and that is enough. The
agreement of the predicate fiery with the subject body according to the principle of contradic-
tion lies in these concepts themselves, regardless of whether they are actual or merely possible
things. I also agree that neither a body nor fire may be actual things, and yet a fiery body is
nonetheless internally possible. But I venture to ask: in this case, is a body in itself possible?
Because you are not basing your answer on experience, you will list for me the data of its
possibility, namely extension, impenetrability, force, and who knows what more, and point out
that there is no contradiction among them. I concede all of this, but you must justify to me why

1
I follow the translation in Kant (1992a) by introducing the dummy noun ‘element’ to translate ‘das
Formale/Reale/Materiale/Logische’ (der Möglichkeit). I put it in brackets, though, to indicate that it does not
correspond to any word in Kant’s text.
2
Ak. 2: 77; cf. MH, Ak. 28: 10. This is a reference to Crusius’s doctrine of the ‘highest material principle
of thought,’ which states that logically consistent predicates that, given the laws of our understanding,
cannot be thought together are not possibly co-instantiated. For instance, <exists> and <does not exist at
any time or in any location> are logically compatible but not possibly co-instantiated. See Ent. }}15, 237,
287; De Usu }}27, 29; and Weg }}250, 261. For critical discussion of Crusius’s influence on Kant, see Wundt
(1924), 52–81, Heimsoeth (1956), 125–88, and Schönfeld (2000), 225; more recently, this influence has
been emphasized by Hogan (2009a) and (2005).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

you have the right to assume the concept of extension as a datum; if it meant nothing, the
possibility of bodies, which you have assumed, would be an illusion. (OPG, Ak. 2: 80)3

The possibility of (the instantiation of) some complex predicate like <fiery body> is
not settled by the logical compatibility of its constituents. If <fiery body> is possibly
instantiated then its constituents—the predicates of extension, impenetrability, and
force, among others—are themselves possibly instantiated. And what does the
possibility of these constituents amount to? He goes on:
Since you must eventually arrive at something whose possibility cannot be further analyzed,
suppose that it is the concept of extension that you cannot further analyze into simpler data in
order to show that there is nothing contradictory in it. The question then arises, whether
‘space’ or ‘extension’ are empty words or whether they signify [bezeichnen] something.4 The
lack of a contradiction does not settle it; an empty word never designates [bedeutet] something
contradictory. [ . . . ] As long as you prove possibility through the principle of contradiction,
you are depending on the thinkable that is given in things and you are considering only
connections among it according to logical rules. But when you finally consider how this is
given to you, you can base it on nothing other than an existing thing [ein Dasein]. (OPG, Ak. 2:
80–1)

This shows that the material requirement of possibility is not the requirement of real
compatibility among the constituents of a predicate, because the material require-
ment applies also for atomic predicates, i.e. ones that have no constituents. Kant is
not here concerned with compatibility among predicates,5 but with the possibility of
(the instantiation of) the atomic constituents of complex predicates themselves. Kant
is not committed to assuming that the predicates he here discusses, e.g., extension,
are actually logically atomic (“ . . . suppose that you can now no longer break up the
concept of extension into simpler data . . . ”);6 elsewhere, he expresses skepticism that
we could ever discover atomic predicates.7 He is treating them as logically atomic in
order to abstract away from the issue of logical compatibility (and real compatibility,
which he has not invoked in this passage) and isolate the issue of the possibility of the
atomic predicates themselves. Henceforth, unless stated otherwise, ‘predicates’ refer
exclusively to atomic predicates.
In the same paragraph Kant writes: “if space did not exist, or if space were not at
least given as a consequence through something existent, the word ‘space’ would

3
The italics on the predicates fiery and body are mine, not Kant’s; he does italicize data, though.
4
I have changed the structure of this sentence slightly to bring out Kant’s point.
5
See Ch. 3.
6
Chignell (2012), 657 is therefore wrong in claiming that Kant “suggests” that extension is unanalyz-
able; Chignell attributes this suggestion to Kant, only to then accuse him of “lurching . . . into apparent self-
contradiction” (658) a few pages later. As I argue below, though, the text Chignell cites, Ak. 2: 86, does not
even carry the suggestion that extension is unanalyzable.
7
Prize, Ak. 2: 280.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

signify nothing at all [i.e. be impossible].” He is referring to his account of what

makes predicates possible stated two pages earlier:
Either the possible is only thinkable insofar as it itself is actual, and then possibility is given in
what is actual as its determination, or it is possible because something else is actual, i.e. its
internal possibility is given as a consequence [Folge] through another existing thing [ein
anderes Dasein]. (Ak. 2: 79)

By “given in what is actual as its determination” I take Kant to mean: the predicate is
instantiated by an existing substance. So, for any possible predicate, there is a
substance that either instantiates that predicate or that predicate is possible in virtue
of being a consequence of that substance. Kant does not tell us what kind of
‘consequence’ it must be, though; since ‘consequence’ [Folge] is Kant’s generic term
for a thing insofar as it has a ground,8 I will take this to be equivalent to: the substance
is the ground of the possibility of the predicate. He does make clear that the
possibility of predicates for finite substances like <extension> does not depend
upon their being instantiated:
You know that a fiery body, a cunning man, and similar things, are possible. If I require
nothing more than inner possibility you will not think it necessary that a body or a fire, etc.
must exist as the data of their possibility, for they are at least thinkable, and that is enough.
(OPG, Ak. 2: 80)

If the possibility of <fiery> or <body> depended upon the existence of actual fiery
substances or bodies (substances instantiating these predicates), then if those sub-
stances and those bodies did not exist, those predicates would be impossible. Kant is
assuming that the possibility of predicates cannot depend upon the existence of finite
substances like these. So predicates of finite substances like these must be possible in
virtue of being ‘consequences’ of some existing substance.
To make things slightly more precise, we can formulate Kant’s material require-
ment on possibility as follows:
(Material) If F is a possible atomic predicate then there is a substance x such that
either (i) x instantiates F or (ii) the possibility of F is a consequence of x (x grounds
the possibility of F). In either case the possibility of F is ‘given’ in x.9
Kant expresses the material requirement as a requirement on possibility simpliciter;
as I observed in the previous chapter, Kant never uses the term ‘real possibility’ [reale
Möglichkeit] in Beweisgrund (or anywhere else in the published pre-Critical writ-
ings).10 The material requirement entails that if there is no existing substance that
grounds or instantiates an atomic predicate it is logically possible this predicate is

8
See MH (Ak. 28: 12), MV (Ak. 28: 355, 401), MvS (Ak. 28: 485), ML2 (Ak. 28: 549, 625), and MM (Ak.
28: 806 f., 820). Connection (nexus) is the generic term for the relation of ground and consequence.
9
OPG, Ak. 2: 79.
10
It does, though, occur in Refl. 4196 (Ak. 17: 452), which Adickes dates to 1769–70.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

instantiated (because it is atomic) but not possible simpliciter. This means there is at
least a conceptual distinction between possibility simpliciter and logical possibility,
and I have introduced the term (one that Kant will use quite frequently in later
writings) ‘real’ possibility to record the fact that possibility simpliciter has ‘real’ or
‘material’ as well as logical/formal conditions. Henceforth, unless explicitly stated
otherwise, ‘possibility’ in this and the next chapter means real possibility. Kant does
not explicitly state the requirement that an existing ground of possibility must be a
substance (rather than, e.g., a mode of a substance),11 but this is a plausible assump-
tion. If we follow Kant in defining a substance as something that (to use my
terminology) instantiates predicates, and which is not instantiated by anything else,
then it is plausible to assume that predicates do not ground possibilities ‘on their
own’; the substances that instantiate them ground those possibilities (perhaps by
instantiating the predicates in question).12
In II.3 Kant argues from the material requirement to the conclusion that there is a
unique substance in which all possible predicates are given, God. I will reconstruct
Kant’s complex argument for this claim in the next chapter; in the meantime, we
need to understand what the conclusion of this argument means.
Kant argues that all possibilities are given in God: this means that all possible
predicates are either instantiated by God or they are possible in virtue of being
consequences of God.13 Some care is needed in unpacking these notions. They invoke
a distinction between two kinds of predicates, which I will call the ‘fundamental’ and
the ‘derivative’ predicates, respectively. It is crucial to understand that God’s funda-
mental predicates are not made possible by being instantiated by God. That these
predicates are instantiated by God logically entails that they are possible (actuality
entails possibility), but, in the terminology of the previous chapter, neither their
instantiation by God, nor God himself, is an explanatory ground (what Kant calls an
antecedently determining ground) of their possibility. There is no explanatory ground
of God (see Chapter 1), and God’s instantiation of these predicates does not explain
why they are possible, though it does entail it. So when Kant says that God grounds all
possibilities, he should be interpreted to mean: all possible predicates are given in
God, but only some predicates have antecedently determining grounds of their
possibility. Since Kant uses the ‘grounding’ terminology much more often than he
uses the ‘givenness’ terminology, I cannot simply abandon it; sometimes I will use
‘grounding’ in the broader sense to mean ‘giving’ possibility (either by instantiating a
predicate or by being an antecedently determining ground of its possibility) and
sometimes in the narrower sense to refer specifically to being an antecedently
determining ground of possibility.

11
At Ak 2: 84 he argues that the ground of all possibility is not an aggregate of substances, but he does
not argue that it is a substance at all. He seems to think this is obvious.
12
Cf. MH (Ak. 28: 24), ML2 (Ak. 28:562), MD (Ak. 28: 638–9), MK2 (Ak. 28: 779), Pöl.RT (Ak. 28: 1041,
1104 f.), and Danz.RT (Ak. 28: 1308).
13
OPG, Ak. 2: 89; see also 86 (“it follows from this . . . ”), 89, and 125. Cf. MH (Ak. 28: 134).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

The fundamental predicates are instantiated by God. The derivative predicates are
made possible by being ‘consequences’ of God (in some unspecified sense of ‘conse-
quence’). I will take this to mean: they are ‘consequences’ of God in virtue of his
fundamental predicates and represent the ‘consequence’ relation as a relation
between fundamental predicates and derivative predicates. Note that derivative
predicates can in principle include predicates instantiated by God; what makes
these predicates derivative is that they are made possible by being ‘consequences’
of his fundamental predicates, rather than fundamental predicates whose possibility
has no antecedently determining ground.14
Furthermore, we need a generalized version of the relation of a substance to a
possible predicate it gives (grounds in the broad sense) in order to reconstruct Kant’s
argument. I propose that for now we just call this ‘derivation relation’ R and define
the ‘givenness’ of possibility as follows:
(Defn.) Possible predicate F is given in substance x (substance x grounds the
possibility of F in the broad sense) if and only if either (i) x instantiates F and there
is no further antecedently determining ground of the possibility of F (F is a
fundamental predicate of x) or (ii) F is possible in virtue of bearing relation R to
the fundamental predicates of x (x grounds the possibility of F in the narrower
sense).
A completely worked-out theory of the grounding of possibilities would tell us: (i)
what the fundamental predicates are, (ii) what the derivative predicates are, (iii) what
the derivation relation R is (allowing us to generate the complete space of derivative
predicates from the fundamental ones), and (iv) an explanation of why the obtaining
of the derivation relation between a fundamental predicate and a derivative predicate
makes that derivative predicate possible. Our investigation into Kant’s modal meta-
physics will be guided by the attempt to fill in these gaps. Kant argues from the
principle that for any possible predicate, there must be a ground of that predicate (a
substance in which it is given), to the conclusion that there is a unique substance that
grounds all possible predicates (in which they are given). In order to reconstruct this
argument, then, we need to differentiate between being a ground of some possibility
(henceforth, a GSRP) and being a ground of all possibility (henceforth, a GARP),
even though the ultimate conclusion of Kant’s argument will be that anything that
grounds some possibility grounds all possibility (every GSRP is a GARP).15

14
See OPG, Ak. 2: 125–6, where, as I read him, Kant claims that God’s intellect, will, and power are
derivative predicates of the fundamental predicates that make up his essence. Likewise, it is in principle
possible for something other than God to possess a fundamental predicate, but that predicate will not be
fundamental to that other substance, for that predicate is possible in virtue of being instantiated by God,
not in virtue of being instantiated by that substance.
15
(Defn.) A substance x is a ground of some real possibility (GSRP) if and only if there is a possible
predicate F that is given in x.
(Defn.) A substance x is a ground of all real possibility (GARP) if and only if, for any possible predicate F,
F is given in x.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

Possible Predicates

Logically Complex

~D1, D1&D2, D2∨D3, (D1&D2) ∨~D3, F1&F3, F1∨~F3, etc.

Logically Atomic

Derivative predicates: D1, D2, D3, . . . .

groun
d ed in
God
R
y
ed b
t antiat
s
Fundamental predicates: F1, F2, F3, . . . . in

Fig. 4.1. The space of all possible predicates

In the previous chapter we examined Kant’s pre-Critical distinction between

logical grounds (whose relation or ‘nexus’ with their consequence is logically intel-
ligible) and real grounds (whose relation to their consequence is not logically
intelligible). In Beweisgrund Kant claims not only that real possibilities have grounds,
but that they have real (i.e. non-logical) grounds.16 He does not explain, though,
precisely what kind of real grounds are required. As we will see in section 3, the mere
fact that possibilities have real grounds rather than logical grounds will pose prob-
lems for the ‘intellectualist’ model of the grounding of possibility defended by some
commentators. Figure 4.1 represents Kant’s theory of the space of possible predicates.
The outer box represents all possible predicates; I defer discussion of impossible
predicates until section 5. The upper half of the box contains the logically complex
predicates; the lower half, logically atomic ones. For now I will only assume that
atomic predicates have grounds, an assumption I will revisit in the next section. The
fundamental atomic predicates are those that are actually instantiated by God and
which lack antecedently determining grounds of their possibility. The derivative
atomic predicates are those made possible by God (some of which may be instanti-
ated by God). The dashed arrow represents what I have called the derivation relation
between fundamental predicates and derivative predicates in virtue of which deriva-
tive predicates are possible. Since God grounds derivative predicates in virtue of
possessing his fundamental predicates, we can think of the ‘grounding’ relation

16
OPG, Ak. 2: 79, 83, 85, and 87.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

between God and the derivative predicates as the ‘logical product’ of the derivation of
derivative predicates from fundamental predicates and the instantiation of those
fundamental predicates by God. However, we do not yet know exactly what relation
the derivation relation is; determining that will be one of our main tasks.

4.3. Harmony, Power, and Intellect

Perhaps the best place to start in understanding Kant’s model of how God grounds all
real possibilities is with Andrew Chignell’s influential ‘real harmony’ interpretation.17
‘Real harmony’ is Chignell’s term for what I have called ‘real compatibility’: predi-
cates are in ‘real harmony’ just in case they are not only logically compatible but
really possibly co-instantiated. Chignell reads the ‘material requirement’ on real
possibility in Beweisgrund as requiring not only that there must be existing grounds
of really possible predicates, but there must be existing grounds of the real harmony
among really compatible predicates.18 Since the co-instantiation of two predicates
F and G is just the instantiation of the logically complex conjunctive predicate F&G,
the real harmony of predicates is equivalent to the real possibility of conjunctive
predicates like F&G. So Chignell’s interpretation (see Fig. 4.2) of the ‘material
requirement’ on possibility as requiring grounds of real harmony between compat-
ible predicates is equivalent to extending the range of predicates which need real
grounds to arbitrary conjunctions of predicates.
It is not clear whether, on Chignell’s view, other logically complex predicates, like
disjunctive predicates or conditional predicates, also require real grounds, so I will
leave them out of consideration.19 Chignell then reasons that, just as the fundamental
predicates must be instantiated to be really possible, the real harmony of these
fundamental predicates can only be grounded in their co-instantiation: in other
words, all of the fundamental predicates must be co-instantiated by God (equiva-
lently, arbitrary conjunctions are instantiated by God).20 In the next chapter I will
criticize Chignell’s reconstruction of Kant’s argument that all real possibilities are
grounded in a single being, God, but for now I merely want to investigate his
interpretation of what this means.
As an interpretation of Kant, Chignell’s model suffers from a lack of textual support,
as argued extensively by Abaci (2014) and Yong (2014).21 Chignell’s harmony

17
Chignell (2009a) and (2012); for critical reactions see Stang (2012), 290–1, Abaci (2014), Yong
(2014), and Chignell’s response, Chignell (2014a). Chignell uses the term ‘exemplification’ for what I am
calling ‘instantiation.’
18
Chignell (2009a), 176.
19
Though his invocation of the “modal PSR” Chignell (2009a), 176 would seem to require this.
20
Chignell (2009a), 187; (2012), 648.
21
Yong (2014), 2–33; Abaci (2014), 5–11. As both Yong and Abaci point out, the paragraphs leading up
to Kant’s assertion of the material requirement on possibility contain no endorsement of a ‘real harmony’
requirement (Ak. 2: 77–9); Kant does mention Crusius’s view that there are logically but not really
compatible predicates, but does not explicitly endorse it (Ak. 2: 77).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

Possible Predicates

Logically Complex

Conjunctive predicates:
F1& F2 & F3 & . . . . D1&D2, D3&D2, D1&D4

gr
ou
nd
gro

ed
un

in
de d

(in
in/
ins

sta
Logically Atomic tan

nt
t ia

iat
ted

ed
by

by
Derivative predicates: D1, D2, D3, . . . .

?)
groun
ded in God
R
y
ed b
antiat
inst
Fundamental predicates: F1, F2, F3, . . . .

Fig. 4.2. Chignell’s space of all possible predicates

requirement—that predicates must not only be logically consistent but really compat-
ible, or lack ‘subject-canceling real repugnance’—is obliquely referred to at two points
in Beweisgrund: Kant’s claim that thought and extension are logically, but not really,
compatible, and (I argued in Chapter 3) his argument that the ens realissimum cannot
instantiate every logically atomic predicate.22 It is not mentioned in the passages in

22
Both occur at Ak. 2: 85–6. Chignell also detects a reference to ‘subject-canceling real repugnance’
(real incompatibility) at 2: 89 and 190; Abaci and Yong convincingly argue that these passages concern
‘predicate-canceling’ real repugnance (Yong 2014, 31; Abaci 2014, 8–9). Chignell responds in (2014a),
55–60 by pointing to two texts: 2: 89 and 2: 190. At 2: 89 Kant argues that the being that grounds the world
must have intelligence and will because beings in the world have those perfections; if it did not, it would be
less perfect than its effects, and this, Kant claims, is impossible. Chignell is right, then, that <produces a
universe of order, beauty, and perfection> and <is a blindly necessary ground of other things> are
incompatible; but he has not given us a reason to think these predicates are logically compatible, so he
has not shown that this is a case of real incompatibility (subject-canceling real repugnance). Chignell is on
even weaker ground with 2: 190, where Kant claims that, “for something positive which exists to be
cancelled, it is just as necessary that there should be a true real ground” and “in other words, only in so far
as an equal but opposed real ground is combined with the ground of a is it possible for a to be cancelled.”
Kant regards the ground of the cancellation of a as “ a” and he makes clear earlier in Negative Magnitudes
that “ a” refers not to the logical negation of a (~a) but to a reality that is equal in magnitude but ‘opposite
in direction’ to a (e.g., a force in the opposite direction, pain as opposed to pleasure, etc.). Chignell’s
attempt to assimilate the ‘conflict’ between a and – a to the metaphysical conflict between predicates like
being water and being XYZ (2009a, 173) founders on the fact that predicates like being water do not have
‘opposites’ (other than their logical contradictories, which is not relevant here); there is no relevant sense in
which being XYZ is the ‘opposite’ of being water (any more than being cyanide is), so they cannot be values
of a and a in Kant’s schematic description of conflict. As for the first quote (“for something positive
which exists to be cancelled, it is just as necessary that there should be a true real ground”), he needs to read
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

which Kant states and argues for the material requirement on possibility, and is only
evoked after Kant has argued that there is a unique ground of all possibility (Ak. 2: 84).
Furthermore, Chignell provides no textual support for the claim that the compatibility
among predicates (their harmony, in Chignell’s sense) requires a ground (or that this
ground is God, rather than the natures of the predicates in question).
Aside from its lack of textual support,23 there is a deep philosophical problem
with Chignell’s model as well. As Yong (2014) points out, a problem arises when
we apply the co-instantiation principle to the derivative predicates themselves:
what grounds their ‘real harmony’? For obvious reasons, logical consistency will
not do. What about the real harmony of the fundamental predicates of which they
are derivative?24 This leaves us without an answer to the question of why the
derivation relation between fundamental and derivative predicates cannot intro-
duce sources of real incompatibility not present in the fundamental predicates. If
D1 and D2 are derivative predicates of, respectively, fundamental predicates F1 and
F2, and if F1 and F2 are really compatible, why does it follow that D1 and D2 are
really compatible?25
To satisfy the ‘real harmony’ requirement via the co-instantiation principle, then,
Chignell is forced to admit that derivative predicates are really harmonious (con-
junctive derivative predicates are really possible) only if they are co-instantiated (if
the conjunctive predicate is instantiated).26 Since, as we have seen, the instantiation
of a derivative predicate by a finite substance does not ground the possibility of that
predicate, the substance whose co-instantiation of these really harmonious derivative
predicates grounds their real harmony must be God himself. So God must instantiate
not only all of the fundamental predicates but all of the co-harmonious pairs of
derivative predicates. But this is absurd for two reasons. First, it means that if, for
example, being liquid and being extended are really harmonious, God must instan-
tiate both, and thus be extended. Secondly, it leads to inconsistency: there are pairs of
harmonious derivative predicates that are not mutually harmonious. For example,

this not just as extending to the existent but to the possible as well: for something positive which is possible
to be canceled, it is just as necessary that there should be a true real ground. But this is not Kant’s view: the
cancellation of the possibility of some positive determination (e.g., extension) can result from the mere
absence of its ground (e.g., God). I agree with Chignell that subject-canceling real repugnance (real
incompatibility, in my terms) is a part of Kant’s pre-Critical modal theory (as I argued in Ch. 3);
I disagree that it plays any crucial role in Kant’s theistic argument in Beweisgrund.
23
Chignell also cites Kant’s remark in ND that “nothing can be conceived as possible unless whatever is
real in every possible concept exists and indeed exists necessarily” (Ak. 1: 395; Chignell 2012, 177), but in
ND Kant explicitly rejects real non-logical incompatibility: “the principle of contradiction . . . is in fact
nothing but the definition of the impossible” (Ak. 1: 391). If the principle of contradiction is the definition
of the impossible then a fortiori, if no contradiction arises from the co-instantiation of F and G, it is possible
that F and G be co-instantiated.
24
Or the real harmony of fundamental with derivative predicates.
25 26
Yong (2014), 33–4. As Chignell himself admits (2012), 665–7.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

being liquid and being extended are harmonious, and thinking and being attentive
are harmonious, so it would follow that God has the following predicates: liquid,
extended, thinking, and attentive. But extended and thinking, we already know, are
not really compatible, so there would be what Chignell calls a ‘subject-canceling’ real
conflict in God.27
Nor is Chignell correct in assimilating the distinction between fundamental and
derivative predicates to the distinction between logically atomic and logically com-
plex predicates.28 On Chignell’s reading, the derivative predicates are logical ‘con-
structions’ from the fundamental predicates, using the standard eighteenth-century
rationalist operations of negation, conjunction, and limitation. This entails that the
predicates of finite beings are negations, conjunctions, and limitations of the atomic
(hence unlimited and positive) predicates of God. This means that if there is a non-
logical relation of real incompatibility among any of the fundamental (Chignell:
logically atomic) predicates then there is a ‘subject-canceling’ real conflict in God,
and God is impossible.29
In fairness to Chignell, he notes all of these problems and takes them to be Kant’s
reasons, in the Critical period, for rejecting this modal argument as an objectively
valid proof of God’s existence. But Kant himself notes these problems, both in
Beweisgrund itself and in contemporary texts, and offers them as reasons to reject
the view on which God instantiates all logically atomic predicates for the very reasons
Chignell gives: there would be a real incompatibility among God’s predicates.30
In a passage I analyzed in some detail in Chapter 3, Kant explicitly rebukes the
logicists for a “conceptual confusion”:

27
Yong (2014), 34. Chignell (2012), 668 suggests that Kant might be able to avoid this conclusion by
accepting a Spinozistic view of the divine attributes: mutually incompatible predicates are instantiated by
distinct divine attributes. But the very same objection applies as before. If <liquid> and <water> are
possibly co-instantiated and <liquid> and <xyz> are possibly co-instantiated then the same attribute
instantiates <water> and <xyz>, which by hypothesis is impossible. Yong (2014), 35 points out that if
these attributes are instantiated by God, then incompatible predicates instantiated by distinct attributes
lead to incompatible predicates instantiated by God. However, if the relation between attributes and God is
not instantiation then the co-instantiation thesis, which, according to Chignell, drives the argument, is
being jettisoned.
28
Chignell (2009a), 166 n. 19, 182.
29
Chignell’s conflation of atomic (logically fundamental) predicates with (metaphysically) fundamental
predicates leads him to claim (2012, 657–8) that Kant’s claim at Ak. 2: 86 that God cannot possess the
predicate of extension because he has a will, contradicts his “suggestion” (six pages earlier) that extension is
unanalyzable. Kant never claims extension is unanalyzable; Kant’s consistent pre-Critical view is that to be
extended is to be composed of substances in interaction. And his claim that <extended> does not
characterize God only entails that <extended> is analyzable if we are assuming that God instantiates all
unanalyzable (atomic) predicates. But that is precisely what Kant is denying in this passage.
30
Some care is required here. In the passage I quote below Kant asserts a real incompatibility between
extension and thought, something the logicists deny. But his argument, as I reconstructed it in Ch. 3, does
not beg this question: he argues there would be a real conflict (predicate-canceling real conflict) among
God’s predicates, and, because of this, a real incompatibility (subject-canceling real conflict). However, as
I pointed out, even the claim that there is real conflict among divine predicates is potentially question-
begging.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

But this should not be understood to mean that all possible realities belong among its [the
necessary being’s] determinations. This is a confusion of concepts that until now has been
unusually prevalent. One attributes all realities without distinction to God as predicates,
without realizing that they can never occur in the same subject as determinations. The
extension, impenetrability, etc. of bodies cannot be predicates of that which has understanding
and will. Nor can one evade this issue by claiming that the predicates in question are not
genuine realities. (OPG, Ak. 2: 85)31

Kant’s point in this passage is that God cannot possess every reality (every ‘positive’
predicate) because some of them are metaphysically incompatible (e.g., extension
and understanding). But, as Kant well knew, no one in the rationalist tradition
claimed that God possesses every reality period, much less that every reality period
is compatible; Leibniz, Wolff, and Baumgarten claim that God possesses every
logically fundamental reality, every unlimited reality, of which the realities of finite
beings are complex conjunctions, negations, and limitations. So Kant’s point here
would be dialectically irrelevant unless he means: God does not possess every logically
fundamental (atomic) reality. Hence his claim, “nor does it help if one seeks to evade
the issue by maintaining that the quality in question is not regarded as true reality.”
He is envisaging a logicist response: your putative examples of incompatible realities,
Herr Kant, are beside the point because they are not logically fundamental (atomic)
realities; they are complexes of negations and limitations of atomic realities. Kant’s
point is that there would be a real incompatibility among God’s predicates if he
possessed every logically fundamental (atomic) reality. If his point is merely that
there is a conflict among logically derivative realities, then he is diagnosing a
“conceptual confusion” that afflicted precisely no one in the German rationalist
tradition (as he would have known).
Kant then gives a series of examples of positive realities that stand in real conflict
and concludes:
Since, if all realities were present in [the necessary being] as determinations, a conflict of this
sort would arise, they cannot all be in it as predicates. Consequently, because they are all
possible through [the necessary being], they belong either among its determinations or among
its consequences. (Ak. 2: 86)

Chignell is correct that Kant rejected this view of God as instantiating every logically
atomic reality, and he is correct about why: such a God would have a real conflict
among his predicates. But he is wrong about when: Kant had already rejected the view
of God Chignell attributes to him in 1763, and so the developmental story Chignell
tells—Kant realizes that we cannot prove a lack of incompatible predicates in God

31
In the Herder metaphysics lectures from the mid-1760s, the period of the composition of Beweisgrund,
Kant explicitly and repeatedly denies that God instantiates all realities: Ak. 28: 128, 132–3, and 150 (see also
MK2, Ak. 28: 781–2, and 28: 917).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

and thus gives up on the modal theistic argument of Beweisgrund—cannot be

upheld.32
Chignell’s basic interpretive misstep was to ignore the fact that God is a real
ground of possibility and Kant’s pre-Critical view of real grounds: grounds whose
relation to their consequence cannot be understood logically. If the derivative
predicates are conjunctions, negations, and limitations of God’s fundamental predi-
cates then we can understand logically how the former make the latter possible.33
Kant’s primary example of real non-logical grounds, as we saw in the previous
chapter, are causal grounds. This suggests a possible interpretation of the distinction
between fundamental and derivative predicates: derivative predicates are causally
derivative of God’s fundamental predicates and, in virtue of this causal relation, they
are really possible.
On this interpretation (which I will ultimately reject), the fundamental predicates
are God’s fundamental causal predicates; the derivative predicates are the predicates
he can cause to be instantiated, in virtue of possessing these fundamental predicates.
This definition actually fits the text of Beweisgrund quite naturally; Kant argues that
intellect and will must be among the fundamental predicates rather than the deriva-
tive predicates (i.e. each must be possessed by God rather than merely a ‘conse-
quence’ of God), writing:
Leaving aside all the causes which are responsible for the generation of plants and trees, everyone
knows that regular flowerbeds, avenues, and such like, are only possible as a result of an
understanding which conceives the plan and a will which executes it. In the absence of
understanding, no power [Macht] or generative force [Hervorbringungskraft], nor any other
data of possibility, are adequate to render the possibility of such order complete. (OPG, Ak. 2: 88)

On one very natural reading of this passage, Kant is making a familiar theistic claim:
nothing that lacked an intellect and will could create a world with organized living
beings. This lends credence to the causal interpretation of the fundamental/derivative
distinction.
On this reading, God’s fundamental predicates ground the possibility of the
derivative predicates they causally bring about (cause to be instantiated). Recall,
though, that Kant denies that the non-instantiation of a derivative predicate renders
it impossible; if a derivative predicate like <extended> were never instantiated, it
would remain possible. So the derivative predicates grounded in God’s causally
fundamental predicates are not the predicates God actually causes to be instantiated;
if they were, any non-instantiated derivative predicate would be impossible. The
derivative predicates grounded by God’s causally fundamental predicates are the
predicates God could cause to be instantiated. The really possible derivative predicates
are the ones God has the power to cause to be instantiated.

32 33
Chignell (2009a), 188–92. OPG, Ak. 2: 79, 83, 85, and 87.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

There are passages in Beweisgrund that strongly suggest this ‘powers’ interpret-
ation of the distinction between fundamental and derivative predicates, for instance:
“not only the manner of their combination, but the things themselves are only
possible through this being, that is, they can exist only as its effects [Wirkungen]”
(OPG, Ak. 2: 125). Furthermore, Crusius, who was an important influence on Kant in
this period, distinguishes logical from real possibility, and grounds real possibility in
God’s power to produce a given state of affairs: “the ideal [logical] possibility of a
non-existent thing would mean nothing, if it did not at least contain in itself enough
reality that God, at least, would be a sufficient cause of everything in the thing [ . . . ] if
he were to make use of his omnipotence [Allmacht].”34 The Crusian ‘powers’ view of
possibility thus bears a certain similarity to Kant’s view: we can distinguish between
the fundamental predicates (God’s fundamental powers) and the derivative predi-
cates (the powers and predicates that are made possible by God’s powers) they make
possible.
It is crucial to distinguish this view from the voluntarist view on which God’s
actual choices ground what is possible and what is not.35 That voluntarist view,
famously endorsed by Descartes, grounds possibility in God’s will; the ‘powers’ view
grounds possibility in God’s power. The distinction can be thought of in the following
way: God has the power to actualize various worlds, but he wills only to actualize one
of them, this world. God’s will chooses among the space of possibilities, it does not
generate or ground that space. On the powers interpretation, the space of real
possibilities is constituted by what God has the power to choose, not by what he
does choose. Commentators who have argued against the ‘powers’ view by assimi-
lating it to the voluntarist view have conflated an important distinction.36
Kant clearly rejects the Cartesian voluntarist view: “the will makes nothing pos-
sible, but rather chooses things whose possibility is already presupposed” (Ak. 2:
100).37 What is more, in a passage I will analyze in some detail in the next section, he
also rejects the powers view:
But granted that the ground, which underlies not only the essence of all things but also the
essence of wisdom [Weisheit], goodness [Gute], and power [Macht], is a unity, it follows that
all possibility must of necessity harmonize with these predicates. (OPG, Ak. 2: 125)

This is an exceptionally rich passage, and fully analyzing it will take some significant
work in the next section, but for now I just want to point out that Kant is here

34
Ent. }56.
35
Descartes’ voluntarist views about modality can be found in his correspondence (CSMK III: 22–6,
235, 358–9) and the Sixth Objections and Replies (CSMK II: 281, 291–2); for critical discussion, see Curley
(1984), Van Cleve (1994), Kaufman (2002), and Easton (2009).
36
E.g., Chignell (2009a), n. 41. In Stang (2010), 296–7 I explicitly distinguish them. Chignell (2012), 671
n. 52 describes that paper as attributing to Kant a “quasi-voluntarist” view of modality. I no longer hold the
‘powers’ interpretation, but I do think it should be strictly distinguished from the voluntarist view.
37
Cf. the rest of that paragraph and the next, as well as Ak. 2: 91 and 28: 134.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

claiming that the unitary ground of the essence (real possibility) of all things, which is
in God (I take this to refer to the totality of God’s fundamental predicates), also
grounds his own omniscience (Weisheit), his omni-benevolence (Gute), and his
omnipotence (Macht).38 A fortiori, none of them is the ground of possibility; none
of them is what makes everything else possible. In particular, God’s unlimited power
is not the ground of all real possibilities. The reason the ‘powers’ interpretation is
wrong is not, as some commentators have claimed, that it is equivalent to voluntar-
ism (which it clearly is not), or that it is incoherent (which is more debatable); it is
simply not Kant’s view.39
Finally, some readers of Beweisgrund have attributed to Kant an ‘intellectualist’
model of the grounding of possibility, on which God grounds all real possibilities by
thinking of them, and all of the relations among them, as a totality.40 However, the
intellectualist reading is also incorrect, for two reasons. First of all, the intellectualist
model falls prey to the same textual problem that the powers model and the
voluntarist model do: Kant explicitly rejects it. Consider this passage, which imme-
diately precedes the passage quoted above:
If one were to ask ‘how do these natures depend on this being, and in such a way that I can
understand their agreement with the rules of wisdom?’ I would answer ‘they depend on
something in this being, which, in virtue of containing the ground of the possibility of things,
is also the ground of the being’s own wisdom; for its wisdom presupposes the possibility of
things.’ (Ak. 2: 125–6)41

I will analyze both passages in detail in the next section, but for now a few remarks
will suffice. In this passage Kant claims that real possibilities depend upon something
in God upon which his intellect also depends; it follows that his intellect is not the
ground of all real possibility.42 Kant reiterates this point in a footnote:

38
One might think that Kant is claiming that God’s unitary essence makes the wisdom, goodness, and
power of other things possible, but in the previous sentence he makes clear he is talking about God’s own
wisdom: “auch der Grund seiner eigenen Weisheit” (2: 125 f.). By parity of reasoning, I think Kant is also
talking about God’s own goodness and power.
39
I defended the ‘powers’ interpretation in Stang (2010), which has received sharp criticism from
Newlands (2014), 175 n. 52, and Chignell (2009a) and (2012). Note, though, that neither Chignell nor
Newlands present what I take to be the ‘proof texts’ against the ‘powers’ interpretation: Ak. 2: 125–6 and
152–3 (see section 4). My present attitude towards that interpretation is complex: I think it is not
adequate to Kant’s text (as I argue in this chapter), but I will argue in the next chapter that Kant’s
argument is best reconstructed assuming the powers interpretation. On Kant’s view, we are simultan-
eously forbidden and required to think of God as causally grounding all possibilities. This is what I refer
to as the tension in Kant’s pre-Critical modal thought. I discuss it further in the next chapter.
40
E.g., Watkins and Fisher (1998), Insole (2011), and Yong (2014). There are passages in Leibniz,
Wolff, and Baumgarten that support such an intellectualist view, on which God grounds all possibilities by
representing them in his intellect; see Mon. }}43–44, Th. }184, Dt.Met. }974, and Meta. }864. See Newlands
(2013) for critical discussion.
41
Thus, Chignell is wrong to claim that Kant is “more or less silent about why” the intellectualist model
does not work (Chignell (2009a), 182).
42
Cf. Ak. 1: 119: “unity is thus derived from a wise being, but not through his wisdom.”
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

Wisdom presupposes that agreement and unity in the relations [among possible things] are
themselves possible. A being which is of a completely independent nature can only be wise in
virtue of containing in itself the ground of the harmony and perfection it itself has the option of
realizing. If there were no such relation to order and perfection to be found in the possibility of
things, wisdom would be a chimera. If possibility were not grounded in the wise being itself, its
wisdom would not be independent in every respect. (OPG, Ak. 2: 126 n.)

There is a lot going on in this footnote, but Kant’s primary point is that possibility
must be grounded in God himself, otherwise his omniscient intellectual awareness of
all possibilities would depend upon something external to him, the hypothesized
external ground of possibility; this would make God’s intellect dependent upon an
external source, which Kant assumes is impossible. But the first sentence of this
note—“Wisdom presupposes that agreement and unity in the relations are them-
selves possible”—also entails that God’s intellect, his ‘Weisheit’, is not the ground of
possibility itself.43 Instead, Kant here claims, the agreement and unity among poss-
ibles must be independently grounded in God, and God’s intellectual apprehension
of that agreement and unity among possibilities is thus a form of self-cognition.
God’s intellect is not itself the ground of possibility; the unity in his essence is the
ground of all possibilities, and this itself makes his intellect possible; it is because
God’s essence grounds all possibilities that God is capable of having omniscient
intellectual awareness of all possibilities, through self-cognition of his essence.
But the problems with the intellectualist model do not end there; very deep
commitments of Kant’s pre-Critical metaphysics make the intellectualist model
unavailable to him. While it is clear that, according to Kant in the Beweisgrund,
God is a not merely an epistemic ground of real possibility (something through
which real possibilities can be known), it is less commonly appreciated that, given
Kant’s pre-Critical commitments in rational cosmology, this precludes the intellec-
tualist model of how God grounds possibility.
A major controversy of eighteenth-century German metaphysics concerns the
relations substances must bear to one another to be members of a common world
(or ‘world-mates’ in contemporary lingo), a unified totality of finite substances.44
Leibniz (and Leibnizians like Baumgarten) held that pre-established harmony among
the representational states of substances, which they conceived of as mind-like
simples (monads), makes substances into world-mates: the world is a totality of
substances whose representational states harmonize with one another.45 From his
earliest writings on, Kant defends the anti-Leibnizian position that world-mates must
really interact with one another; substances x and y are world-mates if and only if the

43
Noam Hoffer has suggested to me in conversation that ‘Weisheit’ here does not refer to God’s intellect
but to his intellectual representation of the harmony and ‘purposive unity’ (as Kant would later call it in CJ)
among possibilities; I do not have space to respond to this intriguing suggestion here, though. Cf. Hoffer
(forthcoming).
44
See Watkins (2005), 21–100 for a detailed discussion of this issue.
45
AG 33, 42, 46–7, 143–4, Meta. }}448–65; for discussion, see Watkins (2005), 24–8, 74–8.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

forces in x are causal grounds of accidents in y and vice versa.46 Kant rejects the pre-
established harmony theory (PEH) because he believes that a world is a real unity, i.e.
the parts of a world are antecedently determining grounds (not merely epistemic
grounds) of one another’s alterations. The representational or intellectual harmony
that obtains among the states of substances (Leibnizian monads) on the PEH theory
by itself does not make them antecedently determining grounds of one another’s
alterations. The complete representational state of such a substance provides an
appropriately informed agent (e.g., God) with complete epistemic grounds for pre-
dicting the subsequent states of all other substances in the same world; however,
because they do not interact, the states of these substances are not antecedently
determining grounds of one another.47
By parity of reasoning, though, Kant is committed to rejecting the intellectualist
model of how God grounds possible predicates because it would entail that God is
not an antecedently determining (explanatory) ground of their possibility. If God’s
intellect grounds real possibilities solely by representing them, then God’s mind could
be an epistemic ground of beliefs about real possibilities—per impossibile, were we
able to ‘see into’ God’s mind we would know what is and is not really possible—but
his mind does not explain why the represented possibilities are really possible. If
God’s intellectual representation of possibilities is an antecedently determining
ground of their possibility, then, by parity of reasoning, Kant would have to admit
that the representational unity among Leibnizian monads could make them into
antecedently determining grounds of one another’s states, and this would undermine
one of his core objections to pre-established harmony. Kant does not merely happen
not to endorse the intellectualist view of the ground of real possibility; one of his
deepest commitments in metaphysics (rejection of PEH as an explanation of the
world-mate relation) is incompatible with it.48

4.4. “This thought rises far higher than

a created being can reach”
Having argued against all extant interpretations of the notion of a real ground of real
possibility, including my own previous ‘powers’ interpretation, I want to pause and
reassess my conclusions up to this point. Kant builds an entire metaphysics in

46
MH, Ak. 28: 39, 45, and 51–2.
47
Although Kant thinks that world-mates are real grounds of one another’s accidents, his argument
against the PEH theory of the unity of a world does not depend on this assumption.
48
Some ‘intellectualist’ readers might object that Kant grounds the possibility of interacting substances
in God’s intellect in ND (Ak. 1: 413, 414). But Kant there says that the divine intellect is the source of the
existence of substances (“the self-same schema of the divine understanding, which gives existence”) and of
their dynamic relations, so he is not, at least there, claiming that the divine understanding is what makes
substances really possible.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

Beweisgrund on the twin principles that (i) God is the real ground of all possible
predicates, and (ii) possible predicates are either (a) instantiated by God (the
fundamental predicates), in which case their possibility has no antecedently deter-
mining ground, or (b) consequences of God (the derivative predicates), in which case
they are possible because of the grounding relation they bear to God (and his
fundamental predicates). But Kant never tells us what exact relation obtains between
the derivative predicates and the fundamental predicates (or God himself) and why
the obtaining of this relation makes the derivative predicates really possible. This has
spawned an entire secondary literature in which commentators have tried to answer
this crucial interpretive question.
What none of Beweisgrund’s commentators, myself included, have noticed is that
it is no accident that Kant fails to identify what this feature of God is; he explicitly
claims that we cannot say anything more than that it is essential to God to do so. In a
convoluted passage that I began to analyze in the previous section, he rejects several
of the models that have been attributed to him. He begins by raising a question: “how
do these natures depend on this being, and in such a way that I can understand their
agreement with the rules of wisdom?” (Ak. 2: 125). The context is Kant’s argument
that the unity and harmony that we observe in the essences of things (their real
possibility) points to there being a single ground of all possibility. Kant now asks, on
behalf of his reader, what relation is there between these essences and that unique
ground in virtue of which those essences are possible (essences of possible things)?
We might reasonably expect, then, that Kant will tell us (in my terminology) what the
derivation relation is and why it makes derivative predicates (and the essence they
compose) possible. But he does not do so:
I would answer, ‘they depend on something in this being, which, in virtue of containing the
ground of the possibility of things, is also the ground of the being’s own wisdom; for its wisdom
presupposes the possibility of things.’ (OPG, Ak. 2: 125)

I interpret this to mean that God has an essence (a set of fundamental predicates)
that grounds all other possibilities (all other possible predicates) and that this
essence is itself the ground of the possibility of his wisdom (his intellect). So God’s
intellect cannot itself be the ground of possibility. He goes on: “but granted that
the ground, which underlies not only the essence of all things but also the essence
of wisdom, goodness, and power, is a unity, it follows that all possibility must of
necessity harmonize with these predicates” (OPG, Ak. 2: 126). So the divine
essence, the ground of the possibility of all derivative predicates, is also the
ground of God’s goodness (his omni-benevolent will) and his power. Since
these predicates of God are made possible by his essence, they are not themselves
the ground of possibility. Kant is here explicitly repudiating the voluntarist, the
intellectualist, and the ‘powers’ interpretation of how God grounds possibility;
these divine attributes are themselves grounded in some more fundamental
divine essence.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

Not only does Kant explicitly repudiate all of the models for how God grounds
possibility that have been attributed to him in the secondary literature, he elsewhere
claims that we cannot in principle understand how God grounds possibility:
Nor is my wonder at the unity and harmony in things diminished once I have convinced
myself that this is only possible because there is a being which is not only the ground of
actuality but also of possibility. Though one can form, through an analogy with human action,
some concept of how a being can be the cause of everything actual, one can form no concept of
how such a being could contain the ground of the inner possibility of other things. It appears
that this thought rises far higher than a created being can reach. (OPG, Ak. 2: 152–3)

It is important to be clear on what question Kant is here denying we can ever answer.
God grounds possible predicates in two ways: (i) either as determinations (funda-
mental predicates), or (ii) as consequences (derivative predicates). Fundamental
predicates have no ‘antecedently determining’ ground of their possibility; properly
speaking, there is no ground why they are possible, they just are instantiated.
Derivative predicates are possible because of the relation they stand in to God’s
fundamental predicates. I take it that Kant is not merely denying that we can
understand why the obtaining of that derivation relation makes those derivative
predicates possible. He is also denying that we can ever understand what that
derivation relation is, for we can “form no concept of how that being can contain
the ground of the internal possibility of other things.” The way in which possibilities
are grounded in God is literally incomprehensible to us.
This is a dissatisfying result; it is supposed to be. However, it is the beginning of a
story that will end, I hope, in a less dissatisfying result. In the next chapter I reconstruct
Kant’s argument that there is a unique ground of all real possibility. One of the
conclusions of my reconstruction will be that Kant’s argument succeeds, but only if
(i) we grant him certain premises about rational cosmology, and (ii) we implicitly think
of the grounding of real possibility in causal terms. At the end of the next chapter I will
argue that if we look back at the argument we have reconstructed we will see that we—
and Kant himself!—have been implicitly thinking of God as grounding possibility
through his powers, precisely what Kant denies is the case at Ak: 2: 153 (quoted above).
So in Beweisgrund Kant argues that there is a unique ground of real possibility, by
implicitly thinking of this ground as a causal ground of real possibility, but then denies
that we can understand its role in grounding possibilities in causal terms, or any other
terms for that matter. This tension in Kant’s thinking about the ground of possibility
will not be resolved, I will argue, until 1790, in the Critique of Judgment. Explaining
how Kant resolves it, though, will require the rest of this book.

4.5. Possibility, Thought, and Content

Before continuing, in the next chapter, to discuss Kant’s argument that there is a
unique ground of all possibility, I want to briefly comment on Kant’s apparent
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

GROUNDING POSSIBILITY 

equation of the possible with the ‘thinkable’ in Beweisgrund.49 Although Kant

frequently uses these expressions in such a way as to suggest that he thinks of
them as identical, he does not claim that everything possible is thinkable by our
minds, and, as we have seen, even God’s intellect does not ground possibility: rather,
it is because God’s essence determines a certain space of possibilities that God’s
intellect is intentionally directed at those possibilities. Even for God, thinkability
presupposes, and is grounded in, possibility.50
What is the relation between a predicate being thinkable and it being possible?
A thinkable predicate, I take it, is one that can figure in the content of a contentful
thought. It would be highly problematic for Kant to assume that all thinkable
predicates are possible, if, by the ‘possibility’ of a predicate we mean the possibility
of its instantiation,51 for this would entail that every thinkable predicate is possibly
instantiated. For the sake of clarity:
(1) Predicate F is thinkable ⊃ predicate F is possible.
(2) Predicate F is possible ⊃ possibly, predicate F is instantiated.
(3) ∴ Predicate F is thinkable ⊃ possibly, predicate F is instantiated.
The conclusion, (3), is implausible, for at least two reasons. First of all, I can have
prima facie contentful thoughts, e.g., there is a largest prime number, involving
predicates, in this case, <largest prime number>, that are not possibly instantiated.
Secondly, if thoughts involving impossibly instantiated predicates lack content, then
any two such thoughts have the same content, namely, nothing, and that is highly
implausible. For instance, there is a highest prime number and there is a counter-
example to Fermat’s Last Theorem are thoughts involving impossibly instantiated
predicates, but clearly they have distinct contents. After all, they can be disproven (it
can be proved that their content does not accurately represent their subject matter),
and they stand in different inferential relations with uncontroversially contentful
thoughts.
We might be tempted to reject (2), but we should be cautious: Kant must hold on
to some connection between the possibility of a predicate and its possible instanti-
ation. If the possibility of a predicate has nothing to do with its possible instantiation,
then in proving that God is the ground of the possibility of all predicates he would
not thereby have shown anything about their possible instantiation; it would be
compatible with the conclusion of Kant’s argument that no such predicates are
possibly instantiated, which is surely absurd. Fortunately, there is a far less drastic

49
See OPG, Ak. 2: 77, 78, 80, 82, 83, 85, 87, and 162.
50
I was prompted to think again about the relation between thinkability and possibility in Beweisgrund
by the fascinating discussion of these issues in the final chapter of Yong (forthcoming), which connects
them to the opening of Hegel’s Science of Logic.
51
This should not be read to mean: possibly, there is a predicate F. For, as I will shortly argue, there
are impossible predicates: predicates that combine atomic predicates in ways that are necessarily
uninstantiated.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 GROUNDING POSSIBILITY

way to avoid the implausible conclusion, for we can understand the relation between
what is thinkable and what is possible by amending (1) as follows:
(1*) F is a thinkable predicate ⊃ (i) F is atomic and possibly instantiated or (ii)
F is complex, the atomic predicates of which F is composed are possibly instan-
tiated, and it is logically possible that F is instantiated.
Note that (1*) does not include the requirement that the atomic constituents of F be
really compatible. This delivers what I take to be the correct result that we can have
contentful thoughts about impossible co-instantiations of (individually possibly
instantiated) atomic predicates. Assuming for the moment that <extended> and
<thinking> are atomic predicates, the predicate <extended & thinking> is a thinkable
predicate, a predicate of which we can have a contentful thought; after all, the early
modern materialist (e.g., Hobbes) who thinks that some bodies think may be wrong,
indeed necessarily wrong, but his thought is not devoid of content.
What is more, there is good textual ground for attributing this view to Kant; he
consistently describes God as providing the data for thought.52 If we take ‘data’ in its
original meaning as ‘that which is given,’53 we can plausibly read him as claiming that
God is the ground of the material of thought, the possible atomic predicates; the
thinkability of some complex predicate (the contentfulness of thoughts about it)
requires that its data be possible, and that they be combined in a logically consistent
way, but it does not require that they be really compatible. We can have contentful
thoughts about impossibilities as long as the basic contents of our thoughts (atomic
predicates) are really possibly instantiated. A contentful atomic predicate, on the
other hand, must be possibly instantiated.
To anticipate somewhat the discussion of Chapter 6, this also strengthens the
continuity between Beweisgrund and the CPR, for in both works Kant is interested
in the question: what is the ground of thinkable content? Important differences
remain, though. First, in the Critical philosophy there are two importantly different
ways in which thinkable content can be given to us (spontaneously, in concepts, or
receptively, in intuition), and in accounting for thinkable conceptual content (think-
able predicates, in Beweisgrund) we have not yet accounted for how objects (which
instantiate those concepts) are given to the mind. Second, in accounting for the
metaphysical basis of thinkable content—atomic predicates are possibly instantiated,
in virtue of standing in the appropriate relation to God—Kant, in Beweisgrund, has
not yet explained how those predicates come to be contents of concepts available to
discursive intellects like ours. In particular, he has failed to account for how our
concepts come to represent possible (atomic) predicates, and thus acquire content,
without depending upon experience (a priori).

52 53
OPG, Ak. 2: 77, 79, 80, 82, 84, 85, 88, and 100. Cf. OPG, Ak. 2: 81.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

5
Kant’s Modal Argument

5.1. Introduction
In the previous chapter I explored Kant’s notion of a real ground of real possibility
and the meaning of his claim that there is a unique real ground of all real possibilities:
God. In this chapter I reconstruct his argument for that claim. The conclusion of
Chapter 4 was largely negative: Kant’s own explicit view is that we cannot say what it
is about God in virtue of which he grounds all real possibilities. In particular, we
cannot identify the fundamental predicates of God, the derivative predicates (the
predicates made possible by God’s fundamental predicates), or the ‘derivation’
relation that makes the latter possible. My reconstruction of Kant’s argument will
attempt to respect these constraints: as much as possible, it will be based on premises
about the structure of the derivation relation itself that does not attempt to say what
that relation is or why its obtaining makes the derivative predicates possible. Ultim-
ately, though, I will argue that we must violate Kant’s own scruples if the argument is
to succeed; reconstructing a crucial section of the argument relies on implicitly
thinking of grounds of possibility as causal grounds.
In section 2 I explore the other central modal notion of Beweisgrund: absolutely
necessary existence. I argue that this should be interpreted ‘hyperintensionally’: not
all beings that exist necessarily exist with absolute necessity.1 In particular, it should
be interpreted in terms of a counterpossible relation of ‘cancellation’: if a being exists
with absolute necessity, then were it not to exist nothing would be (really) possible. In
section 3 I reconstruct several important steps in Kant’s modal argument for the
existence of a unique ground of real possibility. Given plausible principles, I argue,
Kant’s argument is successful up to a certain point: he can prove relatively unprob-
lematically that there is a unique ground of all real possibility, and that this being
exists with absolute necessity. However, this alone does not prove that this being is a
single substance, rather than a plurality of substances. While this may seem a small

1
The core idea of ‘hyperintensionality’ is that possible worlds are not ‘fine-grained’ enough for all
philosophical purposes; for instance, 2 and {2} exist in all possible worlds, but the latter exists in virtue of
the former. I will mainly be interested in hyperintensional relations between propositions; a relation
R between propositions is hyperintensional just in case for some p and q, p bears R to q but (i) p does not
bear R to q* for some q* that is necessarily equivalent to q, or (ii) p* does not bear R to q, for some p* that is
necessarily equivalent to p. For instance, grounding is a hyperintensional relation: if p grounds q it does not
follow that p* grounds q, or p grounds q* (where p* and q* are as above).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

point, it is potentially devastating, for many metaphysicians would accept that there
is a plurality of entities that collectively ground all real possibilities.2 The ‘plurality’
objection, as many commentators on Beweisgrund have observed, poses a major
threat to Kant’s argument, and his own stated solution is quite weak. In section 4
I survey the secondary literature for solutions to the plurality objection and argue
that none of them succeed. In section 5 I reconstruct a Kantian solution to the
plurality objection using a (non-Humean) modal recombination principle and some
core principles of Kant’s cosmology. In section 6 I observe that, on my reconstruction
of Kant’s argument, we must implicitly think of God as causally grounding all real
possibility, something, as we saw in Chapter 4, Kant explicitly denies. I argue that this
is a consequence of Kant’s claim that we have “no concept” of how God could ground
the real possibility of things: because we do not understand the real grounding
relation in which God stands to real possibility, we have no choice but to model it
on a species of real grounding that we understand better: causal grounding. In section
7 I summarize key points of Kant’s pre-Critical modal metaphysics that will be
crucial to the second half of this study.

5.2. Absolute Necessity

Kant begins Beweisgrund I.3 with a familiar theme, necessary existence:
The absolutely necessary is that whose non-existence is impossible in itself. This is undoubt-
edly a correct nominal definition. However, when I ask: why is it that the non-being of a thing
is absolutely impossible? what I am looking for is the real definition, which alone can serve our
purpose. (OPG, Ak. 2:81)

As I argued in Chapter 1, a real definition of ‘necessary existence’ tells us what it is in

virtue of which a necessarily existent being necessarily exists. One candidate answer
is the logicist one—it necessarily exists because its non-existence is logically
impossible—but Kant rejects it:
All of our concepts of the inner necessity in the predicates of possible things, of whatever kind
they may be, derive from the fact that [laufen darauf hinaus, daß] the opposite contradicts
itself. However, in the case of absolutely necessary existence, one would have little success in
understanding it through that characterization. Existence [Dasein] is not a predicate, nor is the
cancellation of existence the negation of any predicate, through which something in a thing is
cancelled and an internal contradiction might arise. (OPG, Ak. 2: 81)

Kant is referring us back to the First Reflection, where he argued that, because
existence is not a real predicate, nothing exists with logical necessity.

2
For instance, on the view of Fine (1994a), facts about necessity are grounded in the essences of things;
since there are multiple things whose essences ground the necessity facts, in Kantian terminology, the
ground of real possibility is a plurality of beings rather than a simple individual.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

Logically necessary existence is a contradictio in adjecto, so if anything exists

necessarily it exists with real necessity. Kant therefore needs a real definition of really
necessary existence, and he goes on to provide one in terms of the concept of absolute
necessity:
If I now consider for a moment why that which contradicts itself should be absolutely nothing
and impossible, I find that through the cancellation of the law of contradiction, the ultimate
logical ground of all that can be thought, all possibility vanishes, and there is nothing left to
think. The conclusion immediately follows that, when I cancel all existence whatever and the
ultimate real ground of all that can be thought therewith disappears, all possibility likewise
vanishes, and nothing any longer remains to be thought. Accordingly, something may be
absolutely necessary either when the formal element of all that can be thought is cancelled by
means of its opposite, that is to say, when it is self-contradictory; or, alternatively, when its
non-existence eliminates the material element and all the data of all that can be thought. The
former, as has been said, never occurs in the case of existence. It follows that, since there is no
third possibility, either the concept of absolutely necessary existence is a deceptive and false
concept, or it must rest on the fact that the non-being of a thing is at the same time the
negation of all the data that can be thought. (OPG, Ak. 2: 82)

This passage contains a number of important claims and themes and, before con-
tinuing, we need to separate and analyze them.
For Kant, as for the logicists, the PNC is the highest principle of logical possibility.
The principle of non-contradiction grounds the whole space of logical possibility.
Therefore, Kant reasons, if you eliminated or ‘canceled’ the PNC, nothing would be
logically possible. This is what it means to say that the principle of non-contradiction
is absolutely logically necessary. By contrast, a proposition is conditionally logically
necessary just in case it follows logically from the principle of non-contradiction. All
logically necessary propositions, i.e. all propositions whose negations entail contra-
dictions, are consequences of the principle of non-contradiction, and hence condi-
tionally logically necessary. To say that they are conditionally logically necessary is
not to impugn their necessity: it is only to draw a distinction between these logically
necessary truths, whose truth follows from the principle of non-contradiction, and
the principle of non-contradiction itself.3
Kant understands absolute real necessity on the model of absolute logical necessity:
it is absolutely really necessary that p just in case the ‘cancellation’ of p would cancel
all real possibility. An absolutely necessary proposition is a proposition that grounds
the whole space of real possibility. Likewise, an absolutely really necessary being is
one whose existence governs and gives rise to the entire domain of real possibility. If
there is such a being its non-existence would cancel all real possibility; without it,
nothing would be really possible. Conditionally really necessary truths are truths that

3
This is a hyperintensional relation of grounding within the space of logical possibility. See n. 1.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

follow by real necessity from absolutely really necessary truths. There may very well
be really necessary consequences of absolutely really necessary truths; likewise, there
may be beings whose existence follows with real necessity from the existence of the
absolutely necessarily existing being (if there is one).4 This in no way impugns the
necessity of these truths, or the necessary existence of these beings; the propositions
could not have been false, and the beings could not have failed to exist.5 It merely
draws a distinction between those propositions and those beings whose truth and
whose existence follow necessarily from an absolutely necessary being, on the one
hand, and absolutely necessary truths and the absolutely necessary being, without
which nothing would be really possible, on the other.
In light of these points, we can state Kant’s definition of absolute necessity:
(Defn.) It is absolutely necessary that p if and only if ¬p cancels all possibility,
and distinguish two different kinds of absolute necessity:
(Defn.) It is absolutely logically necessary that p if and only if ¬p cancels all logical
possibility.
(Defn.) It is absolutely really necessary that p if and only if ¬p cancels all real
possibility.
As we know from the previous chapter, logical possibility is a necessary but not a
sufficient condition for real possibility. But this appears to entail that anything that
cancels all logical possibility cancels all real possibility, so anything that is absolutely
logically necessary is absolutely really necessary. This cannot be correct, however,
because, while the negation of the principle of contradiction cancels all real possi-
bility, it does so by canceling the logical element in real possibility. Kant is interested
in whether there exists a being whose non-existence would cancel the real non-logical

4
Whether there are necessary consequences of God’s existence is related to the complex issue of
whether God necessarily creates the actual world, or whether there is some contingency in God’s choice
of this world. This, of course, is tied up with difficult questions about divine freedom and whether freedom
requires the possibility of doing otherwise. In Beweisgrund itself Kant does not come down definitely on
either side of this issue; at Ak. 2: 153–4 he appears to argue that necessarily, if God creates a world, he
creates the actual world (because it is the most perfect possible world), but this is far from a full-throated
endorsement. The issue of whether the actual world is the most perfect possible world is one on which Kant
worked extensively in this period. Refl. 3703–5 contain Kant’s notes for a planned essay on Pope’s
optimism, the topic of the 1755 prize essay set by the Berlin Academy; the editors’ Introduction to Kant
(2011) contains a detailed discussion of Kant’s work on optimism in this period (pp. xxxii–xli). Kant
continues to uphold the view that the actual world is the most perfect possible in the mid-1760s; see Refl.
3854, 3880, and 4226. Matters are complicated by the fact that Beweisgrund is a transitional text between
the compatibilist, indeed, even necessitarian, theory of freedom Kant defends in ND (Ak. 1: 398–405) and
the more incompatibilist conception of freedom he will defend in the CPR (see Ch. 7). In a number of
Reflexionen from the 1760s Kant seems to endorse a ‘mixed’ view, which combines an incompatibilist
theory of human freedom and a compatibilist theory of divine freedom; see Refl. 3855–9 (Ak. 17: 313–15,
from the years 1764–8), as well as Refl. 4033, 4034, 4156, 4157, 4180, 4184, 4218, 4226, 4228, 4334, and
4338, which Adickes dates to 1769. If this is correct then the actual world exists necessarily, although it does
not exist absolutely necessarily. For critical discussion, see Kain (forthcoming) and Hogan (2009a).
5
This is a hyperintensional relation within the space of real possibility.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

element in real possibility, so we should amend the definition of absolute real

necessity slightly to read:
(Defn.*) It is absolutely really necessary that p if and only if ¬p cancels the real
element in all real possibility.
What does it mean for possibility to be ‘canceled’? The most straightforward reading
of the relation of ‘canceling’ in this definition is that it refers to a necessary
conditional, i.e. ¬p cancels q if and only if □(¬p ⊃ ¬q). However, interpreting
cancellation in terms of a necessary conditional reduces Kant’s definition of absolute
necessity to necessity simpliciter: for any necessarily true p, □ (¬p ⊃ ¬q) is true for
any q. If ‘cancellation’ is interpreted this way, then any necessary falsehood cancels
any proposition whatsoever (including necessary falsehoods). The ‘necessary condi-
tional’ reading of cancellation eliminates the specific connection between absolute
necessity and the cancellation of all possibility.
I think we can do better if we adopt a counterfactual interpretation of ‘canceling’;
but we have to be careful about how we understand counterfactuals. For example, if
we interpret ‘cancellation’ in counterfactual terms and assume that □ (¬p ⊃ ¬q)
entails ¬p )¬q we still run into problems.6 This assumption is equivalent to assum-
ing that all counterfactuals with impossible antecedents are (vacuously) true. If we
make these two assumptions, then for any necessarily false p, p cancels all real
possibility, which once again reduces absolute necessity to necessity simpliciter: any
necessarily true proposition is absolutely necessary. Consequently, we should not
both accept a counterfactual interpretation of ‘cancellation’ and assume that all
counterfactuals with impossible antecedents are true.
The situation improves if we reject the assumption that all counterfactuals with
necessarily false antecedents are vacuously true and interpret Kant’s notion of a
‘cancellation of possibility’ as involving the notion of a non-vacuous counterpossible
conditional. In other words, on Kant’s conception, some things would be ‘canceled’ if
a necessary truth were false, while others would not be. For instance, there may be a
necessary truth p (6¼ God exists) such that if p were false, some, but not all, real
possibilities would cease to be possible. This means that the relation of cancellation
between propositions is ‘hyperintensional’: necessarily equivalent propositions can-
not be substituted into the sentence ‘p cancels q’ salva veritate. In the idiom of
possible worlds, cancellation is more ‘fine-grained’ than possible worlds.
While the Lewisian orthodoxy that once reigned in analytic metaphysics rejected
hyperintensional relations and non-trivial counterpossible conditionals, they are
finding increasing acceptance.7 One relevant motivation for accepting

6
Where ‘p ) q’ stands for the ‘would’ counterfactual conditional: if p were the case, it would be the case
that q.
7
I call this orthodoxy ‘Lewisian’ because Lewis (1986) identified propositions with sets of possible
worlds. However, other philosophers made room for hyperintensionality by introducing ‘impossible
worlds,’ e.g., D. Nolan (1997), Vander Laan (2004), Kment (2006), and Brogaard and Salerno (2013).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

hyperintensionality is dependence relations between necessary existents. Intuitively,

both the number 2 and the singleton set {2} necessarily exist, but the latter exists in
virtue of the former. This is a non-causal relation of metaphysical dependence; the
number 2, a causally inert object, cannot cause {2}, another causally inert object, to
exist, and the existence of 2 does not logically entail the existence of {2}.8 Translating
this into the Kantian language of cancellation, we could say: per impossibile, the non-
existence of 2 would cancel the existence of {2}. This counterpossible relation of
cancellation is not trivial, though, because it does not hold between the non-existence
of 2 and just any impossible state of affairs: the non-existence of 2 would not cancel
the fact that nothing is simultaneously red all over and green all over. The precise
nature of such hyperintensional relations is a difficult and controversial matter in
contemporary metaphysics.9
This newfound interest in counterpossible conditionals is, in a certain sense, quite
Kantian, because Kant himself, I have been arguing, accepts non-trivial counter-
possible conditionals, and what is more, thinks these counterpossible conditionals
are explained by grounding relations between necessary truths. To jump ahead
slightly, there is a unique absolutely necessary being, God, and the reason God’s
non-existence cancels all real possibility is that God’s existence grounds all real
possibility: a grounding relation between necessary truths (God exists, something is
really possible) explains a counterpossible cancellation relation between them (God’s
non-existence, nothing being really possible).
I have gone on at some length about this because some readers will be tempted to
dismiss the notion of ‘absolute necessity’ as I have defined it and reduce it to necessity
simpliciter.10 This definitely has the merit of making Kant interpretation easier.
However, it does so at the expense of conflating two notions that Kant seems to
want to keep distinct. At various points in Beweisgrund Kant distinguishes between
things whose non-existence cancels some real possibilities and things whose non-
existence cancels all real possibilities. For instance, “that existence, by means of
which the material element of all that can be thought is not given, and in the absence

Lange (2009) develops a non-Lewisian (i.e. not world-based) theory that also allows for non-trivial
counterpossible conditionals. However, there is also a growing interest in hyperintensional metaphysics,
centering around the notion of ‘grounding’: see Fine (1994a), Schaffer (2009), Rosen (2010), Audi (2012a)
and (2012b); skeptics about the grounding relation include Wilson (2014) and Koslicki (forthcoming). For
a survey of these issues, see Bliss and Trogdon (2014) and Raven (forthcoming).
8
That inference requires the axioms of set theory, which are taken by most philosophers to be extra-
logical.
9
For discussion see D. Nolan (2014) and Jago (2014).
10
Cf. Chignell (2009a), 167. By necessity simpliciter I mean necessity independent of source or ground;
in eighteenth-century terms, it is necessary simpliciter that p if and only if either p is in se necessary or is a
necessary consequence of some in se necessary truth. Kant’s theory of absolutely necessary existence is a
theory of in se necessary existence, not of necessary existence simpliciter (which would apply also to the
necessary consequences of an in se necessary being), much less of merely hypothetically necessary
existence.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

of which, therefore, there is still something left to be thought, that is to say, still
something possible—the opposite of such an existence is possible in the real sense,
and in the same sense it is also contingent” (OPG, Ak. 2: 83).11 Since Kant is also
committed to the principle that possibly p entails necessarily possibly p,12 anything
whose non-existence cancels some real possibility exists necessarily. If we collapse
absolute necessity and necessity simpliciter, then by definition, all such beings exist
absolutely necessarily; there would then be no modal distinction between things
whose non-being cancels some possibility and the unique being whose non-existence
cancels all possibilities. In my reconstruction of Kant’s argument in the next section,
I bear in mind the possibility of such a ‘reductive’ reading and argue that it would not
help Kant significantly. Interpretive charity, and fidelity to the text, therefore support
distinguishing absolute necessity and necessity simpliciter.
We are now in a position to understand Kant’s definition of absolute necessity:
(Defn.*) It is absolutely really necessary that p if and only if ¬p cancels the real
element in all real possibility,
which, I have argued, should be understood as:
(1C) It is absolutely really necessary that p if and only if, were it the case that ¬p,
there would be no real element in any real possibility.
An absolutely necessary proposition is a proposition whose negation cancels all
possibility. This entails that:
(1E) For any x, x exists absolutely necessarily if and only if, were x not to exist,
there would be no real element in any real possibility.
Absolutely necessary existence is always absolutely really necessary existence, since
nothing exists with logical necessity. Since I have been focusing, as Kant does in the
first Division of Beweisgrund, on the possibility of predicates, and, specifically, on the
possibility of atomic predicates, the specific version of this claim that we will need to
reconstruct Kant’s argument is:
(1AP) For any x, x exists absolutely necessarily if and only if, were x not to exist,
there would be no really possible atomic predicates.
Principle (1E) is stated at the end of I.3.1, which begins, as we have seen, with Kant
inquiring into the real definition of necessary existence. What is the relation between
(1E) and the project of giving a real definition of necessary existence? (1E) is a
principle about absolutely necessary existence, which is related to necessary existence
simpliciter as follows:

11
Cf. OPG, Ak. 2: 82, 99–100, and 162 f.
12
OPG, Ak. 2: 84. This principle, the characteristic axiom of S5, is discussed in more detail below in }3.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

(2) x necessarily exists if and only if (i) x exists absolutely necessarily or (ii) 9y
such that y exists absolutely necessarily and y is necessarily the ground of the
existence of x (x is conditionally necessary).
In other words, a being exists necessarily just in case it is absolutely necessary or is a
consequence of an absolutely necessary being. Clause (ii) has the paradigm form of a
real definition: if y exists necessarily and is necessarily the ground of the existence of x
then we have an explanation of why x exists necessarily. We might then ask ‘why does
y exist necessarily?’ and for this we are referred back to (i). But is (i) (or for that
matter (1AP)) a real definition? No, for it does not tell us, of a being that exists with
absolute necessity, why that being exists with absolute necessity.
The real definition of absolutely necessary existence is provided by Kant’s concept
of a ground of all real possibility (GARP): an absolutely necessary being is absolutely
necessary (its non-existence cancels all real possibility) because it is a ground of all
real possibility. The concept of a ground of all real possibility can provide a real
definition of absolutely necessary existence only if the grounding relation in which
God stands to all real possibility explains (is the ground of) the counterpossible
‘cancellation’ relation in which God stands to all real possibility. This means that
grounding entails cancellation:
(3) For any x, if x is a ground of all real possibility (GARP) then the non-
existence of x would cancel all really possible atomic predicates.
But if being a GARP is going to be the real definition of absolutely necessary existence
it must be the case that all absolutely necessary beings are GARPs, for otherwise there
could be a being whose absolute necessity is not explained by its being a GARP. In
other words:
(4) For any x, if the non-existence of x would cancel all really possible atomic
predicates, then x is a ground of all real possibility (GARP).
With these two pieces in place, we can state the real definition of absolutely necessary
existence:
(Real. Def.) For any x, if x exists absolutely necessarily, then this is so in virtue of
the fact that x is a ground of all real possibility (GARP).
This, combined with the definition of necessary existence simpliciter from above,
constitutes Kant’s complete alternative to the logicist real definition of necessary
existence (the containment of existence within a being’s essence).

5.3. The Only Possible Ground of Proof

In this section I reconstruct Kant’s modal argument for the existence of God. This is
what Kant calls a ‘proof-ground,’ a set of premises that entail the existence of God,
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

but which have not yet been turned into an explicit demonstration (a logically valid
argument) for that conclusion.13 Each sub-section corresponds to a sub-section of
Kant’s argument in Beweisgrund I.2–3.

(i) It is absolutely necessary that something exists (I.2.3)

The first thing Kant does is argue that it is absolutely necessary that something exists,
which he does in I.2.3, after having introduced the ‘material’ requirement on
possibility. His argument is relatively straightforward:
(1) It is absolutely necessary that p if and only if, were it the case that ¬p, no
atomic predicates would be really possible. [By definition – see section 2.]
(2) If F is a really possible atomic predicate, then there exists a substance that
grounds its real possibility.
(3) If F is a really possible atomic predicate and a substance grounds its real
possibility, then the non-existence of that substance cancels the real possibility
of F.
(4) ∴ If nothing were to exist, no atomic predicates would be really possible.
[From (2) and (3).]
(5) ∴ It is absolutely necessary that something exists. [From (1) and (4).]
The most important claim here is (1); since Kant is effectively introducing the
concept of absolute necessity through this premise, I think we should understand it
as a stipulative definition, and grant it to him. The other main premise in this
argument is (2) the principle that possibility requires an existing ground in a
substance. While not philosophically uncontroversial, it is the central assumption
of his argument, so I think we should grant it to Kant and see whether the argument
he builds upon it succeeds. The justification for (3) is that, if a predicate is possible,
then it is necessarily possible (Ak. 2: 84) and it is plausible to assume that possibility
facts have their grounds essentially; nothing but the actual ground of F could ground
its real possibility.14 At this point in the argument, no specific view about what the
existing grounds of real possibility are needs to be appealed to, other than that they
are substances (rather than predicates of substances); even someone who holds the
‘intellectualist’ model (see Chapter 4.3) should accept (5).15

13
As Kant explains at Ak. 2: 66, this work does not contain a demonstration of the existence of God but
only the ‘proof-ground’ (Beweisgrund) for a possible demonstration (cf. Ak 2: 65, 67). As Kant explains in
his logic lectures, this refers to a technical distinction between the proposition to be proved, the proof-
ground (argument), and the logical form by which the conclusion follows from the proof-ground (WL, Ak.
24: 892; cf. JL, Ak. 9: 71, and LPö, Ak. 24: 561). Chignell (2009a), 161 n. 9 points out that this is a discussion
of }191 of Meier’s Auszug. In this book, Kant is only giving a Beweisgrund, a set of premises from which a
conclusion (the existence of God) follows, without giving it the ‘form’ of a demonstration, i.e. without
turning into a formally valid proof (cf. Ak. 2: 88–9).
14
Thanks to Catharine Diehl and Tobias Rosefeldt for pressing me on this point.
15
Cf. Schönfeld (2000), 201–6.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

(ii) There is an absolutely necessary being (I.3.2)

Having proven the de dicto claim that (absolutely) necessarily something exists, Kant’s
next argumentative task is to prove the de re claim that something necessarily
exists.16 But, given Kantian premises, this is relatively straightforward. Note the
short space of text in which Kant moves from the de dicto to the de re claim:
all possibility presupposes something actual, in and through which the thinkable is given.
Therefore [demnach], there is a certain actuality whose cancellation would itself cancel all
possibility. However, that whose cancellation or negation eliminates all possibility, is absolutely
necessary. Therefore, something exists absolutely necessarily [existiert etwas absolut notwen-
digerweise] (OPG, Ak. 2: 83)

The first sentence should be read as making the de dicto claim that (absolutely)
necessarily something exists (if nothing existed, nothing would be really possible),
the conclusion of the immediately preceding section; but by the end of this passage,
without much argument, Kant is making the de re claim that some being exists
(absolutely) necessarily. This may give the impression that Kant, disappointingly, is
just conflating these claims, but this is unlikely for two reasons. First, he made the
de dicto claim several pages earlier in I.2.3, a section whose title is “It is absolutely
impossible that nothing exists,” which is naturally read de dicto. The quoted passage
occurs under the heading “there is an absolutely necessary being,” which is naturally
read de re. Secondly, as I will now argue, given his premises, the de re claim is an
almost immediate consequence of the de dicto claim.
If we take all of the beings that are grounds of some real possibility (GSRPs in the
terminology of the previous chapter) and treat them collectively as one being, then
that being is a ground of all real possibility (GARP) and exists with absolute necessity.
All really possible atomic predicates are grounded by some part of this collection and
if this collection were to be completely annihilated (if all of its parts were annihilated)
it would cease to exist. Some care is required here. We need this collection to be such
that annihilating any proper part of it does not annihilate the collection itself; we
need it to be such that the only way of annihilating it is to annihilate every one of its
proper parts. In contemporary metaphysics there are various ways of doing this;
perhaps the easiest is to take what Fine (1994b) calls the aggregate of the GSRPs.17
The aggregate of the Xs exists if and only if at least one of the Xs exists; unlike
classical mereological fusions, aggregates are as mereologically non-rigid as can be:
they can exist if even one of their parts exists. Call the aggregate of all GSRPs Ω. The

16
As Abaci (2014), 14–15, points out.
17
See Fine (1994b) and (1999). As I am using the term, the aggregate of the Xs exists if and only if at
least one of the Xs exists. Fine originally introduced aggregation in terms of tense: the aggregate of the Xs
exists at time t if and only if at least one of the Xs exist at t. My notion of aggregate is a generalized non-
temporal version of the Finean notion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

absolute necessity of Ω is secured by the following principle, which was premise (3)
in the argument of I.2.3:

(3) If F is a really possible atomic predicate and substance x grounds its real
possibility, the non-existence of x cancels the real possibility of F.

By definition, Ω exists if and only if at least one GSRP exists, so the cancellation of Ω
is equivalent to the annihilation of all GSRPs, which would cancel all really possible
atomic predicates. So it follows that Ω exists with absolute necessity. Since Ω contains
all GSRPs, it grounds all real possibilities (it is a GARP).

(iii) It is unique (I.3.3)

Kant’s next step is to prove that there is a unique absolutely necessary being:

Let A be a necessary being and B another. By definition [vermöge der Erklärung] B is only
possible, insofar as it is given as the consequence of another that grounds it, A. By assumption,
B itself is necessary, so its possibility is in it as a predicate and not as a consequence of another,
but yet, according to what was just shown, it is a consequence, and this is a contradiction.
(Ak. 2: 84)

Kant formulates this argument in terms of one substance grounding another, but his
modal metaphysics up to this point has been formulated in terms of a substance
grounding the possibility of predicates. Let us attempt, then, to reconstruct this
argument in terms of substances grounding one another’s predicates:
(1) Assume A and B are absolutely necessary beings and A6¼B.
(2) Anything that exists absolutely necessarily is a ground of all real possibility
(GARP). [See section 2.]
(3) If x is a GARP then for any possible predicate F either (i) F is instantiated by x
and the possibility of F has no ground (F is fundamental) or (ii) the possibility
of F is grounded in x (F is derivative). [Defn. of GARP—see Ch. 4.2.]
(4) ∴ The fundamental predicates of A have no ground and the fundamental
predicates of A are grounded in B. [From (1)–(3).] This is a contradiction.
(5) ∴ If A and B are absolutely necessary beings then A=B.
In section 2 I argued that Kant is entitled to (2). However, even without (2) he can
prove directly that there cannot be two distinct GARPs, by using (3)–(5) to reduce to
absurdity the assumption that:
(1*) Let A and B be GARPs and A6¼B.
Note, though, that the reductive reading, on which absolute necessity is equivalent to
necessity simpliciter, should reject (2) as a mistake on Kant’s part. For (2), on the
reductive reading, entails:
(2*) Any necessary being is a GARP.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

But this has the problematic consequence that if x is a GARP and x necessarily causes
y to exist, then y is a GARP as well. This is not a consequence that Kant, or any
metaphysician, should embrace. For it would entail that, if God necessarily creates
the actual world, as, on some theories of the divine will, he does, then the world itself
is a GARP. While Kant is not committed to such a necessitarian theory of creation in
Beweisgrund,18 he should not be forced to the conclusion that if God necessarily
creates the actual world then the actual world is a GARP. A charitable reading of Kant
should eschew commitment to (2*); consequently, we should avoid if possible a
reading of Kant on which he is committed to identifying absolute necessity with
necessity simpliciter.

(iv) It is simple19
In particular, Kant must show that the GARP is a single substance, rather than an
aggregate of substances. Unless he can show this, Kant will not have shown anything
that many atheist metaphysicians would reject. For instance, some contemporary
metaphysicians (e.g., Fine) hold that possibilities are grounded in essences; the aggre-
gate of these essences is, in Kant’s terminology, absolutely necessary: if you cancel all
of them, you cancel all possibility. But few such metaphysicians would think there is a
single such essence that grounds all real possibility.20 This is the final stage of the
argument that I will reconstruct; I will leave out of consideration Kant’s further
argument that the GARP has the other traditional divine attributes: intellect, will, etc.
Kant must argue against an alternative picture on which possibility is parceled out
among a plurality of substances, each of which grounds some possibilities (GSRP) but
not all possibilities (GARP). I will call this modal pluralism. Kant’s principal argu-
ment against the modal pluralist is the following:
If one were to appeal to the definition of the necessary being and say that in each part the data
of some inner possibility is given, but in all of the parts together all possibility is given, one
would be imagining something wholly, albeit covertly, incoherent. For if one then thought that
some inner possibilities could be canceled, while others, given through other parts, remain, one
would have to suppose that it is in itself possible for inner possibility to be negated or canceled.
But it is absolutely unthinkable and contradictory that something be nothing, and this means
that canceling any inner possibility eliminates all that is thinkable. It is apparent from this that
the data for all that is thinkable must be given in the thing whose cancellation is the negation of

18
See Ak. 2: 153–4; cf. Refl. 3830.
19
Kant represents the conclusion of this argument as the simplicity [Einfachheit] of the unique GARP
but he never argues that the unique GARP lacks parts entirely; he argues that the unique GARP is not
composed of parts, each of which grounds some subset of the space of possible predicates. It is compatible
with everything for which Kant explicitly argues that the unique GARP has parts, as long as they play no
separable role in its grounding of possibility. Thanks to Manish Oza for pressing me on this point.
20
Whether they would agree that essences are substances in Kant’s sense—bearers of predicates/
properties that are not themselves predicates/properties—is another matter (see Fine 1995, 66–7). But
such metaphysicians might be able to accept all of Kant’s arguments up to this point provided we delete the
assumption that the grounds of possibility are substances (rather than bundles of predicates/properties).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

all possibility; therefore, that which contains the ultimate ground of any inner possibility
contains the ground of all possibility whatsoever, and this ground cannot be divided into
distinct substances. (OPG, Ak. 2: 84)

Kant argues against the plurality view, in this passage, by claiming that if something’s
non-existence cancels some possibilities then it cancels all possibilities, and conclud-
ing from this that if something grounds some possibilities then it grounds all
possibilities (it is the GARP). More formally, his reasoning could be represented as
follows:
(1) Let x be a ground of some real possibilities (GSRP).
(2) If x grounds some real possibilities, its non-existence cancels some real
possibilities (those it grounds). [By premise (3) in section (i) above.]
(3) If x’s non-existence cancels some real possibilities, its non-existence cancels
all real possibilities (“ . . . this means that canceling any inner possibility
eliminates all that is thinkable”).
(4) ∴ If x grounds some real possibilities, it exists absolutely necessarily. [From
(2) and (3) and the definition of absolutely necessary existence from }2.]
(5) If x exists absolutely necessarily, then it is a ground of all real possibility. (“It
is apparent from this that the data for all that is thinkable must be given the
thing whose cancellation is the negation of all possibility.”)
(6) ∴ For all x, if x grounds some real possibilities, it grounds all real possibilities
(“that which contains the ultimate ground of any inner possibility contains
the ground of all possibility whatsoever”). [From (4) and (5).]21
From this conclusion, Kant can easily prove that Ω, from above, has no parts. From
(6) it follows that any GSRP is a GARP and Kant has already proven (see section (iii))
that there cannot be two GARPs. So there is only one GSRP and Ω is identical to it (it
is the aggregate of only one object). In fact, (1)–(6), if successful, would allow Kant to
bypass entirely the proof in (ii) that required the introduction of Ω in the first place,
for it would allow Kant to argue as follows:
(7) Assume: some atomic predicate F is really possible.
(8) If F is really possible there exists an x such that x is the ground of the
possibility of F (GSRP).
(9) Any GSRP is a GARP. [Conclusion of previous argument.]
(10) There cannot be more than one ground of all real possibility. [Proved in
section (iii).]
(11) ∴ There is a unique GARP.
However, the original argument (1)–(6) above is not successful, so the point is moot.
Premise (3), in particular, begs the question against the pluralist. According to

21
Cf. the corresponding argument in ND, Ak. 1: 395.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

pluralism, the complete space of possibilities is ‘parceled out’ among a plurality of

grounds, so that the non-existence of any one of them would cancel only some
possibilities; it would beg the question to assume that anything that cancels some
possibilities cancels all of them. Kant does not succeed in establishing his key claim,
(6), at Ak. 2: 85.22
A Kantian rejoinder to this objection might go as follows. In the passage quoted
above, Kant writes: “it is entirely inconceivable and self-contradictory that something
should be nothing.” Since ‘something,’ in the context of Beweisgrund, means ‘some-
thing (really) possible,’ and ‘nothing’ means ‘not (really) possible,’ we could interpret
this as a variant of the S5 axiom in modal logic: all possibilities are necessarily
possible.23 If there is a ground of some real possibilities (a GSRP) then its non-
existence cancels some possibility, and since possibilities are necessarily possible, the
non-existence of this ground is impossible. So any GSRP exists necessarily. A being
exists necessarily just in case its non-existence cancels all possibility. If a being is such
that its non-existence cancels all possibility, this is because it is a ground of all
possibility; so any being that grounds some possibility grounds all possibility.
There are several places, however, where this argument is vulnerable. Using the S5
axiom we can prove straightforwardly that any GSRP exists necessarily; that much of
the argument is valid. But to prove that it exists with absolute necessity we must
either explicitly collapse absolute necessity and necessity simpliciter or make assump-
tions that entail this collapse. For instance, if we assume that, if a GSRP exists
necessarily then the non-existence of a given GSRP stands in the counterpossible
cancellation relation to any proposition whatsoever (including the proposition that
something is really possible),24 then we can prove that any GSRP exists with absolute
necessity. It is true that in the context of this argument Kant does seem to collapse
this distinction (“this is tantamount to saying that canceling an internal possibility is
the same thing as eliminating all that can be thought”). But even if we grant Kant this
modal collapse the argument still begs the question against the pluralist. For the
modal pluralist can grant that any GSRP exists with absolute necessity but very
plausibly deny that it grounds all real possibility.25 The idea would be this: there are
multiple grounds of possibility, the cancellation of any one of which would cancel
some (real) possibilities; since possibilities are necessarily possible (S5 axiom), these
grounds of possibility each exist necessarily. The non-existence of any of them is
impossible, and any counterfactual with an impossible antecedent is trivially true
(thus collapsing the necessary/absolutely necessary distinction), so each of these
grounds of possibility satisfies Kant’s definition of absolute necessity. However,
although each of these grounds stands in a counterfactual relation to all of possibility,

22
This problem is pointed out by Adams (2000), 433.
23
◇p ⊃ □◇p. Cf. Ak. 17: 252.
24
Equivalently, assuming that □(¬p ⊃ ¬q) entails ¬p ) ¬q.
25
The modal pluralist does not need to accept (4) from }2. Adams (2000) makes an equivalent point.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

none of them stands in a grounding relation to all of possibility. The grounding

relation between these substances and possibilities is more fine-grained than the
counterfactual relation; each substance bears the former relation only to a subset of
the possibilities. So reducing absolute necessity to necessity simpliciter and adopting
the S5 axiom does not actually help Kant’s argument here. I think, therefore, that the
most charitable interpretation is that, although Kant collapses that distinction in this
passage (“this is tantamount to saying that canceling an internal possibility is the
same thing as eliminating all that can be thought”) he does so because he (mis-
takenly) thinks it will help his argument, and, in light of the many considerations that
support this distinction (section 2), his considered view is that absolute necessity is
not equivalent to necessity simpliciter.26
I am not the first commentator to notice that the ‘plurality’ objection poses
significant problems for Kant’s argument. In the next section I explore some solu-
tions that have been proposed in the secondary literature.

5.4. Prior Replies to the Plurality Objection

The secondary literature on Beweisgrund has spawned some sophisticated proposals
for answering the plurality objection on Kant’s behalf.27 Perhaps the best place to
start is Chignell’s solution, which rests on his ‘real harmony’ principle:
(Real Harmony) If F and G are fundamental predicates and F and G are really
compatible then the real compatibility of F and G must be grounded in their co-
instantiation by a single substance.28
In the previous chapter I criticized Chignell’s interpretation on textual and philo-
sophical grounds. In this context it suffices to point out that Real Harmony will only
defeat the plurality objection if we assume additionally:
(Compatibility) If F and G are fundamental predicates (predicates whose possi-
bility is given in a substance that instantiates them) then F and G are really
compatible (it is really possible that they are co-instantiated).

26
Nor is the slightly different argument Kant offers earlier in the same paragraph any more successful
against modal pluralism (Ak. 2: 84); it is a variant of the idea, deriving from the fifth of Aquinas’ ‘five ways’
(Summa Theologica I.q2.a3), that if all beings are contingent then it is possible that nothing exists. For
reasons of brevity I forgo further discussion of it here.
27
To my knowledge, the first to point out that it is a problem were Watkins and Fisher (1998), who
write: “while a contingent being can ground some possibilities, that is, those that arise from the material
elements given through it, it cannot ground its own possibility or the absolute possibility of anything else”
(Watkins and Fisher 1998, 375). However, this appeals to a different conception of contingency. Kant
needs to show that an absolutely contingent being—a being whose non-existence does not cancel all
possibility—cannot ground some possibilities. Adams (2000) also contains a highly influential discussion
of this problem.
28
Chignell (2009a), 187; (2012), 648.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

But Chignell gives no reason why the pluralist must admit this. Note that Compati-
bility does not say merely that some fundamental predicates are really compatible; it
says that all fundamental predicates are collectively compatible, that is, they can all be
instantiated by the same substance. So Compatibility cannot be established by
pointing to a few examples, such as the compatibility of understanding and will,
which Kant asserts at Ak. 2: 87. Far from being a weak or uncontroversial principle,
Compatibility is in fact extremely strong, and Kant would need a strong argument to
justify it. In particular, the pluralist might hold that distinct fundamental predicates
are instantiated by distinct substances (distinct GSRPs) and are not compatibile.
Perhaps Chignell would respond to the pluralist by asking what grounds the real
harmony of derivative predicates on this picture. As we saw in the previous chapter,
this is precisely the point on which the coherence of Chignell’s own view is threat-
ened.29 But notice the pluralist does not need to admit that derivative predicates are
compatible if they are grounded in different GSRPs. The pluralist could hold a
modified Cartesian view on which each of the attributes is a separate substance in
its own right that instantiates various fundamental predicates and grounds various
derivative predicates. So, for instance, the attribute of Thought can ground the
possibility of various modes of thought, and the attribute of Extension can ground
the possibility of various modes of extension, but no thought mode is compatible
with an extension mode. This is not an especially attractive picture of how possibil-
ities are grounded, but Kant claims not merely that it is rationally unappealing but
that it is demonstrably false.
Yong (2014) steps into the breach with an innovative proposal: Kant is searching
for the ground not merely of individual possibilities (really possible predicates) but of
the space of possibilities as a totality and, thus, the modal relations among possibil-
ities (such as compatibility, incompatibility, entailment, etc.). Yong makes some
intriguing remarks about the idea that possibility as such constitutes a unified
totality, but it is unclear how to make this idea dialectically effective against the
pluralist, who, after all, might deny that the space of possibilities constitutes a unified
totality.30
Yong’s more concrete proposal appeals to the doctrines of Kant’s pre-Critical
rational cosmology to defeat the plurality objection.31 Let us assume that A and B are
distinct grounds of real possibility, so A grounds some really possible predicates (call
them the a-predicates) and B grounds a distinct set of predicates (call them the
b-predicates). Yong points out that A and B both exist (by definition of ground of

29
Cf. Yong (2014), 33–4, and Chignell (2012), 665–7.
30
Yong (2014), 38–40. Yong’s reconstruction also threatens to collapse the distinction between Kant’s
proof from the mere fact of possibility (contained in Beweisgrund I) and the proof from the unity in the
essences of things (contained in Beweisgrund II). See Ak. 2: 155 for more on the distinction. The harmony
among possibilities is presented as a premise only of the latter, while Yong claims it as a premise of the
former as well.
31
Yong (2014), 41–4.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

possibility) and therefore stand in relation to one another. He then appeals to Kant’s
cosmological doctrine that substances’ relations to one another are not brute facts but
must be grounded in some causal relation between A and B. In particular, for A and
B to be related to one another, their relation must be grounded in a third substance
C. Yong goes on to argue that, on pain of infinite regress (what grounds the relations
among A, B, and C?), we should identify A and B and accept that there is only one
ground of possibility.
While I think Yong is right to invoke Kant’s pre-Critical cosmology to respond to
the plurality objection, the details of his reconstruction are problematic. First of all,
Kant’s doctrine that substances do not ‘relate’ to one another solely in virtue of
existing, but require an external ground of their ‘relation,’ applies only to substances
that are members of the same world. In other words, Kant’s notion of relation is more
substantive than the contemporary post-Russellian logical notion of relational predi-
cate. In fact, it is one of Kant’s consistent pre-Critical contentions that merely in
creating a plurality of existing substances God does not thereby make them into
members of a single world.32 But Yong has given no argument that the pluralist must
admit that distinct grounds of possibility are members of the same world in Kant’s
specific technical sense: a totality of mutually interacting substances.
It is crucial to realize that Kant’s notion of a world (k-world) is very different from
the notion of a ‘possible world’ in modal logic (m-world). The modal logical concept
of a possible world is ultimately the concept of a ‘modal index,’ an element in a model
theory for modal logic. By contrast, a Kantian world is a concrete object: a collection
of substances unified by mutual causal interaction. Perhaps the clearest way to see the
difference is to note that the necessity of God’s existence means that God exists ‘in’
every m-world (every counterfactual situation) but God is not a part of any possible
k-world (because he does not reciprocally interact with other substances—their
causal dependence on him is one-way). Similarly, Kant holds that God can create
two separate non-interacting k-worlds; in contemporary terms, this means there is an
m-world in which there are two k-worlds (and God).33 K-worlds are not modal
indices (the original function of m-worlds in modal logic) or even truth-makers of
modal claims (a service into which they were pressed by post-Kripkean discussions of
modal metaphysics).34 To return to Yong and the plurality objection, if two grounds
of possibility exist then they are members of the same m-world (trivially), but it does
not follow that they are members of the same k-world. In the next section I try to
improve on Yong’s proposal and explain how Kant’s pre-Critical cosmology can be
used to answer the plurality objection. In what follows, ‘world’ will always mean

32 33
LF, Ak. 1: 21–2; ND, Ak. 1:410, 415; ID, Ak. 2:407, 410. Ak. 1: 414.
34
This does not mean concrete worlds cannot play the role of modal truth-makers; in David Lewis’s
modal theory, possible worlds are conceived in a way very similar to k-worlds (maximal mereological sums
of spatiotemporally connected concrete objects) and modal operators are quantifiers over these worlds.
The differences between these two non-Kantian notions of ‘possible worlds’ are brought out very clearly in
Burgess (2012), 143–56.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

k-world (neither possible worlds as modal indices, nor possible worlds as the truth-
makers of modal sentences, will play any role in my argument).

5.5. E Pluribus Unum

In the previous section I criticized Yong for assuming that if there are multiple
grounds of possibility (GSRPs) then they exist in the same world. I will argue that a
Kantian response to the plurality objection should appeal instead to a weaker but
similar principle:
(Recombination) It is really possible for there to be a world in which all the
fundamental atomic predicates are instantiated (though not necessarily
co-instantiated).
Intuitively, the fundamentality of predicates entails that they do not stand in modal
exclusion relations with one another: if fundamental predicate F is instantiated in a
k-world, this does not necessitate that fundamental predicate G is not instantiated in
that world (though it may necessitate that F and G are not instantiated by the same
object). The pluralist view is that there is a plurality of distinct substances, each of
which possesses some of the fundamental predicates, and no single substance that
possesses all of them (no GARP). I do not think it begs the question against that view
to assume Recombination.
In fact, we need to reformulate Recombination slightly because, as stated, it begs
the question against Kant himself, for Kant thinks that God cannot be part of any
world, so any fundamental predicate that only God can possess will be a counter-
example to Recombination as stated. Fortunately the repair is quite simple:
(Recombination*) It is really possible for there to be a world in which all of the
fundamental predicates that can be instantiated by a being other than God (a
GARP that lacks GSRP-parts) are instantiated (though not necessarily co-
instantiated).35
Since the only GARP the pluralist accepts is an aggregate of all the GSRPs, this claim
is equivalent, for the pluralist, to Recombination above. It is this recombination
principle that I will use to reconstruct my response to pluralism. Furthermore, I take
the following constraint on fundamental predicates to be prima facie plausible and
not question-begging against the pluralist:
(Fundamentality) If F is a fundamental predicate and x is the ground of the
existence of something that instantiates F then x instantiates F.

35
Recombination should not be confused with various stronger Humean Recombination principles that
Kant would reject, e.g., that “patching together parts of different possible worlds yields another possible
world” (Lewis 1986, 87–8). Neither Recombination nor Recombination* entail Humean Recombination.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

This is merely the idea that nothing that lacks a fundamental predicate can ‘intro-
duce’ that fundamental predicate into the space of possible predicates (e.g., by
grounding the existence of an instance). The pluralist should embrace this principle,
for it excludes a view on which the GSRPs that instantiate various fundamental
predicates themselves have grounds that lack these predicates.
Before I explain my own reply to the plurality objection, I want to explore in a
little more detail two principles of Kant’s pre-Critical cosmology that will be crucial
to my argument. The first, which we have already touched upon, is Kant’s definition
of world: a world is a totality of mutually interacting substances. That a world is a
totality means that it is maximal: anything that interacts with a part of a world is a
part of that world.36 That substances in worlds mutually interact means that they
are reciprocal causes of one another’s accidental predicates.37 The second major
piece of Kant’s pre-Critical cosmology to which I will appeal is a principle about
the possibility of mutually interacting substances, which he calls the ‘principle of
co-existence’:
Finite substances do not, in virtue of their existence alone, stand in a relationship with each
other, nor are they linked together by any interaction at all, except insofar as the common
principle of their existence, namely the divine understanding, maintains them in a state of
harmony in their reciprocal relations. (Ak. 1: 412–13)38

This might not seem like an auspicious premise for Kant’s Beweisgrund argument,
because it explicitly mentions God, the very being whose existence he is trying to
establish. But Kant’s reasons for holding this principle do not depend upon the
conclusion of that argument. He holds the principle of co-existence in order to
solve a standard early modern problem in the metaphysics of causation: how to
reconcile the modal independence of finite substances with the possibility of caus-
ation as a necessary connection between them.39 His solution is that causation
between finite substances is not a necessary connection tout court; finite substances
can exist without interacting with one another. For a set of finite substances to
interact, there must be a substance that causes them to exist and ‘places’ them in
interaction. Kant, of course, thinks this third substance that grounds the interaction

36
MH, Ak. 28: 40. God is saved from world-membership by the fact that his causal influence is one-way:
he causally influences finite substances, but they do not influence him.
37
MH, Ak. 28: 25–6.
38
The principle of succession (Ak. 1: 410), which states that a finite substance can undergo internal
change only if it is in interaction with other finite substances, will not play a role in my reconstruction of
Kant’s argument, so I do not discuss it here.
39
See ND, Ak. 1: 413, where Kant argues for the principle of Co-existence, which I reconstruct in what
follows; cf. the parallel discussion of the principle of co-existence in MH (Ak. 28: 51–2, 54). I agree with
Watkins (2005), 140–9 that the crucial issue is the modal independence of substances (rather than the
reducibility of relations); Langton (1998), 107–9 focuses on the reducibility of relations and interprets
Kant’s primary target as Leibniz (rather than Crusius’s doctrine of existential grounds—see Watkins
(2005), 145. For further critical discussion see Laywine (1993), 37–49, Schönfeld (2000), 149–54, and
Insole (2011).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

among finite substances is God, but for our purposes we do not need to assume that,
nor do we need to understand how God grounds the interaction of substances.40 All
we need is the principle that if substances interact then there is a substance that is the
ground of their existence. In fact, we need a modal corollary of that principle:
(Modal corollary) If there exists a plurality of substances that really possibly
interact then there exists a substance that is the ground of their existence.
This is a natural corollary of the Co-existence principle, given Kant’s background
commitment to the principle that real possibilities require existing grounds: if A and
B possibly interact, some substance makes this possible; by the Interaction principle it
is not A or B themselves, so it must be some being that is the ground of their
existence. This reasoning can be easily extended to an indefinite plurality of inter-
acting substances. This is not a principle that the modal pluralist as such is commit-
ted to rejecting, so, by appealing to it, Kant would not be begging any crucial
questions against the pluralist.
With the Interaction principle and the Recombination principle in place, it is not
difficult to formulate a Kantian argument against modal pluralism:41
(1) Let S be the set of all fundamental predicates instantiated by all the GSRPs.
(2) ∴ It is really possible that there is a world in which all of the predicates in S are
instantiated. [By Recombination*.]
(3) Necessarily, if the predicates in S are instantiated in a world, then their
grounds (the GSRPs) exist in that world and are world-mates. [Assumption.]
(4) Necessarily, any two world-mates interact. [By the definition of ‘world’.]
(5) ∴ Possibly, the GSRPs interact. [From (2)–(4).]
(6) ∴ There is a substance that grounds the existence of the GSRPs. [From (5) and
the Modal Corollary.]
(7) Every fundamental predicate is instantiated by some GSRP. [By definition.]
(8) ∴ There is a substance that instantiates every fundamental predicate. [From
(6) and (7), by Fundamentality.]42
Call this substance θ. To show further that θ is a GARP Kant would have to show
not only that it instantiates every fundamental predicate but that every fundamental
predicate is fundamental to it, that is to say, that there is no distinct being that makes
those fundamental predicates possible by instantiating them and grounding the
existence of θ. But Kant does not need to show that θ itself is a GARP; he only
needs to show that there is a GARP. What Kant needs to exclude is the possibility that
there is an infinite ascending chain of beings, each of which instantiates every

40
Kant attributes this to a “schema of the divine understanding” at Ak. 1: 414. For discussion see
Watkins (2005), 149–55 and Insole (2011), 413–19.
41
This argument is offered as a refinement of Yong (2014)’s proposal.
42
Compare my reconstruction of this argument to Kant’s refutation of Manicheanism in ND, Ak. 1:
414–15.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

Fundamental predicates Derivative predicates

F1 D1

F2 D2

...... θ3 θ2 θ1 θ0 F3 D3
. .
grounding . .
. .
R
instantiation
Fig. 5.1. An infinite ascending chain of maximal beings

Fundamental predicates Derivative predicates

F1 D1

F2 D2

θ ... θ2 θ1 θ0 F3 D3
. .
grounding . .
. .
R
instantiation
Fig. 5.2. A series of maximal beings terminating in a GARP

fundamental predicate; for the sake of simplicity, I will refer to such beings as
‘maximal’ beings. Kant needs to exclude, therefore, the view shown in Figure 5.1.
From the Kantian perspective, none of the beings on the left is absolutely necessary;43
if you canceled any one of them, the beings further to the left would still exist, and every
really possible predicate would still be possible (fundamental predicates would still be
instantiated, and the derivation relations would still hold). If Figure 5.1 represents the
metaphysical structure of the grounds of real possibility, there is no GARP.
To exclude such a view, Kant would have to appeal to a strong form of the
principle of sufficient reason, for note that in Figure 5.1 every derivative predicate
is grounded in some fundamental predicate, every fundamental predicate is instan-
tiated, and every maximal being is grounded in another maximal being. So Kant
would have to appeal not only to the principle that everything has a ground but one
to the effect that there are no infinite ascending chains of grounds, or, in the
terminology he will deploy in the CPR, that every series of conditions (grounds)
terminates in an unconditioned condition: a ground that has no ground.44 This
would license Kant in concluding that the structure of grounding relations is as
shown in Figure 5.2.

43 44
Though the aggregate of them is; see section 3. A416–8/B444–6.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

The being that terminates the chain of grounds on this picture is a GARP: it
instantiates every fundamental predicate and it has no ground. If it were not to exist,
nothing would be really possible (because none of the maximal beings would exist,
and thus none of the fundamental predicates would be instantiated).
Before continuing, I want to consider an objection to my reconstructed Kantian
argument against modal pluralism. The most questionable premise in this argument,
from the pluralist point of view, is (3). Why must the pluralist admit that if
substances A and B ground the instantiation of F and G in the same world (by
substances that are world-mates) then A and B are members of that same world?45
The pluralist could claim instead that the finite substances whose instantiations of
F and G are grounded in A and B are world-mates but A and B are not parts of any
world (a pluralist version of the Kantian doctrine that God grounds the existence of
substances in worlds but is not a part of any world). A and B, after all, do not interact;
their grounding role is one-way: they ground instantiations of F and G, and deriva-
tive predicates (see Fig. 5.3).
Although on such a view A and B do not technically interact with any finite
substances (or with another) because interaction is mutual and finite substances are
not the grounds of any accidents in A and B, Kant could point out that they bear the
following relation to one another: they are the grounds of substances that interact,
and he could reasonably ask what is the ground of the fact that they stand in this
relation to one another. After all, the fact that A and B are able to ground substances
that causally interact cannot itself be groundless. There must be some ground of this
channel of influence and it cannot be A or B (because the pluralist is arguing that
A and B do not interact). Kant will reply that it must be some third substance that
grounds the existence of A and B.46
Perhaps a better pluralist response would be to reject the Recombination principle
altogether and claim that fundamental predicates are never instantiated by

A (instantiates F) B (instantiates G)

S1 S2

Interaction

Fig. 5.3. Extramundane grounds of possibility

45
This is a version of the problem I pointed out for Yong (2014)’s reconstruction.
46
Cf. “suppose that [the substances that compose a world] are caused by a number of necessary beings;
the effects, the causes of which are free from reciprocal relation, would not be in interaction” (ID, Ak. 2:
408).
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

substances that compose worlds but are instead instantiated by extramundane

substances that stand outside of worlds but ground substances in worlds. In
Figure 5.3, this means thinking of substances S1 and S2 as instantiating only deriva-
tive predicates. But the Kantian response to this move will be the same: although on
this view substances A and B do not mutually interact (they do not cause alterations
in one another), they are grounds of substances that interact. What grounds the fact
that the very substances whose predicates (e.g., modes of thought) are grounded by
the fundamental predicates of substance A (e.g., the attribute of thought) can
interact with beings whose derivative predicates (e.g., modes of extension) are
grounded by the fundamental predicates of substance B (modes of thought)? The
pluralist might reject the assumption that these modes interact at all, but the pre-
Critical Kant takes himself to have proven, contra Leibniz and other rationalists,
that members of the same world must mutually interact. So there must be a ground
of the fact that A and B can ground members of the same world, and, Kant will
argue, the only being capable of grounding this fact is a being that grounds both of
them. As he says in the Herder metaphysics lectures: “if several causes [Urheber]
posit several things, then these causes must be connected among themselves
[verknüpft unter sich], because their consequences are connected, and thus they
must not be absolutely necessary, because they have a common cause [alle zusam-
men eine Urheber haben]” (Ak. 28: 129). Since A and B, by hypothesis, instantiate
fundamental predicates, their common ground must possess both predicates (see
Fig. 5.4).
Obviously, this argument rests on Kant’s own pre-Critical cosmology, so any
pluralist who rejects that metaphysical theory can escape this argument. However,
as I argued above, that cosmology is not prima facie question-begging against the
modal pluralist. The point of this reconstruction, though, is not to show that Kant
decisively demonstrated that there is a unique ground of all real possibilities. It is to
show that he could have taken himself, plausibly, to have good reasons, from within

C (instantiates F and G)

A (instantiates F) B (instantiates G)

S1 S2

Interaction

Fig. 5.4. Extramundane grounds with a common ground

OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

his own metaphysical system, for asserting that possibility could not be ‘parceled out’
among a plurality of grounds.47

5.6. How (Not) to Represent God

My Kantian rejoinder to the plurality objection appealed to a modal corollary of the
Co-existence principle and a constraint on fundamentality:
(Modal corollary) If there exists a plurality of substances that possibly interact
then there exists a substance that is the ground of their existence.
(Fundamentality) If F is a fundamental predicate and x is the ground of the
existence of something that instantiates F, then x instantiates F.
But note that the modal corollary is a principle about causal grounding: there must be
a common cause of the existence of a plurality of (possibly) interacting substances.48
If the argument from section 5 is to be valid, then Fundamentality must also apply to
causal grounding (though it may also apply to grounding in general). But this has the
effect of making God causally fundamental with respect to the GSRPs and making
God’s fundamental predicates his causally fundamental predicates. The reason,
according to this argument, why the fundamental predicates instantiated by GSRPs
are not fundamental to them (the instantiation by GSRPS of those predicates is not
what makes them possible) is that they are not causally fundamental to them; the
existence of those substances, and their instantiation of those properties, is causally
grounded in God and his instantiation of those fundamental predicates. This has the
effect of treating God’s fundamental predicates as his causally fundamental predi-
cates, the predicates without which he would not be able to cause the instantiation of
these predicates (Fundamentality), and thus be unable to causally ground the possi-
bility of the interaction of GSRPs that instantiate them (Modal corollary). So this
argument works by implicitly treating God’s modally fundamental predicates as his
causally fundamental predicates and thus treating God as the causal ground of all
really possible predicates.
If God’s modally fundamental predicates are his causally fundamental predicates,
then derivative predicates (predicates that are possible in virtue of being ‘conse-
quences’ of God’s fundamental predicates) are causally derivative of God. But this
leaves us with two options: either the (really possible) derivative predicates are
actually caused to be instantiated by God or they are possible in virtue of being
such that God has the power to cause them to be instantiated. If we take the former

47
The argument of the previous three paragraphs can be seen as a supplement to the argument of Yong
(2014).
48
In ND Kant attributes their common ground to the “schema of the divine understanding” which he
describes as the “origin of existence” and as “preservation” (Ak. 1: 414). I take this to mean that the
common ground of their existence is the divine intellectual apprehension that causes them to exist and
causes them to continue to exist (preservation).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

route, then there will be no non-instantiated but possibly instantiated predicates. But
Kant himself claims that a derivative predicate like <extended> is still possible even if
it is not instantiated; in this context this means that <extended> is possible even if
God does not actually cause any instantiations of it. So the grounding of derivative
predicates in God through actual causation is incompatible with Kant’s own views
about the possibility of non-instantiated predicates.
The alternative is to hold that derivative predicates are grounded, not in God’s
actually causing them to be instantiated, but in God’s power to cause them to be
instantiated and that God has the power to cause instantiations of predicates that are
not actually instantiated. The conclusion of this line of reasoning is that if the
argument in section 5 is indeed Kant’s best possible reply to modal pluralism and
it implicitly treats God’s fundamental predicates as causally fundamental then, given
that Kant does not want to be committed to the view that (necessarily) there are no
contingently uninstantiated derivative predicates, we must implicitly think of God’s
fundamental predicates as his powers and think of derivative predicates as predicates
God has the power to cause to be actualized.
But this, of course, is precisely how, according to Kant, we should not represent
God as grounding all possible predicates. As I discussed in the previous chapter, Kant
explicitly says that the essence of God, in virtue of which he grounds all really
possible predicates, itself makes possible his power [Allmacht]; I take this to mean
that God’s power is itself a derivative predicate, something that is grounded in
whatever fundamental predicates of God ground all possibilities. Intuitively, God
has the power to actualize various possibilities because they are possibilities; the fact
that God has the power to cause predicate D to be instantiated is itself partly
grounded in the fact that D is possibly instantiated, not constitutive of that fact.49
Elsewhere in Beweisgrund Kant makes clear that we cannot conceive of God ground-
ing possibility through his powers:
Though one can form, through an analogy with human action, some concept of how a being
can be the cause of everything actual, one can form no concept of how such a being could
contain the ground of the inner possibility of other things. It appears that this thought rises far
higher than a created being can reach. (Ak. 2: 152–3)

Kant’s point is that we can coherently represent God as the ground of the existence of
everything actual through an analogy with our will (God’s will stands to the actual
world in a relation analogous to that between our wills and the products of our
actions), but he denies that this analogy can be extended to represent God as the
ground of the possibility of things. The modal analogy Kant has in mind, I take it, is
that, just as various actions are possible for us because we have the power to perform
them, so various things are possible because God has the power to create or actualize

49
As Newlands (2013) and Chignell (2012), quite rightly, objected to the ‘powers’ interpretation in
Stang (2010).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

them. But this is precisely the model of God’s grounding of possibility that, I have
argued, the reply to modal pluralism in section 5 relies on: conceiving of God’s
fundamental predicates as causally fundamental powers. Kant does not explicitly
state why we cannot, using an analogy with human action, form a concept of how
God contains the inner ground of the possibility of all things, but his reasoning is not
difficult to reconstruct: plausibly, it is just the idea, discussed above, that God’s own
power to create something is itself partly explained by the fact that this thing is possible
and hence cannot be part of the explanation of that fact (on pain of circularity).
This is the basic tension in Kant’s pre-Critical modal metaphysics: in order to
understand his modal argument we need, implicitly, to think of God as grounding
real possibility through his causal power, but we have independent reason to think
this cannot be the case. Some readers might wonder whether the requirement to
think of God grounding possibility through his powers arises only on my idiosyn-
cratic reconstruction of Kant’s argument against modal pluralism, that is, whether
this is a tension of my own construction. However, recall that Kant maintains in
Beweisgrund that God is the real ground of the real element in real possibility, but
fails to specify what kind of real grounding is involved here (causal or otherwise). Nor
do Kant’s other writings (published or unpublished) contain an answer to this
question. What is more, it is not clear what category of real grounding other than
causal grounding is available to Kant, for he does not develop a theory of real non-
causal grounds in any of these writings.50 It is no wonder, then, that in reconstructing
Kant’s arguments we need to rely implicitly on the notion of causal grounding. Kant
says that God is the real ground of possibility, but explicitly forbids us from
representing God’s relation to real possibility using the only species of real grounding
about which he has anything substantive to say. But if we take seriously Kant’s claim
that we “can form no concept of how [God] could contain the ground of the inner
possibility of other things,” then we are left with the conclusion that, through implicitly
assimilating it to causal grounding, we obtain no adequate way of representing God’s
relation to all possibility. Beweisgrund, therefore, raises a question about our capacity to
represent possibilities that it does not answer: how is it possible for us to represent
the relation between the space of real possibility and its unique ground in such a way

50
Kant’s examples of real grounds in NG (Ak. 2: 190, 202) and MH (Ak. 28: 12, 24–6, 37, 43, 49, 103,
844–5, 888) are all examples of causal grounds (primarily, forces—at Ak. 28: 102 he even appears to equate
these concepts). Nor is the other main source for Kant’s notion of real ground, Crusius, particularly helpful
on this score. In Ent. }36 Crusius divides real grounds into efficacious causes and “inefficacious causes or
existential grounds.” Clearly, the former category will be no help in understanding non-causal real
grounds; his main examples of existential grounds are a wedge that has the power to overcome a resistance,
whether or not it ever exercises this power, and “the three sides of a triangle and their relations to each
other [which] constitute a real ground of the size of the angle” (Ent. }36, translation from Watkins 2005,
84–5; see also Ent. }79). The wedge example is clearly useless in understanding non-causal real grounding,
and, while I think the mathematical example is a clear precursor to Kant’s notion of a non-causal real
ground of possibility (ratio essendi) in the Critical period, it is not particularly helpful to Kant in the 1760s,
for he has not yet developed his Critical theory of space as the form of intuition that grounds the possibility
of mathematical objects. See Ch. 7 for more discussion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

that would allow us to formulate and understand Kant’s own argument for the existence
of such a ground? In the second half of this study I will argue that Kant does not fully
answer this question until 1790, in the Critique of Judgment.

5.7. Kant’s Pre-Critical Modal Metaphysics

As I pointed out in section 1.3, Kant sometimes uses ‘ontological proof ’ in a wider
sense to refer to any ontotheological argument for the existence of God, one that
attempts to prove his existence using modal concepts alone. In this broader sense,
Kant’s modal argument in Beweisgrund is an ontological argument because it is an
argument in ontotheology.51 But the Beweisgrund argument is an argument from a
consequence—that something is possible—to an antecedently determining ground of
that consequence, a real ground that explains why it obtains. Kant’s argument
provides a posterioriG knowledge of God’s existence: it demonstrates his existence
on the basis of a consequence of that existence, the fact that there are really possible
predicates. This is because Kant’s view is that there is no such a prioriG (antecedently
determining) ground of God’s existence. As he says in the Herder metaphysics
lectures: “that which absolutely necessarily exists does not exist because of a cause
[Ursache], but rather because its non-existence cannot be thought. But this is merely
a ground of knowledge [Erkenntnisgrund] and not an antecedently determining
ground [ratio antecedenter determinans]. In short: it exists” (MH, Ak. 28: 55).
Nor is there an antecedently determining ground of God’s possibility:
Now the necessary being is the sufficient real ground of everything else which is possible, apart
from itself. Consequently, the necessary being will possess that predicate, through which
everything else, apart from itself, can become actual, in agreement with these relations
[order, beauty, perfection]. (OPG, Ak. 2: 88)
Its own possibility is originally given in its existence. It is of other possibilities that the
necessary being contains the real ground. It follows from this in accordance with the law of
contradiction that it cannot be the real ground of the possibility of the most real being itself [ . . . ]
(OPG, Ak. 2: 86)52

In the first passage, Kant draws attention to the fact that God is not the ground of his
own possibility; God is “the sufficient real ground of everything else which is possible,
apart from itself.” In the second passage Kant writes that “its own possibility is
originally given in its existence.” I take this to mean: God exists, therefore God is
possible. However, if God’s existence is not an antecedently determining ground of

51
See OPG (Ak. 2: 155), MV (Ak. 28: 454), and ML2 (Ak. 28: 599).
52
The next paragraph begins “accordingly, the possibility of all other things, in respect of that which is
real in them, rests on the necessary being” (Ak. 2: 87) Cf. Ak. 2: 88 and ND, Ak. 1: 396, where Kant writes,
“of all beings, God is the only one in which existence is prior to, or, if you prefer, identical with possibility.”
Later on the same page he draws the conclusion that God is exempt from the law that everything has a
determining ground.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 KANT ’ S MODAL ARGUMENT

his possibility, in what sense is God’s possibility ‘given’ in his existence? The relation
between God’s existence and his possibility, for Kant, is epistemic and logical: God’s
existence entails that he is possible and is thus is a ground of knowing that he is
possible (a ‘consequently determining’ ground).53 There is no ground that explains
why God is possible, because such a ground would have to be either God himself or a
being distinct from God. No being distinct from God can ground the possibility of
God, since God is himself the ground of the possibility of all other beings. Nor can
God be the ground of his own possibility, since Kant is adamant that nothing can
ground itself.54 Therefore, just as there is no a priori ground of God’s existence, there
is no a priori ground of his possibility.55
The logicist view was that God exists necessarily in virtue of his essence containing
existence. Kant rejects this as a ground of his necessary existence and offers instead a
new concept, that of absolutely necessary existence, and a real definition of it: God
exists absolutely necessarily (= his non-existence cancels all possibility) in virtue of
the fact that he is the real ground of all possibility. On this view, God has a
hyperintensional counterfactual relation to possibility in virtue of the fact that he
has a grounding relation to possibility. It is because God instantiates the fundamental
predicates that, were he not to exist, nothing would be really possible. There is no
ground of God’s existence, but there is a ground of God’s possession of the funda-
mental predicates; it follows from God’s essence that he possesses them (although
that is not how Kant derives them in his argument; God’s essence appears not to be
an epistemic ground for us of his fundamental predicates). Given that God exists and
has these predicates, this grounds his absolute necessity.
So far, we have explored in some detail Kant’s rejection of crucial aspects of the
logicist account of modality. These include:
(Anti-Logicism 1) The idea of a being whose existence is logically grounded in its
essence is incoherent. Consequently, no existential proposition is logically neces-
sary. If there are necessarily existing beings, they do not exist necessarily in virtue
of a logical relationship between their essence and existence. This is because
existence is not a determination; the idea of merely possible but non-existent
objects is incoherent.

53
Cf. Kant’s claim that God’s possibility “is in him as a predicate” (Ak. 2: 84).
54
Cf. ND, Ak. 1: 394 and MH, Ak. 28: 13–14.
55
There is at least one passage where Kant suggests that we can have a priori knowledge of God’s
existence: “the argument [Beweisgrund] that we are giving is based merely on the fact that something is
possible. Consequently, it is a proof which can be conducted entirely a priori” (Ak. 2: 91). However, I think
Kant’s point here is simply the point he makes in the passage quoted above, where he writes: “all grounds of
proof [Beweisgrund] for the existence of God must derive either from the concepts of the understanding of
the merely possible or from the empirical concept of the existent” (Ak. 2: 155–6). When Kant claims that his
argument for the existence of God is a priori he means that it “derives from concepts of the understanding.”
I take this to mean that his argument for the existence of a first real ground of possibility is a priori. It does
not depend upon experience of the determinate predicates of existing things, like their harmonious causal
powers, as his alternate argument in Division Two of Beweisgrund does.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

KANT ’ S MODAL ARGUMENT 

(Anti-Logicism 2) There is real, non-logical incompatibility between predicates.

Consequently, some states of affairs are logically consistent but not possible.
But in this and the previous chapter we have also seen Kant’s positive modal
metaphysics in the pre-Critical period. We can represent this modal metaphysics,
first, as the conjunction of three very general claims about real possibility:
(Non-Logical ) Real possibility is distinct from logical possibility. That some prop-
osition p is logically consistent (¬p does not entail a contradiction) does not entail
that p is really possible. So logical possibility and real possibility are at least
conceptually distinct; it is not a conceptual truth that ◇Lp iff ◇Rp. In fact, some
logical possibilities are not really possible; for some propositions p, ◇Lp & ¬◇Rp.
Equivalently, for some p, ¬□Lp & □Rp.
(Ground ) Real possibilities have real grounds in actuality. If it is really possible
that p, then there is actually an existing object that grounds the fact that it is really
possible that p. The relation between grounds of real possibility and the real possi-
bilities they ground is not a logical relation; if x grounds the real possibility that p, this
is not because of the conceptual containment of p within the concept of x.
(Worldly) Real possibility is a form of metaphysical, or world-based, possibility.
Real possibilities in general are grounded in facts about the world that do not
depend upon how our minds are constituted or how we experience or conceptu-
alize the world.
(GARP) There is a unique real ground of all real possibility (God). Because it is
the real ground of all real possibility, this being exists absolutely necessarily.56
I will argue that, in the Critical period, Kant retains Non-Logical and Ground as
requirements on real possibility in general and retains Worldliness as a requirement
on some, but not all, kinds of real possibility. Furthermore, I will argue in Chapter 9
that Kant retains GARP not as an article of knowledge, but as a belief grounded in the
nature of our faculty of theoretical reason. I have also argued that Kant is committed
to, but does not explicitly endorse:
(Tension) Given our cognitive constitution, we cannot understand how the
GARP (God) grounds all real possibilities. However, given those cognitive limita-
tions, we must think of God as grounding possibility through his powers.
In Chapter 10 I will argue that within the Critical philosophy Kant not only remains
committed to this, but embeds it within a sophisticated theory of the discursive
nature of our intellect, as opposed to the intuitive nature of God’s, and why it obliges
us to think of God as grounding possibility in this way.

56
I have formulated these as principles about the possibility of propositions to facilitate comparison
with Kant’s Critical views, for, I will argue, Kant gives up his pre-Critical fixation on the possibility,
specifically, of predicates to consider the predicative structure of possibility: the possibility of an object
instantiating a concept.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

PART II
Kant’s Critical Modal
Metaphysics
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

6
Real Possibility and
the Critical Turn

6.1. Introduction
As we have seen in detail in the previous chapters, Kant rejects the logicist meta-
physics of modality. By doing so, he must also reject the logicist account of our
knowledge of modality. If not all logically consistent concepts are really possibly
instantiated, by analyzing our concepts and ascertaining their logically consistency
we do not thereby come to know that they are really possibly instantiated. What then
is the source of our knowledge of real possibility? This problem in modal epistem-
ology poses a significant challenge to metaphysics, as Kant and his rationalist
predecessors had practiced it. Ontology (metaphysica generalis) was conceived by
Wolff and Baumgarten and, in some texts, by Leibniz himself, as the science of all
possible beings qua possible beings. Consequently, a challenge to our modal epis-
temology, for both Kant and his predecessors, is also a challenge to the very idea that
metaphysics is a science [Wissenschaft], a body of knowledge [Wissen], rather than
merely rationally grounded conjectures or hypotheses.
This chapter traces Kant’s struggle with this and related problems, from the 1760s
through the publication of the Critique of Pure Reason in 1781 (and the B edition in
1787). I begin, in section 2, with Kant’s modal epistemology in the ‘Prize Essay’ of
1764, and argue that the epistemology and methodology for metaphysics Kant
defends in that essay cannot be squared with the positive metaphysical theory he
defends in Beweisgrund. In section 3 I examine Kant’s 1770 Inaugural Dissertation in
light of his continuing engagement with problems about modality. While the dis-
tinction between sensible and intellectual cognition Kant makes in that work will
permanently transform how he conceives of these problems, the Inaugural Disserta-
tion does not contain a satisfactory answer to the modal problems Kant had been
engaged with since at least the mid-1760s. In section 4 I argue that the problem of the
‘relation’ of a priori concepts to objects that Kant described to Marcus Herz in his
famous letter of 1772 is in fact a question about the relation of a priori concepts (the
categories) to really possible objects. That problem (what grounds the ‘relation’ of a
representation to a really possible object?) becomes the problem of how the categor-
ies, the fundamental concepts of metaphysics, can be used to ‘cognize’ [erkennen]
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

sensibly given objects in the CPR. In section 5 I argue that, because <possible> and
<exists> are not real predicates of objects, these modal questions should be formu-
lated at the level of concepts, e.g., how do we know a priori that our concepts are
really possibly instantiated? In section 6 I return to where this study began in the
Introduction: Kant’s remarks about the centrality of modal concepts to ‘transcen-
dental philosophy’ in the CPR. I argue that questions about our capacity to represent
real possibilities we can prove to be really possible are at the center of Kantian
transcendental philosophy. In section 7 I reconstruct Kant’s positive account of
how a priori knowledge of really possible phenomena (objects that can be given in
sensible intuition) is possible. In section 8 I reconstruct his negative argument that we
cannot know what is really possible for ‘noumena,’ objects that cannot be given in
sensible intuition. It follows that we cannot know all (really) possible beings qua
possible (the objects of general metaphysics or ontology), and, in particular, that we
cannot know the real possibility of any of the objects of ‘special’ metaphysics: the
soul, the cosmos, or God.1

6.2. Modal Epistemology in the Prize Essay

In the ‘Prize Essay,’2 written at the same time as Negative Magnitudes and Beweis-
grund but published in 1764, Kant offers a methodology for metaphysics. However,
that methodology fails to explain how, given his anti-logicist modal metaphysics, we
can come to know modal facts.
Kant distinguishes the method of mathematics from the method of philosophy in
general, and metaphysics more specifically, which he describes as “nothing other
than the philosophy of the fundamental principles of our cognition” (Prize, Ak. 2:
283). Mathematics proceeds synthetically; it starts with a small stock of undefined
concepts and undemonstrated principles (axioms), defines new concepts by arbitrar-
ily combining (synthesizing) the undefined concepts, and proves theorems involving
those concepts by using their definitions and the axioms. In philosophy, we analyze
concepts given by the nature of our intellect (e.g., <substance>, <cause>, <space>),
rendering them more distinct, that is, making apparent their logical structure.3
Consequently, definition plays a different role in the two disciplines. Mathematics
begins by defining its concepts, while a definition of a philosophical concept, if
attainable at all, is attained at the end of a process of philosophical analysis.4

1
My argument in this chapter, that Kant’s doctrine of noumenal ignorance is motivated by concerns in
modal epistemology, agrees in broad outlines with Chignell (2010), (2011), and (2014b). However, there
are many differences of detail in our interpretations, which I do not have space here to discuss fully.
2
This essay is Kant’s submission for the Berlin Academy’s 1763 essay competition. Kant received
second place; first place went to Moses Mendelssohn’s essay “On Evidence in Metaphysical Sciences”
(Mendelssohn 1997, 251–306). See the editor’s introduction to Kant (1992a), lxii–lxiv for a detailed
account of the Prize competition.
3 4
Prize, Ak. 2: 276. Prize, Ak. 2: 284.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

We cannot begin our philosophical inquiry by defining our concepts because

defining them is precisely the task of philosophy. But neither can we expect that the
outcome of philosophy will be a definition of all of our concepts; we will eventually
come upon concepts we do not know how to analyze any further. Having reached
the limit of analysis, we temporarily suspend the project of definition and consider
indemonstrable propositions involving these unanalyzable concepts.5 These pro-
positions will either be indemonstrable merely by us, because we cannot further
analyze their constituent concepts, or because there is no logical analysis of their
constituent concepts that would show them to be true. In the context of Kant’s
rejection of logicism, it is important to note that he allows that there are indem-
onstrable necessary truths: necessarily true propositions which are not made true
by the internal logical structure of their concepts. These are the precursors of
synthetic a priori judgments in the CPR, but Kant has not yet developed his
sophisticated Critical theory of how such judgments are possible. In the Prize
Essay he claims only that we “immediately perceive” them.6
Kant’s proposed methodology for metaphysics in the Prize Essay fits some of the
remarks he makes about his methodology in Beweisgrund. In that work, Kant
eschewed the task of presenting a demonstration of God’s existence in strict logical
form in favor of presenting the matter of such a demonstration, the premises that
entail the conclusion (a Beweisgrund or ‘proof-ground’).7 He spares himself the task
of defining various concepts in Beweisgrund, claiming that, even without a complete
analysis of them (which may be impossible for us), certain fundamental propositions
involving them are immediately certain to us, albeit indemonstrable; in the termin-
ology of the Prize Essay they are the indemonstrable fundamental judgments of
reason. One such instance is the principle, discussed in Chapter 4, that understand-
ing and will are compatible; Kant gives no argument for this, but claims merely that
“an immediate judgment of the understanding forces one to admit the truth of this
contention, even though it cannot properly speaking be given the distinction
required of a logically complete proof ” (OPG, Ak. 2: 87). His analysis of existence
in Division One also conforms to his epistemology of conceptual analysis in the Prize
Essay.8 He analyzes the concept <existence> into the slightly simpler concept of
absolute positing, not because he thinks such analyses are in general required in
philosophy, but because “the present case is one in which [omitting such an analysis]
could occasion confusion and lead to serious errors” (Ak. 2: 70).9
Nonetheless, Kant’s Prize Essay epistemology has severe limits when the anti-
logicist metaphysics of Beweisgrund is taken into account. Kant credits us with an

5 6 7
Prize, Ak. 2: 280. Prize, Ak. 2: 280. OPG, Ak. 2: 65–7.
8
See esp. I.1.1, which reproduces the methodology of the Prize Essay exactly.
9
Cf. OPG, Ak. 2: 70. However, given Kant’s theory of absolute positing as the positing of an object for a
concept rather than the positing of a conceptual mark, it is not clear whether the idea of a further
conceptual analysis of absolute positing even makes sense.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

ability to immediately know real non-logical modal relations between predicates, for
instance, that <extension> and <thinking> are really incompatible.10 But Kant does
not explain how we have access to these non-logical real modal relations among
predicates. As we saw in Chapters 4 and 5, Kant thinks that real possibilities have real
grounds. A real ground is one whose effects, or range of possible effects, is not
contained in or logically derivable from its concept. We cannot come to know the real
grounds of possibility, or the way in which they ground real possibilities, merely by
analyzing our concepts, or performing any other broadly logical operation. In
Negative Magnitudes Kant’s answer to how we know non-logical grounding relations
is more of a restatement of the question than anything: “all of our cognitions of [the
relation of a real ground to its consequence] reduce to simple, unanalyzable concepts
of real grounds, the relation of which to their consequences cannot be rendered
distinct at all” (Ak. 2: 204). We simply do know that some grounds posit their
consequences, even though they do not logically entail their consequences, and a
fortiori we do not know this positing relation through logical analysis of the
ground.11

6.3. Modal Epistemology in the Inaugural Dissertation

For our purposes, the most important text Kant published between the Prize Essay
(1764) and the ‘silent’ decade that preceded the publication of the CPR in 1781 is De
mundi sensibilis atque intelligibilis forma et principiis, the Latin dissertation he
composed on the occasion of his appointment as Professor Ordinarius of Logic
and Metaphysics at the University of Königsberg in 1770.12 The ‘Inaugural Disser-
tation’ (ID), as it is commonly known, is a crucial text in Kant’s development, for it is
the first publication which contains the distinction between concepts and intuitions,
the doctrine that space and time are a priori forms of intuition, and the doctrine that
they present objects as they appear to our sensible faculty, not as they are in
themselves. It thus anticipates many of the distinctively Critical themes that will
occupy us in the rest of this study. However, it makes few advances on the problems
of Beweisgrund itself: the nature of real possibility and how it is given to our minds.
The concept/intuition distinction, and the doctrine that space and time are pure
forms of intuition, will be discussed in further detail in sections 5–7, but for now
I want to discuss them briefly in order to note a modal theme that is sounded in the

10
OPG, Ak. 2: 85 f.
11
Strictly speaking, Kant’s theory of real grounds is compatible with the concept of a real ground G
containing the concept of its effect E, although this conceptual containment relation does not explain the
grounding relation between G and E. Similarly, it might be contained in the concept of (the possibility of ) E
that it is grounded by G, but this conceptual relation would not explain why G grounds E, assuming that G
is a real ground of (the possibility of ) E. See Chapter 3.2.
12
A more complete study would also include discussion of Kant’s 1766 work, Dreams of a Spirit-seer
Elucidated Through Dreams of Metaphysics; for reasons of space, though, I forgo discussion of Dreams.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

ID and will resurface later in the CPR. Kant distinguishes concepts, which represent
indefinite pluralities of objects in virtue of their sharing common marks, from
intuitions, which immediately present particular objects to our minds.13 He points
out that there are concepts whose instances cannot be intuitively presented to us, but
that we cannot infer from this that these concepts are meaningless or that they are
not possibly instantiated. For instance, we have the concept of an infinite compos-
ite whole but we cannot be intuitively presented with an instance; because the
successive intuition of parts of a whole takes place in time, we can only ever intuit a
finite series of the parts of the whole.14 We have to distinguish, for any given
concept, two kinds of modal questions: ‘is the object of this concept possible?’ and
‘can this concept be given an object in sensible intuition?’ This distinction in modal
orders—which we might think of as a distinction between ‘possibility in its own
nature’ and ‘presentability-to-the-mind’—is made especially clearly in the ID, and
will become an important theme in the CPR.
Consonant with Beweisgrund, Kant further distinguishes the first question—
whether a given concept is possibly instantiated—into two distinct questions: is it
logically possible that it is instantiated? And, is it possibly instantiated? Although
Kant does not explicitly introduce the language of ‘real’ possibility or the ‘real’
element of possibility in the ID, it is clear this is what he has in mind when he argues
that we cannot infer possibility from lack of internal contradiction (logical possibil-
ity).15 He makes an associated distinction between the logical use of the understand-
ing, which orders concepts in logical relations according to the principle of
contradiction,16 and the real use of the understanding, of which he writes: “the
fundamental concepts of things and of [their] relations, and the axioms themselves,
are given in a fundamental fashion by the [real use of] pure understanding itself ”
(Ak. 2: 411). Although Kant does not make this connection explicitly, I think it is
clear that the real use of the understanding provides us cognitive access to real
possibility; if what is really possible for things is determined by their fundamental
concepts (essences) and principles (their grounds of possibility) then in cognizing
those concepts and principles we cognize what is really possible for them.
However, Kant does little more than drop a few suggestive hints about what the
real use of the understanding is and how it is possible. These hints foreshadow
important Critical doctrines that will concern us in later chapters, so I want to
mention them briefly. Throughout the ID Kant contrasts our receptive form of
intuition—objects are given to us in intuition by causally affecting our sensible
faculty—with the purely spontaneous manner in which God intuits objects.17 Kant

13
ID, Ak. 2: 397, 399, 402, and 405.
14
ID, Ak. 2: 388–9. It is clear from Kant’s discussion that the infinite whole in question is an infinite
compositum, an infinite whole that depends upon its parts; an infinite totum, a whole that is prior to its
parts, is given to us in our a priori intuition of space. See A438/B466.
15 16 17
ID, Ak. 2: 416. ID, Ak. 2: 411. E.g., ID, Ak. 2: 389 n., 396.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

notes two main differences between our intuition and God’s. First of all, because God
is a completely independent being, his intuition of the world does not depend upon
anything external to him; God spontaneously creates and intuits the world and, in
some passages, Kant seems to identify God’s creation of the world with his intuition
of it. Secondly, God immediately intuits the whole world as a whole, unlike us, who
must successively intuit parts of the world and synthesize them into (necessarily
limited) intuitions of (successively larger parts of) the world. The theistic cognitive-
semantic problem posed by Beweisgrund (how we represent God’s grounding of
possibility) is not addressed here, but Kant does make an interesting new sugges-
tion.18 He concludes the fourth section by writing that “[Malebranche’s view],
namely, that we intuit all things in God, is very close indeed to the one which is
expounded here” (Ak. 2: 410), but does not further explain in what respects Male-
branche’s view is similar to his own. The Malebranchean view that we perceive
objects through ideas in God’s mind, translated into this context, means that we
cognize real possibilities through the real understanding by participating, in some
limited fashion, in divine intellectual intuition. This is significant, because Kant’s
Critical account of our cognition of real possibility will rest on rejecting such a
Malebranchean view; I return to it below in section 8.

6.4. Relation to an Object

Sometime after 1770 Kant discovered a problem in the foundations of metaphysics,
which he first poses to Marcus Herz in his famous letter of 1772:
I noticed that I still lacked something essential, something which, in my long metaphysical
studies, I, as well as others, had failed to consider, and which in fact constitutes the key to the
whole secret of metaphysics, hitherto hidden even from itself. I asked myself this question:
what is the ground of the relation [Beziehung]19 of that in us which we call ‘representation’ to
the object? (Corr., Ak. 10: 129)

Kant here raises a problem about ‘the relation’ of representations to objects and goes
on to consider two possible models of this relation: either the object is the ground of
the representation, or the representation is the ground of the object. He concludes
that in the case of the ‘intellectual’ concepts involved in the real use of the under-
standing, concepts given by the nature of our mind itself rather than by abstraction
from experienced instances, no explanation has yet been given of their relation to
objects.

18
ID, Ak. 2: 126 n.
19
Translation (with minor modifications) from Kant (1999), 133. Ellington (Kant (1977), 117) trans-
lates Beziehung as ‘reference.’ However, I think this mistakenly prejudges which Beziehung Kant has in
mind, because ‘reference’ has a more specific philosophical meaning in English than Beziehung does in
German.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

This passage, and the dichotomy Kant poses (either the object grounds the
representation, or vice versa), parallel the opening of the “Transition to the tran-
scendental deduction of the categories” in the CPR:
There are only two possible cases in which synthetic representation and its object can come
together, relate to one another necessarily, and, as it were, meet each other: either if the object
alone makes the representation possible, or if the representation alone makes the object
possible. (A92/B124 f.).20

Kant goes on to apply this dichotomy to a priori concepts (concepts given by the
nature of the mind itself) and their relation to objects. The problem of the relation of
a priori concepts to their objects is what Kant will call, in the CPR, the problem of the
‘objective validity’ of the categories. The Transcendental Deduction will argue that
categories have ‘relation’ to (objective validity for) sensibly given objects and, what is
more, that they have ‘relation’ only to (objective validity for) sensibly given objects.
This close textual parallel has led many commentators to read the problem of the
‘objective validity’ of the categories in the CPR as the successor of the problem of the
‘relation’ of intellectual representations to objects in the Herz letter.21 On this
reading, the problem that Kant describes to Herz in the 1772 letter is the problem
that motivates the Transcendental Deduction. In the remainder of this chapter I am
going to follow this interpretive tradition and argue that the original problem in the
Herz letter and the problem of the objective validity of the categories is a modal
problem: how is it possible for us to know a priori that it is really possible for objects
to instantiate the categories?
In both texts Kant raises a question about representation in general, but, at least
since ID, he has distinguished two different kinds of representation: concepts and
intuitions. He regards the ‘relation’ of a priori intuitions to their objects as less
problematic,22 so I will follow Kant in focusing on the more problematic case of
concepts and their relation to their objects. It is quite unclear, however, both in the
Herz letter and in A92/B124, what the ‘relation’ in question is.23 One natural thought
would be that the relation in question, of representation to object, is simply the
relation x represents y. There are good reasons to reject this natural thought, though.
First, in the CPR Kant will never deny that categories represent non-sensible objects

20
Cf. the parallel passage in MM, Ak. 29: 796.
21
E.g., Cassirer (1981), 126; Kemp Smith (1984), 186–7; Guyer (1987), 25; Longuenesse (1998), 17; and
Förster (2012), 2.
22
Since he had already explained this relation, in ID, through the Critical doctrine that space and time
are forms of intuition. See section 3.
23
Nor is it uncontroversial what objects are in question here. Carl (1989b) argues that the problem Kant
poses in the letter to Herz is the problem of how intellectual concepts relate to empirical objects; Beck
(1989) argues, on the contrary, that the problem is how intellectual concepts relate to intelligible objects
(noumena). However, Ciovacki (1991) shows that Carl and Beck are both mistaken: in the letter to Herz,
Kant is raising a perfectly general problem about the relation of intellectual concepts to objects überhaupt.
Both restrictions of the problem are misguided.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

in the minimal sense that they can be used to think about such objects and thus have
some (however minimal) intentional content for those objects. But he will deny that
the categories relate to (in the relevant sense) objects that cannot be given in sensible
intuition. Secondly, consider the list of concepts Kant explicitly considers in the Herz
letter: mathematical concepts of magnitude, purely intellectual concepts (like <sub-
stance>), and empirical concepts. One thing all of these concepts have in common,
for the pre-Critical Kant, is that they are concepts of really possible objects. If Kant
were interested in representation in general he would presumably be just as interested
in how logically consistent concepts überhaupt represent their objects, but he is not.
The ‘relation’ in question seems to arise only for representations of really possible
objects. Another natural thought would be that the ‘relation’ in question here is the
relation between an object and a concept it instantiates. However, as I will argue
more fully below, Kant holds that mathematical concepts have ‘relation’ to objects in
virtue of being possibly instantiated by objects given in intuition, whether or not they
are actually instantiated by any objects. So the ‘relation’ in question cannot be actual
instantiation of a concept by an object.
I take this to be sufficient reason to entertain the hypothesis that Kant is con-
cerned, in both passages, with the question of how a representation ‘relates’ to really
possible objects: what is the ground of possibility of our representation of really
possible objects? But this suggestion needs to be refined further. If the ‘relation’
between representation and object at issue in the Herz letter is merely being the
representation of a really possible object then it does not have an epistemic dimension;
a concept, for instance, can be the concept of a really possible object without our
knowing it, or even being in principle capable of knowing it. The relation could
obtain ‘behind our backs,’ so to speak. Kant considers one such model, and rejects it
on the basis that it is circular.24 A few paragraphs later in the Herz letter, Kant
describes Malebranche and Crusius (as well as Plato) as explaining the ‘relation’
between representation and objects by appeal to a harmony implanted by God. If the
relation in question is simply possible instantiation, this means: our concepts are
possibly instantiated because God has created us with concepts of really possible
objects (Crusius) or allowed us to see ideas in his mind of really possible objects
(Malebranche). But Kant’s objection—that it is circular—then becomes unintelli-
gible: Malebranche and Crusius do not offer a circular explanation of why our
concepts are concepts of really possible objects.
Kant’s circularity objection comes into clearer focus, though, if we recognize that
the ‘relation’ in question is not purely modal but has an epistemic dimension as well:
what is the ground of the possibility of our representing really possible objects that we
can know a priori to be really possible? Malebranche and Crusius (according to Kant)
attempt to prove the real possibility of the objects of our concepts by assuming that

24
Corr., Ak. 10: 131.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

some of our other concepts, e.g., <God> and <cause>, are actually instantiated. I will
examine the nature of this circularity objection in detail in section 8 below, but for
now I just want to point out that on this modal-epistemic interpretation of the
‘relation’ of representation to object, Kant’s ‘circularity’ objection becomes at least
intelligible (though not necessarily decisive).
If this is correct, then the question whose discovery motivated the ‘Critical’ turn
during the ‘silent decade’ is implicitly a modal question and, what is more, the modal-
epistemic question that I argued is elicited by, but not explicitly posed in, Beweis-
grund.25 This also clarifies Kant’s claim in the Herz letter that “I, and others, in my
long metaphysical labors had neglected” the question of the relation of representa-
tions to objects. It is implausible that Kant would accuse his predecessors of neglect-
ing to consider how representations represent objects, because Kant’s predecessors in
the German rationalist tradition developed extensive theories of representation.26
This remark becomes clearer, however, if we interpret the relation as I have done,
for we can read Kant as casting an eye back on his logicist predecessors and his own
pre-Critical writings and saying: “they assumed that by representing something as
logically possible one thereby represents it as really possible, and although I carefully
distinguished these things, I did not have an account of how we represent something
as really possible or how we can prove of really possible objects that they are really
possible objects.” If this is the ‘relation’ between representation and object that Kant
is here concerned with, he is right that he, as well as many others (though not perhaps
all), neglected it.
One term Kant uses in the CPR for representations of knowably really possible
objects is cognition:
To cognize an object it is required that I be able to prove its possibility (whether by the
testimony of experience from its actuality or a priori through reason). But I can think whatever
I like, as long as I do not contradict myself, i.e. as long as my concept is a possible thought, even
if I cannot give any assurance whether or not there is a corresponding object somewhere within
the sum total of possibilities. But in order to ascribe objective validity to such a concept (real
possibility, for the first sort was merely logical) something more is required. (Bxxvi n.)27

A concept is a cognition, according to this characterization, only if I can prove that it

is really possible for it to have an object (i.e. to be instantiated). Kant characterizes
cognition in different terms in other contexts (e.g., as thinking an intuited object

25
This is one possible route from ID to the CPR during the ‘silent decade.’ In this chapter I cannot hope
to do justice either to all of the other factors motivating Kant’s thought in this period (e.g., his rereading of
Hume, or the discovery of the Antinomies), nor can I engage in detail with the vast scholarly literature on
the subject. For other narratives of Kant’s development in this period see de Vleeschauwer (1962), Cassirer
(1981), Werkmeister (1981), Carl (1989a), Kreimendahl (1990), Guyer (1987), and Kuehn (2001).
26
Most notably, Leibniz’s theory of representation in terms of expression; see L 339.
27
Kant goes on to claim that there can be practical as well as theoretical sources of cognition, but in this
book I am focusing almost exclusively on theoretical cognition. See Chignell (2010) and Schafer (unpub-
lished) for more on how practical reason can fill the ‘real possibility’ requirement on cognition.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

under a concept), but for now I want to focus on this modal concept of cognition (m-
cognition, for short). The problem of the Herz letter can now be stated in Critical
terms as follows: how is it possible for the categories, concepts given by the nature of
the mind rather than abstracted from experienced instances, to be a priori cogni-
tions? This problem will be our guiding thread for the rest of this chapter.

6.5. Intuition, Existence, and Possibility

Before continuing, I want to make an observation about the transformative role that
the intuition/concept distinction will play in how Kant so much as formulates
questions about possibility. To do so, I need to explain a little more about what
intuitions are, and what distinguishes them from concepts.
Kant defines intuitions as singular and immediate representations of objects,
whereas concepts are general and mediate representation of objects.28 A concept is
a general representation because it represents all of the objects that share certain
common marks; these objects make up the extension (in the contemporary sense)29
of the concept, while the marks of the concept make up its intension.30 We can use a
concept to pick out a single object by combining the concept with a demonstrative or
indexical expression (e.g., ‘that chair’) but the concept itself remains a general
representation; ‘singular concept,’ for Kant, is a contradiction in terms.31 By contrast,
intuitions are singular in their very content. I interpret this to mean that an intuition
is ‘of ’ a single object. To any intuition we can ask ‘what is the object of that intuition?’
and the answer must be a single object, rather than a plurality of objects (those two
ducks) or a conceptually specified set of objects (all the ducks in the pond).
I interpret the immediacy of intuitions to mean that an intuition does not
represent its object via some mediating representation; the immediate object of an
intuition is an object itself, not a further representation of that object (e.g., a mental
image).32 Concepts, by contrast, represent their objects by means of some mediating
representation, either a concept or another intuition.33 The clearest case of mediating
representations are the conceptual ‘marks’ that compose a concept. For instance,
<body> represents bodies by means of the marks of which it is composed—
<extended>, <impenetrable>, and <lifeless>—which themselves represent objects

28
A19/B33, A50/B74, A320/B77, and A713/B741. On the singularity and immediacy of intuitions see
also WL (Ak. 24: 904–5), MM (Ak. 29: 880), and MK3 (Ak. 29: 971). There is a vast literature on the
Kantian notion of intuition, which I cannot hope to engage with in detail here: Hintikka (1969), Parsons
(1969) and (1992), Thompson (1973), Howell (1973), Falkenstein (1995), Smit (2000), Hanna (2008) and
(2011), Allais (2009) and (2015), McLear (2011), (2015), and (forthcoming), Tolley (2013) and (forth-
coming), and Stephenson (forthcoming).
29
Kant sometimes identifies the extension of a concept with the set of concepts in which it is contained
as a mark, rather than the set of objects that it applies to; see JL (Ak. 9: 95–100), LB (Ak. 24: 240), and LP
(Ak. 24: 569), as well as Longuenesse (1998), 86–7 and Anderson (2004), 512–13 for discussion.
30 31
JL, Ak. 9: 95–6. JL, Ak. 9: 91.
32 33
Cf. Smit (2000), 263 and Allais (2015). A19/B33 and A68/B93.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

(perhaps through the mediation of further marks). Kant does not explain how
intuitions can play the mediating function that conceptual marks do, but the
mediating role of intuitions will not be crucial for my account so I forgo further
discussion of it here.34
The first thing I want to explore is the crucial connection between intuition and
existence as ‘absolute positing.’ The immediacy of intuitions means that intuitions
can never be ‘uninstantiated’ the way that mediate concepts can be; an ‘uninstan-
tiated’ intuition would not be about anything, so it would not be a representation of a
singular object, contrary to its definition. This entails that for every intuition there is
an object of that intuition, which entails that the object of that intuition (a definite
description, due to the singularity of intuition) exists in the ‘absolute positing’
sense.35 Furthermore, intuition is a necessary epistemic basis for making judgments
of ‘absolute positing.’ This is made especially clear in “On the impossibility of an
ontological proof of God’s existence” in the CPR, in which Kant restates the ‘absolute
positing’ theory of existence from Beweisgrund:
Now if I take the subject (God) together with all his predicates (among which omnipotence
belongs), and say God is, or there is a God, then I add no new predicate to the concept of God,
but only posit the subject with all its predicates, and indeed posit the object in relation to my
concept. (A599/B627)

By contrast, when I judge that God is omnipotent I only posit the ‘relation’ between two
concepts: anything that falls under <God> falls under <omnipotent> (because the latter
is a mark of the former). For us, the only way to make an existential judgment with
epistemic warrant is through intuition, which for us is necessarily sensible (receptive):
Thus whatever and however much our concept of an object may contain, we have to go out
beyond it in order to attribute existence to it [diesem die Existenz zu erteilen]. With objects of

34
Atomic concepts, concepts that are not logically composed of further concepts, by definition do not
have marks, hence their representation of their objects is not mediated by marks. Since atomic concepts’
representation of their objects is not mediated by conceptual marks, it stands to reason that it would be
mediated by intuition. On this interpretation, atomic concepts represent objects by being abstracted from
intuitions that present instances of them. For instance, the concept <red>, assuming it is atomic, represents
objects via an intuitive representation of a red object: it represents all objects that resemble the intuited
instance in respect of color, but lacks any marks. While Kant is skeptical of our ability to discover any
atomic concepts (concepts composed of no marks), his account of conceptual representation should not be
inconsistent with their existence. Cf. Smit (2000).
35
The idea that the object of an intuition always exists is liable to provoke the objection that in various
passages Kant seems to allow that the object of an intuition might not exist (e.g., when we hallucinate or
dream); B278, Prol. (Ak.4: 281–2), B151, and Anthr. (Ak. 9: 153, 167–8). But notice that the notion of
existence involved in those passages is not mere absolute positing: it is the positing of an object that exists
in space outside the subject and is causally efficacious. But that notion of existence is a real predicate, for it
“adds to the concept of an object.” I am not claiming every intuition has an object that exists in that sense;
I am making the more minimal claim that for every intuition there is (absolute positing) an object of that
intuition. In hallucination, we mischaracterize the nature of the object; we mistake, for instance, a mere
visual image for an object that exists in space outside us. See Stephenson (forthcoming), McLear
(forthcoming), and Allais (2015) for discussion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

sense this happens through the connection [Zusammenhang] with some perception [Wahr-
nehmung] of mine in accordance with empirical laws; but for objects [Objecte] of pure
thinking there is no means whatever for cognizing their existence, because it would have to
be cognized entirely a priori, but our consciousness of all existence (whether immediately
through perception or through inferences connecting something with perception) belongs
entirely and without exception to the unity of experience, and though an existence outside
this field cannot be declared impossible, it is a presupposition that we cannot justify through
anything. (A601/B629)

Epistemic warrant for an act of absolute positing (i.e. the judgment there is an F, or,
equivalently, an F exists) must come from a consciously apprehended intuition
(perception)36 of the object absolutely posited, or from a perception of a distinct
object, from whose existence we can infer the existence of another object, through
known empirical laws (“connection [Zusammenhang] with some perception of mine
in accordance with empirical laws”). Since, by definition, the former is lacking for
objects that cannot come before our senses (noumena—see section 7), and we cannot
satisfy the latter for such objects because we cannot cognize the concept <cause–
effect> when applied to them (the argument I reconstruct in section 8), we lack
epistemic warrant for acts of absolute positing with respect to objects that cannot be
sensibly presented to us (like God). But we should not, therefore, assume there are no
such objects, or, equivalently, that they do not exist (in the absolute positing sense).
Note that these passages occur in the section of the CPR where Kant raises his
famous objection against the ontological argument (reconstructed in Chapters 1 and
2): “being is obviously not a real predicate” (A598/B626). This is because the
epistemic indispensability of intuition is the downstream consequence of the denial
of ontological proofs: if ontological proofs were possible, it would be possible to
justifiedly make at least one existential judgment (there is a God) without intuiting
any objects. Conceptual analysis of internally logically consistent objects would be
enough for at least some existential judgments.
Having sharply separated existence (the absolute positing of an object) from
possibility, and sharply separated intuitions of objects from concepts, Kant now
has to interpret questions about possibility differently. The question of whether, for
instance, spatiotemporal substances are possible is not a question about objects (of
absolute positing) but a question about the concept <spatiotemporal substance>: is it
possibly instantiated by an object? We cannot intelligibly interpret this question as a
question about objects of absolute positing (the objects there are/the objects that
exist) because that would be tantamount to interpreting it as the question: are the
spatiotemporal substances there are possible or impossible? That question would
either receive a trivial ‘yes’ (every object there is is possible) or it would involve

36
See Tolley (forthcoming) on Kant’s notion of Wahrnehmung.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

entertaining the absolute positing of impossible objects, e.g., there are spatiotemporal
substances, but are they possible?
This means that <possibility> is not a real predicate of objects any more than
<existence> is: “the modal categories have this special feature: they do not augment
the concept to which they are ascribed, as determinations of the object, but rather
express only the relation to the faculty of cognition” (A219/B266).37 It will take the
rest of this book to fully explicate what it means that the modal categories “express
the relation to the faculty of cognition,” but for now I want to focus on the first part of
this sentence. That the modal categories do not “augment the concept to which they
are ascribed, as determinations of the object” means that the modal categories are not
determinations or ‘real predicates,’ or, in the terms of Kant’s definition later in the
CPR, they are not “predicates that go beyond the concept of the subject and enlarge
it” (A598/B626). Recall my interpretation of what a ‘real predicate’ is from Chapter 1:
a real predicate is one such that, possibly there is an object (absolute positing) that
does not instantiate it. Assuming my interpretation there was right, this means: it is
not possible for there to be (absolute positing) an object that does not instantiate the
concept <possible>. This means that questions about possibility are not first-order
questions about the objects there are (objects of absolute positing) but second-order
questions about concepts: is a given concept possibly instantiated, i.e. is it possible for
there to be objects of absolute positing that instantiate it? In other words, the
distinction between the possible and the impossible is not an extensional distinction
between two domains of objects (of absolute positing) but a distinction between
concepts: a distinction between concepts possibly instantiated by objects of absolute
positing, and concepts not possibly instantiated by such objects. Likewise, the
distinction between the existent and the non-existent is not a distinction between
two domains of objects but one between concepts: those that are instantiated and
those that are not.38
Our epistemic access to absolute positing is via sensible intuition, or so Kant will
argue. We must be careful not to assume at the outset that all objects there are are
sensibly intuited or intuitable by us; there may be objects that could be absolutely
posited by a being with an intellect different from ours (e.g., a non-sensible or
‘intuitive’ intellect). And we must not assume at the outset that a concept is possibly
instantiated if and only if it is possibly instantiated by objects of the kind we have
epistemic warrant to posit absolutely: sensible objects. And we should not assume at
the outset that if there are objects of a certain kind then there is an intellect to posit
them absolutely. That there are such objects means that if some intellect were to posit
them absolutely then that act of positing would be correct or true; it does not
(without additional argument) entail that any intellect does posit them or even
could posit them. Consequently, when I refer to objects as ‘objects of absolute

37 38
Cf. A74/B100. Cf. Abaci (2013), 18.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

positing’ I am reminding the reader that ‘object’ here means: appropriate target of
absolute positing (q-object, cf. Chapter 1.3). I am not necessarily claiming that any
intellect engages in an act of absolutely positing these objects (e.g., judging that there
are such objects).39

6.6. Transcendental Philosophy and

the Concept of an Object
In thinking about the continuity, and differences, between Kant’s conception of
modality in the CPR, his pre-Critical works of the 1760s, and the older logicist
metaphysics, a particularly productive place to start is the ‘Table of Nothings,’ a
short section at the end of the Transcendental Analytic, which has not received the
attention I think it deserves from scholars.40 That section begins:
The highest concept with which one is accustomed to begin a transcendental philosophy is
generally the division into the possible and the impossible. Since all division, however,
presupposes a concept that is divided, a yet higher concept must be given, and this is the
concept of an object in general (taken problematically, and unspecified whether it is something
or nothing). (A290/B346)41

Kant’s claim that the highest distinction in transcendental philosophy is “generally”

[gemeiniglich] taken to be that between the possible and the impossible is a reference
back to Wolff and Baumgarten; as we saw in Chapter 1, the highest concept of
ontology (metaphysica generalis) is the concept ens/Ding, possible thing or possible
being, and Wolff and Baumgarten begin by dividing possible beings from impossible
ones (nihil, Nichts).42 But ‘transcendental philosophy’ is Kant’s term for the philo-
sophical discipline whose preparatory critique is being given in the CPR,43 so Kant is
here announcing that the CPR is both a successor to ontology and an improvement in
this respect: it begins with the more general concept <object>, which it then further
divides into <possible> and <impossible>.44
It is important from the outset to realize that these are not extensional distinctions
between different sets of objects; Kant is not claiming that the extension of <object>
includes both possible and impossible objects, just as Wolff and Baumgarten are not

39
There may be forms of absolute positing that are not judgmental; e.g., the intuitive intellect may
absolutely posit its objects without making judgments (because judgment is a form of discursive unity). See
Ch. 10 for more.
40
It merits only one paragraph in Willaschek (1998) and is not mentioned in Guyer (2010);
Longuenesse (1998), 303–5 is a welcome exception.
41
The ‘Table of Nothings’ is mirrored by several similar passages in Kant’s lectures on metaphysics in
the 1780s, corresponding to Baumgarten’s discussion of possibility and impossibility in Meta. }}7–18; see
MV (Ak. 28: 414), ML2 (Ak. 28: 543–4), MD (Ak. 28: 628), MM (Ak. 29: 811–12), and MK3 (Ak. 29: 960–2).
42
Meta. }7, Ont. }57, Dt.Met. }28. 43
A12/B25–6.
44
Cf. Kant’s claim at A246–7/B303 that the “proud name of ontology . . . must give way to the modest
one of a mere analytic of the pure understanding.” Cf. Ak. 29: 752.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

claiming that impossible beings are a kind of being. He is making an intensional

distinction among concepts in virtue of their representational content: those that are
possibly instantiated (concepts of something) and those that are not (concepts of
nothing). This also means that Kantian transcendental philosophy operates at one
level of abstraction higher than Wolffian ontology; rather than concerning itself in
the first instance with objects (beings) it concerns itself with our representations of
objects. Its highest genus is <object of representation>, which Kant then divides into
the concepts <possible object (of a representation)/something> and <impossible object
(of a representation)/nothing>.
Kant then divides the concept of nothing—the concept <impossible object (of a
representation)>—into four sub-concepts, organized according to the categories.
While Kant does not draw this out, each of these ‘concepts of nothing’ is actually
a different distinction between concepts of impossible and possible objects. For
instance, the first ‘concept of nothing’ is the concept of an ens rationis, which Kant
glosses as “the concept of an object to which no intuition corresponds” (A290/
B347). This corresponds to the distinction between concepts for which correspond-
ing objects cannot be given in intuition (ens rationis, nothing), and concepts for
which a corresponding object can be given (something). Kant organizes his ‘table of
nothings’ according to the four moments of the table of categories (quantity,
quality, relation, and modality),45 but for our purposes a different principle of
organization is more relevant: the different distinctions between possible and
impossible objects are logically subordinate to one another. For instance, a concept
of an object that can be given in intuition (something that is not a mere ens rationis)
is by definition logically consistent, so it is not a nihil negativum (the object of a
logically inconsistent concept), but the converse does not hold; some ‘logical’
somethings (objects of logically inconsistent concepts) are merely entia rationis
(e.g., Kant’s example of a fictional force) for which objects in intuition cannot be
given. If we unwind these logical subordination relations among different possible
contents we arrive at the following table of divisions of possible and impossible
objects (summarized in Figure 6.1):
1. Logical something vs. logical nothing (nihil negativum). This is the distinction
between concepts that are logically possibly instantiated and those that are not.
2. Real something vs. real nothing (ens rationis). This is a distinction within the
‘logical something,’ and I discuss it in detail below.
3. Material something vs. formal nothing (ens imaginarium). This is the distinc-
tion, drawn within the real something (objects that are not entia rationis),
between material objects that can be given in space and time through sensory
experience, and the a priori forms of space and time themselves.

45
A290/B346.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

Object

Logical something Logical nothing (nihil negativum)

Real something Real nothing (ens rationis)

Material something Formal nothing (ens imaginarium)

Positive something Privative nothing (nihil privativum)

Fig. 6.1. The table of nothings (organized by logical precedence)

4. Positive something vs. privative nothing (nihil privativum). This is the distinc-
tion, drawn within material objects in space and time (objects that are not entia
imaginaria), between positive determinations (light, heat) and a mere absence
of positive determination (darkness, cold).
I have not attempted to characterize the second pair of distinctions in explicitly
modal terms, because to do so would presuppose the details of Kant’s Critical
epistemology, and because it is not crucial for my purposes here. Instead, I want to
focus on the first two something/nothing distinctions.
In his explanation of the distinction between an ens rationis and what I have called
a ‘real something,’ Kant writes:
thus the object [Gegenstand] of a concept, to which no corresponding intuition can be given
[gar keine anzugebende Anschauung correspondiert], = nothing, i.e. a concept without an
object, like the noumena, which cannot be counted among the possibilities, although also for
that very reason must not be assumed to be impossible (ens rationis); or like certain new forces,
that, although they are thought without contradiction, lack examples in experience, and
therefore must not be counted among the possibilities. (A290/B347)

Two points are important here. First, this is a distinction that concerns real possi-
bility; the mere logical possibility of a concept does not entail that its object is
possible. Secondly, it is not the distinction between what is really possible and what
is not really possible; it is the distinction between what we can know to be really
possible and what we cannot know to be really possible.46 The concept of a nou-
menon, like the invented concept of a force that has never been encountered, is a
concept of ‘nothing’ because we cannot prove that its object is really possible. We
should not, Kant tells us, conclude that such objects are therefore impossible.
This explains why the highest genus of transcendental philosophy is <object (of
representation)>. Kant defines ‘transcendental’ at the beginning of the Transcenden-
tal Logic as follows: “I call all cognition transcendental that is occupied not so much
with objects but rather with our mode of cognition of objects, insofar as this is to be

46
Cf. MK3 (Ak. 29: 961).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

possible a priori” (B25).47 The most general question for transcendental philosophy
is, how can we represent any object whatsoever a priori? To answer that question we
must distinguish representing a logically possible object from representing a logically
impossible one (hence the first division of the Table of Nothings, when it is organized
logically: logical nothing versus logical something), and ask, what is the ground of the
possibility of our representing a priori a logically possible object? The answer to this
first transcendental question is relatively straightforward: by representing something
that obeys the principle of non-contradiction (PNC) we represent a logically possible
object.
The second question of transcendental philosophy, according to my reorganiza-
tion of the Table of Nothings, would be, what is the ground of the possibility of
our representing something really possible that we can know to be really possible
(a real something)? Because the answer to the first transcendental question is
relatively trivial (PNC), this has a claim to being the first substantive question of
transcendental philosophy. This characterization of transcendental philosophy may
seem unfamiliar to many readers, but recall Kant’s characterization of cognition in
explicitly modal terms from Bxxvi n. (quoted in section 4): to cognize an object we
must be able to prove that the object is really possible. But that is equivalent to
representing a real something (in the Table of Nothings). So if this is correct, the
first substantive question of transcendental philosophy is, how is a priori cognition
possible?
Ontology, as conceived by Wolff and Baumgarten, is, the science of all possible
beings insofar as they are possible. Transcendental philosophy, the Kantian successor
to ontology, differs in two crucial respects. First, it is not a first-order science of some
domain of objects, but a second-order inquiry into how representation of objects is
possible. Secondly, it distinguishes between representing a (knowably) really possible
object and representing a (knowably) logically possible object, and inquires specific-
ally into the grounds of possibility of the former. Note that this distinction does
not rest on assuming that logical possibility and real possibility differ in extension,
i.e. that there is some p such that it is logically possible that p but not really possible
that p. For even if logical possibility and real possibility are the same in extension they
differ in intension. The content of the concepts <logically possible> and <possible>
are different; otherwise, the logicist view that they are co-extensive would be unin-
formative. Consequently, representing an object as really possible is distinct from

47
This is a modification from Kant’s original definition of transcendental cognition at A11: “I call all
cognition transcendental that is occupied not so much with objects but rather with our a priori concepts of
objects in general.” However, this modification in the B edition merely brings Kant’s definition of
transcendental cognition into line with his definition at A57/B80. The CPR, Kant remarks in both the
A and the B Introductions (A12/B25–6), does not contain the complete system of transcendental
philosophy because it gives neither a complete inventory, nor an exhaustive analysis, of a priori concepts;
instead it contains a ‘critique’ of our rational faculty, an analysis of the fundamental principles and modes
of operation which make it possible for us to cognize objects a priori, as well as an evaluation of their
legitimate scope and use.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

representing it as logically possible, and proving that it is really possible is different

from proving it to be really possible. By answering the transcendental question about
logical possibility, the logicists have not yet answered the transcendental question
about real possibility.
The question elicited, but not explicitly raised, by Beweisgrund—how are real
possibilities given to the mind?—becomes the principal question of transcendental
philosophy in the CPR: how is it possible for us to represent a priori knowably really
possible objects? This encompasses three interrelated questions, each of which will
receive its answer in the course of the CPR:
(1) What is it to represent an object as really possible? That is, what is the content
of the representation of an object as really possible?
(2) What is it for an object to be really possible? What are the grounds of real
possibility? (At least for objects whose real possibility we can represent (1) and
prove (3).)
(3) Of a really possible object we can represent as really possible, how can we
prove that it is really possible? And, what’s more, prove this a priori?
These are questions, respectively, about the cognitive semantics, the metaphysics,
and the epistemology of modality. In the next section I am going to reconstruct
Kant’s positive Critical theory of cognition of really possible objects (his answer to (1)
and part of his answer to (3)), so before continuing I want to make a brief remark
about the concept of an object.
To relate this back to section 5, the most general concept of object in transcen-
dental philosophy is not ‘object of absolute positing.’ For us, there are two kinds of
representations: concepts and intuitions. Intuitions give us objects of absolute pos-
iting (though we may not be consciously aware of every object we intuit, and hence
we may not engage in an act of absolute positing for each of them). When we say that
a concept C is the concept of an impossible object we are not saying there is an object
(of absolute positing) that instantiates C but that object is impossible.48 Instead we
are saying something about the content of C: because of this content it is impossible
for C to be instantiated by an object (of absolute positing). So we have a notion of
object that is, so to speak, internal to representational content, sometimes called
the ‘intentional’ object of the representation. I will call this notion of object ‘object
of representation’ (r-object, for short) and distinguish it from the notion of an ‘object
of absolute positing’ (q-object). Talk of r-objects is really shorthand for talk of

48
Interpreting the idiom ‘C is a concept of a really impossible object’ in this way is unattractive for at least
three reasons. First, it would populate our theory of what there is (our ‘ontology’ in the Quinean sense) with a
host of absurd beings: the largest prime number, bodies that think, round square, etc. Secondly, by parity of
reasoning, it would entail that ‘<C & ~C> is the concept of a logically impossible object’ entails that there is an
object, that is both C and ~C, which would require that the PNC be restricted to logically possible objects.
Thirdly, there is a more natural interpretation available, given in the body of the text.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

conceptual contents. This means that we can more perspicuously formulate ques-
tions (1)–(3) in terms of concepts, as follows:
(1*) What is it to represent a concept as really possibly instantiated by a q-object?
(2*) What is it for a concept to be really possibly instantiated by a q-object? What
are the grounds of real possibility?
(3*) How can we prove a concept to be really possibly instantiated by a q-object?
And, what’s more, prove this a priori?
Sometimes I will talk about the possibility of the object of a concept; in every case,
this means: the possibility that the concept is instantiated by a q-object.49

6.7. A priori Cognition of Phenomena

In this section I reconstruct Kant’s reasons for thinking that we can have a priori
(modal) cognition of phenomena, objects that can be given to us in sensible intuition.
Contrary to the method of the rest of this book, in this section I am going to be
somewhat naive about what exactly (real) possibility means; my aim is to start with
some stock of clear Kantian principles about possibility (e.g., that logical possibility is
not sufficient for real possibility) and construct, over the course of this section, a clearer
idea of what notion of real possibility is at work in Kant’s positive theory of a priori
cognition (which I call formal possibility). In Chapters 7 and 8 I distinguish formal
possibility from several other notions of real possibility in Kant’s Critical system.
Cognition [Erkenntnis] is a technical term and, as Kant scholars are increasingly
appreciating, does not correspond very well to our concept ‘knowledge.’50 First of all,
knowledge is standardly taken to be propositional, while Kant regards intuitions and
concepts (under the right conditions) as cognitions, but neither of them has a
propositional object.51 Secondly, Kant appears to countenance the possibility of
false cognition,52 but the notion of false knowledge is prima facie incoherent. Thirdly,
Kant’s definition of cognition is very different from his definition of knowledge
[Wissen]. Kant gives two kinds of characterizations of cognition. First, a modal one
we have already seen: a representation is a cognition just in case we can prove the real
possibility of its object. Secondly, a ‘semantic’ one: to cognize an object is to think a
given object under a (empirical or a priori) concept.53 By contrast, Kant defines

49
For more on Kant’s concept of an object, and the distinction between r- and q-objects, see my
unpublished paper “Kant and the Concept of an Object,” available on my website (see Notes on the Text).
50
I have been especially influenced here by Schafer (unpublished) and (forthcoming); the following
remarks on the difference between cognition and knowledge are deeply indebted to him. Related ideas are
developed by Smit (2000) and Hanna (2006).
51
The construction ‘anschauen daß’ does not (to my knowledge) appear anywhere in Kant’s writings.
52
A58/B83.
53
E.g., A50/B74. I call this the ‘semantic’ characterization of cognition because it describes cognition in
terms of the interplay of two kinds of representations (concept, intuition), which are themselves defined in
terms of their different manner of ‘relating to their object’ (their content).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

knowledge [Wissen] in terms of an epistemic relation between a subject’s grounds for

‘holding [a judgment] to be true’ [Fürwahrhalten] and the objective grounds of the
judgment’s truth.54 This is very different from the definition of cognition, which is
not specified in terms of the holding-to-be-true of a judgment at all. There are even
passages where Kant explicitly contrasts Erkenntnis and Wissen, for instance, “there
can be a science of that of which our cognition [Erkenntnis] is not knowledge
[Wissen]” (JL, Ak. 9: 72).
On this interpretation, what defines cognition is not an epistemic warrant but the
way in which it represents its content; cognition is not a purely epistemic but a partly
semantic kind. Focusing for the moment on the modal definition, cognition of an
object represents it as the really possible object it is; this is why Kant imposes the
additional requirement that cognizing an object involves being able to differentiate it
from other possible objects.55 Cognition requires an epistemic warrant for knowing
the object to be really possible, but it is not itself that epistemic warrant or knowledge
or proof.
As I noted above, Kant gives two different kinds of characterizations of cognition,
a modal one (m-cognition, for short) and a semantic one (s-cognition, for short). We
saw m-cognition in the previous section, and we have been discussing s-cognition
here: thinking a given object under a concept. In some contexts Kant emphasizes m-
cognition,56 in other contexts, s-cognition,57 and this naturally leads to the question
of which one is fundamental and which one is derivative. I think that m-cognition is
fundamental and s-cognition (thinking a given object under a concept) is logically
downstream of it. I have three main reasons for thinking this.
First, the dialectical role that the notion of cognition plays in Kant’s critique of
traditional metaphysics. Kant will argue that we cannot cognize non-sensible objects
(noumena) and hence that we cannot have knowledge (Wissen) of them.58 Conse-
quently, metaphysics, conceived as a scientifically organized body of knowledge
(a science, Wissenschaft) of noumena, is impossible for us. This makes perfect

54
A822/B850. I examine the different modes of ‘Fürwahrhalten’ in detail in Ch. 9.
55
A58/B83, A242 n. Thanks to Karl Schafer for bringing this important point to my attention. Cf. Smit
(2000).
56
The locus classicus here is Bxxvi n. But I also think that a concept having ‘objective validity’ is
equivalent to it being a cognition, for Kant, so all of the passages that explicitly connect ‘objective validity’
with real possibility also count as ‘modal characterizations’ of cognition. See A239/B298.
57
E.g., A50–1/B74–5, B146.
58
There is a complication here. Kant takes himself to have sufficient warrant to make certain claims
about noumena: that there are noumena, that they are non-spatiotemporal, and that they causally affect us
(see Prol. 4: 315, A251–2, Bxxii n., A26/B42, Disc. Ak. 8: 215). So either these claims do not count as Wissen
in Kant’s sense (although we do possess sufficient rational warrant to justify asserting them) or Kant is
willing to allow some minimal Wissen of noumena in the Critical period. While I favor the former view, the
latter view is compatible with the larger point in the text: even if we know that noumena exist and are
therefore really possible, this leaves no room for the highly determinate and systematic knowledge of what
is possible for noumena required for a general metaphysics (ontology) of noumena. See Smit (2000),
Schafer (forthcoming), and Chignell (2014b) for further discussion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

sense with m-cognition, because Kant’s rationalist predecessors agree that to know
about some objects in a domain we must be able to prove that those objects are
possible.59 So Leibniz, Wolff, and Baumgarten would agree that if m-cognition of
noumena is impossible, then noumenal knowledge is impossible for us. What is
more, if m-cognition of noumena is impossible, then metaphysics, the science of all
possible beings qua possible, is impossible for us, since those beings include nou-
mena. By contrast, it would be question-begging for Kant to assume that s-cognition
is a necessary condition for knowledge. Leibniz, for instance, did not think that we
could have immediate singular thoughts about particular monads (except perhaps in
the case of the monad that I myself am), but did not think that was a barrier to
knowing about them. So m-cognition is dialectically better suited to Kant’s critique of
noumenal metaphysics.
Secondly, the concept of s-cognition is implicitly modal to begin with. Kant is not
claiming that for a concept to be an s-cognition we must be given an actual instance
of the concept in intuition. Mathematical concepts, for instance, are cognitions,
whether or not they have actual instances (as long as it is really possible that they
have instances).60 Even in passages where Kant is working with the notion of s-
cognition, he often makes clear that he means the concept relates to an object possibly
given in intuition.61 If s-cognition is implicitly modal then we should ask, what is
important about concepts of objects that are possibly given in intuition? As I will
argue in this chapter and the next, they are the very concepts we can know to be really
possibly instantiated.
This leads to my third point: concepts of objects that are possibly given in
intuition are concepts whose objects we can prove to be really possible (objects
of m-cognition). The converse also holds: objects of m-cognition are objects that
are possibly given in intuition. M- and s-cognition are co-extensive, or so I will
argue. While this might be thought to undermine the importance of the distinction,
I think, on the contrary, that it shows that the semantic notion is logically
downstream of the modal notion. In the rest of this section I will reconstruct
Kant’s reasoning from the notion of m-cognition to the notion of s-cognition. He
begins with the problem of m-cognition and arrives at the problem of s-cognition.
This will take two steps. In the first I reconstruct Kant’s reasoning that if a concept
is the concept of an object that can be given in intuition (s-cognition) then that
concept is an m-cognition. In the next section I reconstruct Kant’s argument that a
concept is an m-cognition (a concept of a knowably really possible object) only if it
is an s-cognition (a concept of an object that can be given in intuition). Kant begins

59
Recall Leibniz’s objection to the Cartesian ontological argument: we must first prove that <God> is
the concept of a possible object. See Ch. 1.3 for more on Leibniz’s idea that knowledge in metaphysics rests
on a prioriG knowledge of the possible instantiation of its concepts.
60
See B147, A156/B195, and esp. A224/B271.
61
E.g., Bxxvi, A3/B6, A4/B8, A63/B88, B146, A239/B298, A245, A247/B304, B383, B391, and B479.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

with the modal problem—how do we represent knowably really possible objects a

priori?—and proves that it is equivalent to a problem in cognitive semantics—how
can a priori concepts represent objects that can be given in sensible intuition?
If we begin by assuming the modal characterization of cognition, and we assume
further that our basic representations of objects are either concepts or intuitions,62
then it is relatively straightforward to explain how empirical cognition is possible. We
intuit a given object and we abstract a concept from the object by generalizing from
one of its features (the object is red and we form the general concept <red>). It is
actually the case, hence really possible, that an object instantiates the concept; to
know its real possibility we need only know that it instantiates the concept, and know
that actuality entails real possibility. There might be some difficult work to do in
spelling out exactly how empirical concept formation works, or how intuition of an
object puts one in a position to know that the object instantiates the concept, but
Kant does not regard it as mysterious how any of this is possible in the first place. But
this empirical model will not account for a priori cognition or for cognition of
unactualized possibility.
Kant’s theory of a priori intuition gives him some purchase on the problem of a
priori cognition and cognition of unactualized possibility, but not enough. We intuit
space and time a priori, which I take to mean: independently of how we are sensorily
affected by objects (hence a priori) we are immediately presented with (intuit) the
spatiotemporal structure in which any object given to us must appear. So we have a
priori cognition of the real possibility of at least two objects: space and time.
Furthermore, we can use this model to account for a priori mathematical cognition:
a mathematical concept is an a priori cognition just in case I can construct its object
in pure intuition and thereby prove that such an object can be given in space and
time.63 While I cannot go into the details of Kant’s theory of construction in pure
intuition, I want to point out that what ‘can and cannot be constructed’ in pure
intuition is itself a modal notion, so we need some account of it. Let us say that it is
intuitively possible that a given concept is instantiated just in case it is compatible
with the forms of intuition, space and time, that we intuit an instance of that concept.
In the next chapter I will have more to say about the nature of this compatibility
relation.
A priori constructibility in intuition by itself has serious limitations, though, as a
guide to real possibility because instances of the fundamental concepts of metaphys-
ics (e.g., substances, forces, etc.) cannot be presented in pure intuition.64 The
fundamental concepts of metaphysics are concepts of concrete beings with causal
powers, not of abstract formal mathematical objects that lack causal efficacy; conse-
quently, no mathematical construction is a construction of an instance of such a

62
Judgments are not basic because they are composed of concepts.
63
See esp. B147, A224/B271, A714/B742, and A141/B180.
64
A715/B743, A722/B750.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

concept. In the Prize Essay Kant had already pointed out that the concepts of
metaphysics cannot be constructed, but in the context of the CPR this has an
additional methodological consequence: a priori intuition alone will not explain
how cognition through the categories is possible.
This is the point which Kant’s thinking about the problem of a priori cognition
had reached by the time of his 1772 letter to Herz. He does not find empirical
cognition problematic, and he has the rudiments of an explanation of a priori
cognition of pure quantities in mathematics,65 but he cannot explain (or is unwilling
to explain in his correspondence with Herz) how a priori cognition of concrete non-
mathematical objects using the fundamental concepts of metaphysics—what he will
come to call the categories or ‘pure concepts of the understanding’—is possible. But if
we cannot cognize objects using categories (the fundamental concepts of metaphys-
ics) then metaphysics is impossible for us, because cognition, in the modal sense, is a
precondition for knowledge. We cannot, for instance, know the nature of substance
(e.g., whether all substances are simple) unless we can prove that substances are really
(not merely logically) possible. Kant’s ingenious solution in the CPR is to come up
with (i) an account of what it is to represent an object given in sensible intuition as
falling under a category, that (ii) specifies this in terms of the a priori forms of
intuition, space and time, such that (iii) representation of such objects is really
possible, according to a notion of real possibility that will be articulated in the course
of executing tasks (i) and (ii).
Kant’s answer to (i) is provided by his view of the relation between the logical
functions of judging and the categories. Judgments, for Kant, are hylomorphic
complexes of matter (their constituent concepts) and form (their principle of
unity).66 The logical functions of judging are principles of unity of judgments, the
manner in which concepts can be arranged to make judgments. The table of the
logical functions of judging (A70/B95) lists these different principles of unity,
organized according to four different moments: quantity, quality, relation, and
modality. By specifying the quantity, quality, relation, and modality we specify the
principle of unity of a judgment; judgments that agree in quantity, quality, relation,
and modality can only differ in their matter: the concepts that are combined into a
unified judgment according to the same form.67
Kant gives the logical functions a double duty, not merely uniting concepts into
judgments, but also uniting the manifold of representations in an intuition: “the very
same function that gives unity to different representations in a judgment also gives
unity to the different representations in an intuition, which, expressed generally, is

65
Corr., Ak. 10: 131.
66
For more on Kant’s ‘hylomorphic’ conception of judgment see Stang (2014a).
67
As Strawson (1966), 78 points out, hypothetical judgments (if p then q) do not have a quantity or
quality (but p and q do, assuming they are not further hypothetical judgments), so the specification of the
form of a judgment will be slightly more complicated than this.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

called the pure concept of the understanding” (A79/B104–5).68 This brings in two
crucial Kantian notions: the manifold and synthesis. An intuition represents a single
object but contains in itself a manifold (plurality) of parts, which are themselves
representations of the parts of the object.69 To intuit an object we do not need to
become conscious of the manifold of its parts, or to combine them in any way; we
can be given, in intuition, objects of which we are not conscious.70 But if we
consciously intuit an object then we are conscious of the manifold of representa-
tions in our intuition of it and it is possible for us to think of that manifold as
composing an intuition of a single object, i.e. to represent that intuition as a
complex whole composed of those parts. Representing a complex whole of parts
as a complex whole is what Kant calls ‘synthesis’ or combination.71 To represent a
complex whole of parts as such, to ‘synthesize’ it, we also need to represent the
principle of unity of that complex, the relation that holds among its parts in virtue
of which that complex is the complex it is.72 So for any intuition of an object of
which we are conscious, we can be conscious of the manifold of representations in
that intuition, and it is possible for us to combine or synthesize that manifold and
think of that intuition as a complex but unified intuition of a single object. For each
logical function of judging there is a corresponding principle of unity for synthesis
of manifolds in intuition. Just as logical functions are principles of unity for
judgments, these ‘schemata’ are principles of unity of conscious acts of combination
of the manifold of parts in an intuition. To think of a manifold of intuition as
unified according to a given schema (as having that principle of unity) is to think
of the object of that intuition as falling under the corresponding category.
For instance, I have a temporally extended intuition of an object, whose parts are
temporal parts or ‘time slices’ of the whole intuition. If I represent those ‘time slices’
as constituting a unified intuition of a single object that persists through time while
its appearance changes then I am representing that intuition as having a certain
principle of unity: representation of the changing accidents in a persisting object.
This corresponds, according to Kant, to the logical form of categorical judgment
and the category <substance-accident>. If I combine that manifold according to that
principle of unity then I am thinking of the object of that intuition as a substance,
that which persists while its accidents change.
I interpret this to mean that (a) for each logical function of judging there is a
corresponding rule for synthesizing manifolds (pluralities) of representations in

68
Cf. “I will merely precede with this explanation of the categories. They are concepts of an object in
general, by means of which its intuition is regarded as determined with regard to one of the logical
functions for judgments” (B128). My interpretation here is deeply indebted to Longuenesse (1998).
69 70 71
B136 n. See JL (Ak. 9: 35) and Anthr. (Ak. 7: 135). B129–131.
72
The qualification “to represent a complex whole of parts as such” is crucial—we do not need to
synthesize every whole, for, given that space and time are infinite given wholes, this is an infinite, hence
impossible, task. We are given space and time as infinite wholes. To think of their parts as parts of a whole,
though, we need to engage in an activity of synthesis. Cf. McLear (2015).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

intuitions, (b) we synthesize manifolds of representation in intuitions by apply-

ing these rules, and (c) because of this, the object of that intuition is repre-
sented as instantiating the corresponding category. These rules for synthesis are
what Kant calls the ‘schemata’ of the categories. This gives us an account of
what it is to represent an object as falling under a pure concept of the
understanding, Kant’s answer to question (i) from above: to represent an object
as falling under a category is to synthesize the manifold of representations in an
intuition of that object according to the schema of the corresponding logical
function of judging.
Kant’s answer to (ii) and (iii) follows almost immediately from this initial move.
The schemata of the categories are given in temporal terms; I interpret this to mean
that the manifolds of representation that are being synthesized are temporally
extended manifolds, that is, manifolds of outer intuition that occur at different
times, and the rules are specified in terms of the temporal properties of these
manifolds. To return to the example of <substance>, the schema for this concept is
the rule of organizing a temporally extended series of outer representations as
representations of the changing accidents of a persisting substance.
Finally, we need an answer to question (iii) from earlier: we need to show that it is
really possible for us to synthesize a manifold in an intuition according to these
logical functions of judging. This raises, of course, the pressing question of what
exactly real possibility means here. I will examine that question in detail in the next
chapter, but for now it suffices to point out that the very minimal notion of real
possibility articulated earlier—compatibility with the a priori forms of intuition,
space and time—will not suffice, for the simple reason that this combination of a
manifold of representations is a spontaneous activity of the understanding, not of the
passive faculty of sensibility. If all we have done is demonstrate that it is compatible
with the spatiotemporal structure of intuition that a temporally extended manifold of
intuition be represented as successive apprehensions of the changing accidents of a
persisting object, we will not have shown that it is really possible for the understand-
ing to so unify that manifold. Kant needs a more complete notion of real possibility to
answer this question.
The answer lies in Kant’s doctrine of the unity of apperception, one of the most
puzzling and complex ideas in the whole Transcendental Deduction, which, by
Kant’s own admission, is both the most difficult (and most important) part of the
Critique itself.73 I cannot offer a complete interpretation of the Deduction here, for it
would quickly engulf this entire book.74 For our purposes, a few points will have to
suffice. The unity of apperception is the unity of consciousness, the unity that obtains

73
Axvi.
74
Nor, for the same reason, can I engage with the extensive secondary literature on the Deduction. My
discussion here is deeply indebted to Longuenesse (1998), as well as Strawson (1966), Henrich (1968–9)
and (1976), Allison (1983) and (2004), Hanna (2001), and Allais (2015).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

among all of my conscious representations in virtue of being the conscious repre-

sentations of a single conscious subject. Kant claims for the unity of apperception in
relation to the understanding the very status he had earlier claimed for space and
time in relation to intuition, that it is a ‘highest’ ground of (real) possibility:
The supreme principle of all intuition in relation to sensibility was, according to the Tran-
scendental Aesthetic, that all the manifold of sensibility stand under the formal conditions of
space and time. The supreme principle of all intuition in relation to the understanding is that
all the manifold of intuition stand under conditions of the original synthetic unity of apper-
ception. (B136)75

Just as the ‘highest’ grounds of the possibility of objects of intuition are space and
time, the ‘highest’ ground of the possibility of those objects insofar as they are
thought by the understanding is what Kant here calls the “original synthetic unity
of apperception.”
Let me first explain what that term refers to. The analytic unity of apperception is
Kant’s term for the unity that obtains among my conscious states in virtue of their
being mine: “the ‘I think’ must be able to accompany all of my representations”
(B131). He argues further that the analytic unity among the manifold of representa-
tions in an intuition is only possible if a further synthetic unity holds among them,
and this synthetic unity is itself defined in terms of the possibility of a synthesis: “only
because I can comprehend their manifold in one consciousness do I call them all
together my representations; for otherwise I would have as multicolored, diverse a
self as I have representations of which I am conscious” (B134).
I interpret this to mean for any intuition whose object I can think about (con-
sciously subsume its object under a concept), the manifold of parts in that intuition
must have synthetic unity of consciousness, which means not that I actually con-
sciously synthesize that manifold, but that it is possible that I do so. The synthetic
unity of apperception refers to a necessary possibility: necessarily, if a sensibly given
object is thinkable by me, then possibly I synthesize (combine) the manifold of the
intuition in which that object is given.
That the synthetic unity of apperception is the ‘highest’ ground of the possibility of
objects of the understanding means, I take it, that the synthetic unity of apperception
is the ultimate ground of the possibility of using the understanding to think about
given objects. This means two things: anything that is a condition of the possibility of
thinking about given objects is either among the grounds of the synthetic unity of
apperception, or among its consequences. Everything that conditions the possibility
of thinking sensibly given objects must ‘go through’ the synthetic unity of appercep-
tion: it must be either explanatorily downstream or upstream of it. But notice there
are also conditions of the possibility of being given objects: space and time. So we

75
Cf. A107: “the numerical unity of this apperception grounds all concepts a priori just as the
manifoldness of space and time grounds the intuitions of sensibility.”
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

have two distinct sets of conditions of possibility: the conditions of being given
objects (space and time) and the conditions of thinking about them (the synthetic
unity of apperception, and its conditions). To experience an object is to think about a
sensibly given object using a concept. So we need to supplement our former con-
ception of real possibility—compatibility with the forms of intuition—with the
conditions of the possibility of thinking about objects to form a conception of what
we might call formal (real) possibility:
(Formal ) It is formally possible that we represent a certain (conceptual or
intuitional) content just in case it is compatible with the forms of intuition
(space and time) and the highest principle of the understanding (the synthetic
unity of apperception) that we represent a sensibly given object using that (con-
ceptual or intuitional) content.76
That the synthetic unity of apperception is the highest or ‘first’ ground of real
possibility of thinking about given objects means that any thought that is compatible
with the synthetic unity of apperception is formally possible, provided it also agrees
with the forms of intuition (i.e. it is not a thought of an object that cannot be given in
intuition). Since the synthetic unity of apperception just is the possibility of synthe-
sizing manifolds of intuition, it follows that if the synthetic unity of apperception is
itself possible (i.e. does not have inconsistent conditions of possibility) then it is
formally really possible that we synthesize intuited manifolds. I take it that Kant
assumes in the Deduction not only that the analytic unity of apperception is possible,
but that it is actual (I am aware of myself as the numerically identical subject of my
representations). It follows that the synthetic unity of apperception (which makes
analytic unity possible) is itself possible, from which we can conclude that synthesis
of manifolds of intuition is possible. Now we have already seen that the synthesis of a
manifold of intuition is synthesis according to the logical functions of judging and
their schemata, and this just is representing (specifically, thinking) the object of
intuition under the corresponding category. So if the former is really possible, the
latter is really possible as well. We can conclude, finally, that it is really possible to
think a sensibly given object under the categories, and we have given a preliminary
characterization of what the relevant notion of real possibility is (compatibility with
the intuitional and intellectual conditions of experience of objects—respectively,
space and time, and the synthetic unity of apperception).
This is an explanation of what grounds the real possibility of representing sensibly
given objects under the pure concepts of the understanding. It constitutes an answer to
one question Kant is raising in the Analytic of Concepts: how can we so much as
represent sensibly given objects as falling under the pure concepts of the understanding?

76
Cf. Chignell (2014b), which argues that the notion of real possibility involved in the ‘real possibility
constraint’ on knowledge is compatibility with background knowledge of nature and its laws. See Ch. 7.3
for discussion of the textual evidence for this claim.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

This is ultimately a question about representational content: what representational

content do pure concepts of the understanding have when they are applied to sensibly
given objects and how can that content be applied to them? This does not by itself
answer another question: how can we know a priori that it is really possible that objects
instantiate the categories? The account I have sketched above—which essentially
summarizes some key points in the Metaphysical Deduction, the first half of the
Transcendental Deduction (roughly, up to }23),77 and the Schematism—leaves open
the possibility that, while we can represent sensibly given objects under the categories, it
is impossible for them to actually instantiate those concepts. For instance, a Leibnizian
might allow, although we can think of objects of sensation as substances in an unquali-
fied sense by representing them as bearers of absolutely intrinsic properties (underlying
their perceptible relational-dispositional properties), it is impossible that sensibly given
objects could be substances because they are merely phaenomena bene fundata, the
well-founded phenomena of perception. Likewise, a Humean (though probably not
David Hume himself ) might allow that we can think of sensibly given objects using the
concept <necessary connection> without thereby admitting that it is really possible that
objects could instantiate such a concept.78
To fill this gap between, on the one hand, the real possibility that we think of
sensibly given objects using the categories, and, on the other, the real possibility that
they actually instantiate the categories, Kant must invoke his transcendental idealist
thesis that the objects given in intuition are appearances, not things in themselves.
The meaning of Kant’s transcendental idealism and the distinction between appear-
ances and things in themselves has been controversial since the publication of the
A edition in 1781, and that controversy has continued down to the present day.79 It is
not my intention to enter into that controversy here, for we can understand how
transcendental idealism fills the gap in Kant’s account while remaining neutral
among competing interpretations of transcendental idealism. The key idea in closing
this gap is that the real possibility of thinking a given object under the categories entails
the real possibility of an object falling under the categories.80 The first terms refers to
the real possibility of a certain representational content (thinking of a given object
that it instantiates a category); the second term refers to the real possibility of given
objects instantiating categories.

77
Which naturally raises the question, what is left to do in the second half of the B Deduction? My own
view, which I do not defend here, is that the second half shows the real possibility of sensibly given objects
instantiating the categories. The solution that I offer to that problem in the rest of this section is a more
modest solution than, on my reading, Kant actually gives in }}24–7.
78
I would not attribute this view to Hume himself for, on my reading, Hume denies that we have any
idea of (what Kant would consider) a necessary connection (except insofar as the idea of a necessary
connection is just the idea of a constant succession accompanied by a feeling of certainty), so, in Kantian
terms, we cannot even think of objects under (what Kant considers) the category <cause–effect>.
79
Stang (forthcoming) provides an overview of the current state of the debate; Beiser (2002) covers the
historical background.
80
See A111.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

The reason why this entailment holds is that the objects of experience are appear-
ances, not things in themselves. While there are different ways of understanding this
distinction, they all agree that this entailment holds. For instance, on the phenom-
enalist reading of this distinction ‘appearances’ and ‘things in themselves’ refer to
distinct classes of entities. Appearances are merely the ‘intentional objects’ of
experience, objects whose ‘whole being’ is grounded in the contents of experience;
if it is really possible that we experience appearances a certain way (e.g., falling
under the categories) then it is really possible that they do fall under the categor-
ies.81 On the ‘one object’ reading ‘appearances’ and ‘things in themselves’ do not
refer to different classes of objects but to different properties possessed by one and
the same class of objects.82 There are many different interpretations of what this
distinction amounts to, but I think it should be clear that on any such reading the
entailment holds. For instance, if we take this distinction to be the distinction
between the properties objects can appear to have in experience and the properties
that can never be revealed in experience, the entailment holds: if it is really possible
for us to experience objects as having certain properties (e.g., falling under the
categories) then it follows that those properties are among their ‘empirical’ prop-
erties, which just means it is really possible that appearances (or objects qua
experienced by us) have those properties.
The third class of interpretations of transcendental idealism differ from the first
two in not taking the distinction between appearances and things in themselves to be
a metaphysical distinction (either between two kinds of objects, in the case of the
phenomenalist, or between two kinds of properties, in the second case) but a
methodological distinction between two standpoints we can take on objects: we
can consider them as objects of experience for spatiotemporal discursive intellects
(the ‘empirical standpoint’) or we can consider them as objects of an intellect in
general (the ‘transcendental standpoint’).83 This interpretation also upholds the
entailment, for if it is really possible (compatible with the conditions that have to
be in place for us to experience objects in the first place) for us to represent objects
under the categories, then it is really possible for objects to fall under the categories,
insofar as objects are considered from the empirical standpoint. So even on the
‘methodological’ reading of Kant’s transcendental idealism, the entailment holds.
This is a reconstruction of Kant’s account of how it is possible for us to cognize the
real possibility of sensibly given objects falling under the categories. Note that it will

81
The phenomenalist interpretation of Kant was inaugurated by the (now infamous) Feder-Garve
review, the text of which can be found in Karl Vorländer’s edition of the Prolegomena (Kant 1976), 167–74;
Sassen (2000), 54–8 contains a translation. Sellars (1968) and (1976), Aquila (1979) and (1983), ch. 3, Van
Cleve (1999), and Stang (forthcoming) attempt, in different ways, to rehabilitate the phenomenalist reading
of transcendental idealism, while rejecting the Feder-Garve view.
82
Recent defenders of this view include Allais (2004), (2006), and (2007); Rosefeldt (2007) and
(forthcoming); Marshall (2013); and McDaniel (2013).
83
I have in mind mainly Allison (1983) and (2004).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

only succeed for objects that are appearances, not for things in themselves. For things
in themselves, objects that do not depend upon our manner of sensibly cognizing
them, the entailment does not hold. There is a ‘modal gap’ in the case of things in
themselves, between the real possibility of our representing them under the categor-
ies and the real possibility of their falling under the categories. More fundamentally,
though, the account of what it is to represent an object as falling under the categories
does not apply to things in themselves, because (given Kant’s transcendental idealist
thesis that all objects given to us in intuition are appearances) things in themselves
cannot be given to us in intuition. So this account does not issue in an explanation
either of (a) what it is to represent things in themselves under categories or (b)
whether we can cognize (knowingly represent the real possibility of) things in
themselves instantiating the categories. In section 8 I reconstruct Kant’s argument
that (b) we cannot cognize things in themselves using the categories. In Chapters 9
and 10 I explore his account of (a) what we are doing in representing things in
themselves under the modal categories.
Before continuing, I want to remark on the difference between two distinctions
Kant uses to articulate his transcendental idealism: appearance/thing in itself, and
phenomena/noumena. The appearance/thing in itself distinction is a distinction
between things that depend upon being sensibly intuited (appearances) and things
that do not so depend (things in themselves).84 The phenomena/noumena distinc-
tion is a cognitive distinction between objects that can be given in sensible intuition
(phenomena) and objects that cannot be so given (noumena).85 This distinction
assumes nothing about the metaphysical status of phenomena or of noumena; even a
transcendental realist could accept it. Since all of our intuition is sensible, we intuit
only phenomena, by definition. All objects we intuit are either in space or time, and
Kant argues in the Transcendental Aesthetic that space and time present only
appearances, not the things in themselves that appear. Consequently, <appearance>
and <phenomenon> are co-extensive concepts, though they differ in intension
(meaning).86 By the same reasoning, all things in themselves are noumena, for they
cannot be sensibly intuited. Conversely, a noumenon that is not a thing in itself
would be a being that cannot be sensibly intuited (noumenon) but depends upon
being sensibly intuited (an appearance to sensible intuition), which is absurd.
<Noumena> and <things in themselves> [Dinge an sich selbst] are co-extensive

84
See esp. Kant’s definition of transcendental idealism at A369; cf. A42/B59.
85
A252; this corresponds in the B edition to the ‘negative’ concept of a noumenon (B307).
86
Kant does distinguish appearances from phenomena at A248–9, but I think the dominant meaning of
‘phenomena’ in this section is: possible object of a sensible intuition. One might also wonder whether there
could be phenomena (objects of sensible intuition) for intellects other than ours, but which are not
appearances (whether some discursive intellects intuit things in themselves). Kant claims that sensible
intellects, in virtue of being passive, cannot intuit things in themselves (B68, B152–3, B156, B159), but does
not base his case for transcendental idealism on this ‘short’ argument. See Ameriks (2003) for discussion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

concepts, though they differ in intension.87 Hence, if we cannot cognize noumena, a

fortiori we cannot cognize things in themselves.
Bear in mind, though, that by drawing the distinction between <phenomena=possible
object of sensible intuition> and <noumena=not a possible object of sensible intuition>
Kant is not thereby committing himself to there being noumena. This is his point when
he writes: “the division of objects [Gegenstände] into phenomena and noumena, and
of the world into a world of sense and a world of understanding, can therefore not
be permitted at all in a positive sense, although concepts certainly permit of division
into sensible and intellectual ones” (A255/B311, underlined portion added in the
B edition).88 But we should not, for that reason alone, conclude that noumena are
impossible, for “one cannot assert of sensibility that it is the only possible kind of
intuition” (A255/B310).

6.8. No a priori Cognition of Noumena

In this section I will reconstruct Kant’s reasons for denying that we can (modally)
cognize objects other than phenomena; since cognition is a precondition for know-
ledge [Wissen] of an object, it will follow both that we cannot have knowledge of
noumena (objects that cannot be given in sensible intuition) and thus that we cannot
know all possible objects überhaupt.89 Metaphysics is possible for us only if it is
restricted to phenomena.
Kant typically argues that we cannot cognize noumena by claiming that we can
only cognize objects that we can intuit. However, this strategy of argument relies on a
premise that would not be granted by all of Kant’s interlocutors: that we can only
cognize (know the real possibility of) objects that can be given to us in intuition. This
premise becomes especially problematic when it is paired with Kant’s assumption, for
which he never explicitly argues in the CPR, that all of our intuition is sensible (that
we have discursive intellects). While many of Kant’s commentators find these claims

87
I know these claims will strike some readers as misguided, but there is strong textual evidence that Kant
regards ‘Dinge an sich selbst’ and ‘noumena’ as co-extensive terms: Cf. A254/B310, A256/B312, A259/B315;
Prol. }30 (Ak. 4: 312), }32 (314–15), }33 (315), and }59 (360); Disc. (Ak. 8: 208); and Prog. (Ak. 20: 292, 308).
88
Cf. his comment in the previous sentence that the concept <noumenon> does not allow us to “posit
[read: posit absolutely] anything positive outside” the domain of the senses.
89
The exact interplay between real possibility, cognition, and knowledge is a matter of scholarly debate;
I do not have the space to discuss it fully here. All parties agree that cognition limits in some fashion what
we can know; the question is how to spell this out in detail. Chignell (2014b) points out that Kant seems to
allow some very general knowledge about noumena (e.g., that they are non-spatiotemporal, that they affect
us) while denying that cognition of noumena is possible (see discussion in section 7 of this chapter).
Schafer (forthcoming) argues that, while this entails something very much like modal knowledge of
noumena (e.g., if we know that there are noumena we know a fortiori that they are really possible), it
fails a further requirement on cognition: that we cognize objects determinately (e.g., we know that there are
noumena but nothing about their determinate positive properties). If Schafer is correct that we ‘know’ (in
some sense, though not perhaps Kant’s technical notion of Wissen) some minimal real possibilities for
noumena, my main point still stands: this minimal knowledge of real possibility would not allow for the
robust modal knowledge of noumena required for noumenal metaphysics. See also Smit (2010).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

prima facie plausible they would find universal assent neither among Kant’s early
modern predecessors nor among later philosophers (even those influenced by
Kant).90 Since I am attempting to reconstruct Kant’s best argument that cognition
of noumena is impossible, I will focus on arguments that do not rely on contentious
premises about the intuition-dependence of cognition or on the assumption that our
faculty of intuition as such is sensible.
The closest thing Kant gives to a general argument that cognition of noumena is
impossible, without relying on controversial premises about the intuition dependence
of cognition, is the passage at the beginning of the Transcendental Deduction, from
which I quoted a small section at the beginning of section 4:
There are only two possible cases in which synthetic representations and its objects can come
together, necessarily relate to each other, and, as it were, meet each other: either if [1] the object
alone makes the representation possible, or if [2] the representation alone makes the object
possible. If it is the first [1b], then this relation is only empirical, and the representation is never
possible a priori. And this is the case with appearances in respect of that in it which belongs to
sensation. But if it is the second [2], then since representation in itself (for we are not here
talking about its causality by means of the will) does not [2a] produce its object as far as its
existence is concerned, the representation is still determinant of the object a priori if [2b] it is
possible through it alone to cognize something as an object. (A92/B124–5—bracketed num-
bers refer to my analysis of the argument below)

This parallels the following passage from the letter to Herz:

[O]n what grounds rests the relation [Beziehung] of what in us is called representation
[Vorstellung] to the object? If [1a] the representation contains only the way in which the
subject is affected by the object, then it is easy to see how the representation corresponds to the
object, as an effect to its cause, and how this determination of our mind can represent
something, i.e. how it can have an object. The passive or sensuous representations thus have
a graspable relation to objects, and the principles that are derived from the nature of our soul
have a graspable validity for all things insofar as they might be objects of the senses. Likewise, if
[2a] that in us which is called representation were active with regard to the object, i.e. if the
object were produced by the representation itself (as one thinks of divine cognitions as
archetypes of things), then the conformity of the representations with the objects would also
be able to be understood. And so one can at least understand the possibility of both an
archetypal intellect [intellectus archetypus], upon whose intuition the things themselves are
grounded, as well as ectypal intellect [intellectus ectypus], which attains the data of its logical
activity from the sensuous intuitions of things. But neither is [2a] our understanding by means
of its representations the cause of the object (except in morals, the cause of good ends), nor is
[1a] the object the cause of the representations of the understanding in the real sense (in sensu
reali). The pure concepts of the understanding must, therefore, not be [1a] abstracted from the
sensation of the senses, nor must those concepts express the receptivity of representations

90
E.g., Hegel, who criticizes Kant on exactly this point in his 1802 work, Faith and Knowledge: H.Werke
2: 325–6.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

through sense; but they must, to be sure, have their sources in the nature of the soul, though
not [1a] insofar as they are produced by the object nor [2a] insofar as they bring forth the
object itself. (Corr., Ak. 10: 130)

I argued above, in section 4, that the relation between representation and object with
which Kant is concerned in both passages is cognition: representing a knowably really
possible object. Although his own positive solution depends upon intuition, Kant
raises the question, in both passages, primarily about the pure concepts of the
understanding (categories): how can we cognize objects through pure concepts of
the understanding?91 His reasoning, in both passages, takes the form of a dilemma:
either (1) the object is the ground of the concept, or (2) the concept is the ground of
the object. Whereas, in the Herz letter, Kant concludes pessimistically that a priori
cognition using pure concepts of the understanding seems to be impossible on either
horn of the dilemma, in the CPR he thinks this possibility can be understood through
the second horn (2): the pure concepts of the understanding make their objects
possible. We have already seen Kant’s positive solution, so now I want to reconstruct
his reasons for thinking this dilemma is exhaustive and his reasons for excluding the
other option(s).
Kant’s argument relies on a disjunctive premise:
(Disjunction) Where C is a concept, if C is a cognition then either (1) the object
of C grounds C, or (2) the concept C grounds its object.
This disjunctive premise needs to be fleshed out further, in several respects.92 First of
all, it is not clear what kind of grounding relation is involved. On the one hand, Kant
seems to assert a causal relation between an object and its empirical concept in the
letter to Herz (“the representation corresponds to the object, as an effect to its
cause”); causal grounding is also relevant in the practical case, where the representing
of a particular action as my end is among the causes of the existence of that action
(“neither is our understanding by means of its representations the cause of the object
(except in morals, the cause of good ends)”). On the other hand, the non-causal
grounding of possibility also clearly has a role to play, since Kant’s ultimate conclu-
sion will be that the categories are non-causal grounds of the possibility of objects.
Consequently I propose that we divide each horn of the dilemma into a causal and a
non-causal case, and represent Kant’s disjunctive premise as a table of possible views
(Table 6.1).
Regarding (1b), note that, while in the Herz letter Kant says that the object of an
empirical concept is the cause of the concept, in the CPR he says that the object
makes the concept possible (and this is why the concept is “not possible a priori”).
I explain what this difference might amount to below.

91
This is because he possessed a rudimentary account of a priori intuition and mathematical cognition,
since at least 1770 in ID. See }3 of ID (Ak. 2: 398–406).
92
This section is heavily indebted to Eli Chudnoff, who helped me think through Kant’s options.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

Table 6.1. Kant’s disjunctive premise

(1) The object grounds (2) The concept grounds the

the concept object
(a) Causal grounding (ground Empirical concepts Practical representation
of existence) (Herz letter)
(b) Non-causal grounding Empirical concepts Kant’s positive view: cognition
(ground of possibility) (CPR) of phenomena

However, further refinements are needed before we can reconstruct Kant’s argu-
ments for excluding the other options. In this study I have typically formulated
grounding relations as relations between facts (or between an object and a fact, e.g.,
between God and the fact that some predicate is really possible), so we must
determine what fact about the concept is being grounded by what fact about the
object (and vice versa). The easiest case to consider is the empirical one: an empirical
concept is a cognition because we experience an actual instance, so we know a fortiori
that it is really possibly instantiated. In the empirical case it is only possible for us to
represent the relevant content if there is an actual instance in our environment and
we experience it. This is a point that Kant constantly makes about concepts of
particular forces: we cannot have a concept of a particular force unless we actually
encounter it.93 In the empirical case (horn (1)) we can identify the relata of the
grounding relation as the fact that the concept is instantiated and the fact that we
have a concept with that representational content (that we possess that concept,
assuming concepts are individuated by their content). Assuming the relata of the
grounding relation are parallel in the two horns of the dilemma, then in horn (2) the
relata will be the same (the fact that we possess a concept, and the fact that the
concept is instantiated). In the case of non-causal grounding of possibility (b), what is
grounded is the possibility of the second fact (e.g., that we possess a certain concept,
(1a)). If this is correct, we can reformulate the disjunctive premise as the chart of
alternatives in Table 6.2, where bracketed expressions denote facts.
Kant of course, wants to exclude (1) entirely and conclude that, in the case of a
priori theoretical cognitions, only (2b) remains, which will lead to the conclusion that
the only objects we can cognize are phenomena (objects that are made possible by
our representations of them, appearances). But before we examine his reasons for
rejecting (1) it should be pointed out that even on Kant’s, admittedly idiosyncratic,
way of seeing the problems and their potential solutions there is a third alternative
that he does not consider explicitly in the quoted passages, but which does come in

93
Thus there are no a priori concepts of particular forces; see A207/B252, A222–3/B269–270, A291/
B347.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

Table 6.2. Kant’s initial alternatives

(1) [the concept is instantiated] (2) [we possess the concept]

grounds [we possess the grounds [the concept is
concept] instantiated]
(a) Causal grounding Empirical concepts (Herz letter) Practical cognition
(ground of existence)
(b) Non-causal Empirical concepts (CPR) Kant’s positive view
grounding (ground of
possibility)

Table 6.3. Complete set of alternatives

(1) [the concept is (2) [we possess (3) Common source

instantiated] the concept] grounds both [the
grounds [we grounds [the concept is instantiated]
possess the concept] concept is and [we possess the
instantiated] concept]
(a) Causal grounding Empirical concepts Practical Crusius (see below)
(ground of existence) (Herz letter) cognition
(b) Non-causal Empirical concepts Kant’s positive Malebranche (see below)
grounding (ground of (CPR) view
possibility)

for discussion later, both in the Herz letter and the CPR.94 The third alternative is
that our concepts and their objects both depend upon a common source; in the
context of the eighteenth century, the natural candidate is God. So we should really
add a third column of possible alternatives, as in Table 6.3.
Option (3) actually involves two different grounding relations: one between the
‘common source’ (God) and our concept, and one between the ‘common source’
(God) and objects. We might want to divide (3) into four rather than merely two sub-
cases, but, for the sake of simplicity, I shall make the Kantian assumption that the
‘common’ source is the non-causal ground of the very possibility of objects instan-
tiating our concepts, so we only need to distinguish whether God is the cause of the
existence of our concepts with their representational content or their non-causal
ground. We can quickly dispense with (2a), for, while this is the relation between
concepts and objects in practical cognition, in theoretical cognition our concepts do
not causally bring about the existence of their objects. Kant’s strategy is to exclude

94
Ak. 10: 131 f. and B167–8. I take Kant’s mention of a “preformation system of pure reason” to be a
reference to Crusius; see discussion below.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

each of the options represented in this table as a potential explanation of a priori

cognition, until he is left only with (2b), his own view.
By representing these options in a table I do not mean to be committing Kant to
the assumption that they are exclusive. To take the empirical case (1), it might be that
the presence of gold in my environment is both a non-causal ground of the possibility
that I have the concept <gold> and the causal ground of my actual possession of that
concept; Kant’s argument will be that, if either of them is true for a given concept,
that concept is not an a priori cognition. What is more, Kant’s reasons for rejecting
option (1) and its sub-options are different in kind than his reasons for rejecting
option (3) and its sub-options. Kant’s objection to (1) is that if objects were the
grounds of our possessing the categories then a priori cognition through the cat-
egories would be impossible. His objection to (3), by contrast, is that even if there is
some common source, both of our possession of category-concepts and of their
instantiation by objects, this does not explain why a priori cognition is possible. This
does not allow him to conclude that (3) and its sub-options are false, but merely that
they have no role to play in a transcendental critique of our cognitive faculty: an
inquiry into the grounds of possibility of our a priori representation of knowably
really possible objects.

(i) Kant’s reasons for rejecting object-dependence

Kant’s reasons for excluding (1a)—the instantiation of the concept is (one of) the
causal grounds of our possessing the concept—are fairly straightforward. If my
possession of the concept causally depends upon the concept being instantiated by
an object, then I have experienced an instance and abstracted the concept. This
model might be acceptable for empirical concepts, but it cannot be correct for a priori
concepts like the categories; if my possession of the categories depends causally on
their objects, then the fact that I possess these concepts depends upon the particular
course of my experience, so these concepts would not count as a priori concepts in
Kant’s sense: concepts, the possession of which does not depend upon the particular
course of our experience.95
This option, (1a), corresponds to the causal model Kant rejects in the Herz letter:
“if the representation contains only the way in which the subject is affected by the
object, then it is easy to see how the representation corresponds to the object, as an
effect to its cause” (Ak. 10: 131). However, in the CPR Kant suggests a different
model: “if it is the first [the object alone makes the representation possible], then this
relation is only empirical, and the representation is never possible a priori. And this is
the case with appearances in respect of that in it which belongs to sensation” (A92/
B124–5). Kant here claims that the object of an empirical concept makes that concept
possible; this means that, if I do not experience an instance of the empirical concept,

95
See Chapter 7.3 for more on this notion of the a priori.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

then that empirical concept is not so much as thinkable for me. This is plausible, for
instance, about certain very basic sensory concepts; if one has not yet experienced the
taste of pineapple then one cannot have the concept <taste of pineapple>. Kant makes
the analogous claim about concepts of particular physical forces: unless the force in
question is instantiated by some object, you cannot so much as have the concept of
that force (think about that force in particular).96 Consequently, we cannot so much
as think about particular uninstantiated forces (we can only think about force in
general, and particular instantiated forces).
This is a similar thought to what is now called ‘semantic externalism.’ For instance,
if the clear drinkable stuff in lakes and rivers were XYZ rather than water, we could
not have thoughts about water, but only about XYZ.97 This is relevant to Kant,
because if we apply the idea of semantic externalism back to the categories we obtain
the following possible alternative picture: possessing the concept <substance>, that is,
being able to think about substances, requires being in an environment where there
are substances (likewise for the other categories). This might be thought to face the
same problem that (1a) did—how can we know a priori that we are thinking about
substances, and thus are in a substance-laden environment?—but the externalist may
be able to resolve this problem by appealing to another Kantian doctrine: to think
about objects we must think using the categories. However, on the externalist view
I am sketching, this principle receives a different interpretation: to think about
objects is to use certain concepts (e.g., <substance>), but to use those concepts
requires being in a world where those concepts are instantiated. Now the principle
that “to think about objects we must think using the categories” is a principle that
Kant regards as a priori, and what is more, as a priori whether or not the objects in
question are phenomena or noumena. So the externalist defender of metaphysical
realism could elaborate the following alternative: to think at all is to be intentionally
related to certain conceptual contents, the fundamental concepts of all thought (the
categories), and to possess those concepts requires there actually being objects
(noumena) instantiating them in one’s world. This promises to give us an a priori
proof that the categories are instantiated, but a proof that rests on the opposite of the
ground Kant wants it to: the instantiation of the categories is the ground of the
possibility of our possessing these concepts and, indeed, of thinking at all.98
However, in appealing to the principle that the possibility of category-thoughts
non-causally depends upon the instantiation of the categories in one’s environment,
the ‘metaphysical externalist’ is making a claim about objects überhaupt, including
both phenomena and noumena. Our ability to cognize noumena is precisely what is
at issue, so a circularity worry might arise: this is only an explanation of our ability to

96 97
E.g., B252, MFNS (Ak. 4 :486–7). See Kripke (1972) and Putnam (1975).
98
This ‘externalist’ response to Kant is closely related to the issue of externalism and ‘armchair’
knowledge; see Burge (1988). I think it is also part of Hegel’s idealism in the Science of Logic, but I do
not have space to defend that claim here. Thanks to Eli Chudnoff for helpful discussions on this point.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

cognize noumena if we can cognize (the concepts involved in) the ‘metaphysical
externalist’ principle, but that principle is itself a principle (in part) about noumena
so it is circular to appeal to it. It is hard to know how problematic circularity of this
kind is; I will postpone further discussion of this issue until the final subsection.

(ii) A priori cognition of phenomena

Kant dismisses option (2a) out of hand as an account of theoretical cognition because
our concepts of objects do not cause the existence of their objects. Not even the
categories make it the case that, for instance, substances exist. Kant adamantly
maintains, in the face of those who would assimilate his view to Berkeley’s, that
our representations are among the grounds of the possibility of objects but do not
suffice to ground their existence. For objects of experience to exist there must be
objects affecting us, providing the matter of experience and its objects. Option (2b)
represents Kant’s own positive view of a priori cognition, reconstructed in the
previous section. If the categories are among the grounds of the possibility of their
objects (objects instantiating them), as they are on Kant’s transcendental idealist
theory, then a priori cognition of those objects is possible. But this only explains how
a priori cognition of phenomena, objects that can be given in sensible intuition, is
possible; it does not explain a priori cognition of noumena.

(iii) Kant’s reasons for rejecting ‘common source’ models

The idea behind (3) is that the categories are concepts of really possible objects
because God is the common source of both our concepts and real possibility. Kant
distinguishes two different versions of such a view: a ‘pre-established harmony’ view
(which elsewhere he refers to as the ‘preformation system’ of pure reason—B167)
on which God has created our intellects so that they track real possibility,99 and a
‘hyper-physical’ influx view he associates with Malebranche, on which we currently
stand in connection with God’s mind.100 I will begin by explaining the former view,
which Kant associates with Crusius, and why it should be understood as option (3a)
in my schema; then I’ll explain why the Malebranchean view is an instance of (3b).
I conclude by explaining Kant’s common objection to both views as explanations of
the possibility of a priori cognition: they are circular.101
The highest principle of Crusius’s modal epistemology, which figures in his
explanation of why (to use Kant’s terms) the categories (the basic concepts of
metaphysics) are a priori cognitions, is that conceivability and possibility are co-
extensive: “the very highest distinguishing feature of possible and actual things is the

99
Cf. Kant’s discussion of Crusius in Disc. (Ak. 8: 245 f.), Refl. 4275 (Ak. 17: 492), 4446 (Ak. 17: 554),
4866 (Ak. 18: 14), as well as LP (Ak. 24: 335), MH (Ak. 28: 10–11, 156), MV (Ak. 28: 372), and MvS (Ak. 28:
467–8).
100
Corr., Ak. 10: 131. Cf. Refl. 4275 (Ak. 17: 492), 4446 (Ak. 17: 554), and 4866 (Ak. 18: 14).
101
Kant also includes Plato as a representative of the ‘preformation’ view in the Herz letter, but I will
focus on the early moderns, Crusius and Malebranche.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

essence of the understanding, namely that what cannot be thought as such is not
possible or actual, and that, by contrast, what can be thought is possible.”102 The
scope of what is ‘thinkable’ by our intellects is specified by three fundamental laws:
the principle of contradiction, the principle of the inseparable, and the principle
of the uncombinable. The law of non-contradiction provides the ‘formal’ criterion of
conceivability; anything that violates it is inconceivable, hence impossible. The
second two principles—the principle of the inseparable and the principle of the
uncombinable—are ‘material’ principles of thought because they do not pertain
merely to the logical form of a putative possibility. Some concepts are uncombinable
in thought, even though they are not logically contradictory. Consequently, it is
impossible for them to be co-instantiated. For instance, it is inconceivable, hence
impossible, that a person be in two locations simultaneously, although this is not
logically contradictory.103 Likewise, some concepts are inseparable in thought, even
though there is no contradiction in affirming one, while denying the other. Conse-
quently, they are necessarily co-instantiated, but this necessity is not a logical
necessity. For instance, it is inconceivable, hence impossible, that an alteration in a
substance lack a sufficient ground, although there is nothing contradictory in the
concept of an uncaused event.104 Since not all logically consistent concepts are
conceivable (given these principles), not all logically consistent concepts are possibly
instantiated. Real possibility, for Crusius, as for Kant, is a restriction on logical
possibility: not all logical possibilities are really possible.105
Conceivability (defined in terms of the three laws above) is a guide to possibility
because God has given us intellects that cannot combine in thought really incom-
patible concepts, and can combine in thought really compatible concepts.106 In
Kantian terminology, Crusius’s view is that we know that the categories are really
possibly instantiated because it is conceivable (in the sense described above) that they
are; thinking of a substance with accidents, or a cause that is sufficient to bring about
the existence of an alteration, does not involve combining anything that cannot be
thought together, or separating anything that cannot be thought separately.107
Crusius’s view is most naturally understood as the view that God is the causal
ground of our having the intellects we do, with certain laws of conceivability. In the
context of Kant’s argument, this means that God is the causal ground of our
possession of category concepts, our ability to think about objects as, for instance,
substances (3a). On Malebranche’s view, though, our mind’s dependence on God
takes a more radical form: we represent objects (other than our own mental states) by

102
Ent. }15; translation from Watkins (2009), 140. Cf. Weg }261.
103
De Usu }27. 104
De Usu }27.
105
In fact, it is plausible that Kant borrowed his distinction of logical from real possibility from Crusius.
106
See De Usu }29; Ent. }237, 287. Cf. Heimsoeth (1956), 163.
107
In the letter to Herz, Kant writes of this solution: “Crusius believed in certain implanted rules for the
purpose of forming judgments and ready-made concepts that God implanted in the human soul just as
they had to be in order to harmonize with things” (Corr., Ak. 10: 131).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

perceiving ideas of them in the divine intellect. Without God’s intellect it would not
be so much as possible for any created mind to represent outer objects, much less
represent them as substances.108 “Through his presence God is in close union with
our minds,” writes Malebranche in The Search after Truth, “such that He might be
said to be the place of minds as space is, in a sense, the place of bodies.”109 If we take
this metaphor in a Kantian spirit, then Malebranche’s view would be that the divine
intellect is the complete space of possible thinkable content; the ideas presented to
our minds are merely a limited subset of that space. In the context of Kant’s
argument, this means not only does God cause it to actually be the case that we
have category-concepts (that we perceive those ideas) by creating us and willing that
we do; God’s mind is the non-causal ground of the mere possibility of those concepts
(those ideas). In my way of dividing the options, Malebranche represents (3b): the
possibility of our possessing category-concepts (the possibility of our ‘seeing’ those
ideas) non-causally depends upon God.110 To apply Malebranche’s view to Kant’s
specific problem, the reason our category-concepts are possibly instantiated is that
they are ideas in God’s mind, and divine ideas are essences of possible things.111
Despite this difference, though, Kant levels the same objections against both the
‘preformation’ system (Crusius) and the ‘hyper-physical influx theory’ (Male-
branche): he makes a ‘bad company’ objection, and he accuses them of circularity.
In the Herz letter he says of both views that they “encourage all sorts of wild notions
and every pious and speculative barbarism” (Ak. 10: 131), a point to which he returns
in the Prolegomena: “with the lack of sure criteria for distinguishing an authentic
origin from a spurious one, the use of such a principle [either Crusian divine
preformation or Malebranchean immediate presence to God’s mind] looks very
precarious, since one can never know what the spirit of truth or the father of lies
may have put into us” (Ak. 4: 320 n.). Kant is raising an essentially Cartesian
objection to Crusius and Malebranche: perhaps an evil demon has implanted the
‘laws of thought’ in our minds, or is revealing to us ideas of impossibilities.
However, I am going to focus on Kant’s objection in the Herz letter that the
theories of Malebranche and Crusius are circular: “the deus ex machina [ . . . ] has—
besides its vicious circularity in drawing conclusions concerning our cognitions—
also this additional disadvantage [ . . . ]” (Corr., Ak. 10: 131). Recall the dialectical
context in which Kant discusses these views: we want to explain how (and whether) a
priori cognition of noumena is possible, and cognition is a prerequisite for

108
Malebranche makes the un-Kantian assumption that representation of bodies (outer sense) is in
principle different than representation of states of the mind (inner sense). Awareness of the states of the
mind, according to Malebranche, requires no mediating idea. See Search, 1.2.3–7.
109
Search, 3.2.6 (Malebranche 1980, 230).
110
It should be noted that Crusius also thinks that God is the ground of all (real) possibility, so a fortiori
he is the ground of the possibility of our being able to think of objects as substances. While this is part of
Crusius’s overall metaphysical theory it is not part of his account of how we can know that our thoughts
track real possibility, or in Kant’s terms, his theory of how we can cognize objects a priori.
111
Malebranche (1997), 74.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

knowledge. To prove the real possibility of noumena by appealing to some meta-

physical theory requires, presumably, that that metaphysical theory itself is an object
of knowledge and, therefore, of cognition. But the metaphysical theory to which
Crusius and Malebranche appeal is a theory about noumena themselves: God and his
intellect. Therefore, in arguing that cognition of noumena is possible, Crusius and
Malebranche must first assume that it is. This is the alleged circularity in their
theories.112
However, as anyone familiar with the ‘Cartesian circle’ and the debates surround-
ing it knows, the issue of circularity in epistemological theories is a complex one. We
might think of Malebranche and Crusius as providing sufficient conditions on
knowledge of possibility (cognition), namely:
(3a) An agent S knows that a concept C is really possibly instantiated if (i) C is
logically consistent and contains no elements that cannot be combined by S in
thought, and does not involve separating any elements that cannot be thought
separately by S, (ii) God has created S’s intellect so that if (i) is true then C is really
possibly instantiated, and (iii) S bases her judgment that C is really possibly
instantiated on (i). [Crusius]
(3b) An agent S knows that a concept C is really possibly instantiated if
(i) S perceives C, (ii) C is in God’s mind and (iii) S bases her judgment that C is
really possibly instantiated on (i). [Malebranche]113
These principles do not entail that, if S knows that some concept of noumena is really
possibly instantiated, then S has some antecedent knowledge of God (or about any
other noumena), in particular. In this sense, they are not circular. To borrow a slogan
from Van Cleve (1979), to know real possibility I do not need to know (3a) (or (3b)),
I just need to fall under it (in particular, I need to satisfy the right-hand side). In
particular, to satisfy the cognition requirement with respect to some metaphysical
theory about noumena, I do not need to know the metaphysical theory of God that
underlies (3a) (or (3b)); I just need to satisfy it, in particular, the right-hand side of
(3a) (or (3b)).114
But a problem will arise if we think of (3a) or (3b) as providing a justification for
claims of knowledge of real possibility for noumena. By using (3a) or (3b), and their
right sides, to infer, for some concept C, that I know that C is really possibly

112
Kant’s objection is thus very similar to the problem of the ‘Cartesian circle.’ My argument in the rest
of this section is indebted to Van Cleve (1979), who argues, convincingly, that the problem of the Cartesian
circle is the problem of the criterion.
113
I have formulated both (3a) and (3b) as externalist principles on which an agent need not base her
judgment of possibility on the grounds that make her concept the concept of a possible object: divine
preformation/vision in God. Presumably, Crusius and Malebranche would both agree that agents had
knowledge of possibility before they ever offered their theories. This does not affect, I think, the point I go
on to make about epistemic principles.
114
Van Cleve (1979), 70.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

instantiated, I do not confer any justification on the conclusion of that logically valid
inference unless the premises themselves are justified. This becomes even more
problematic when we recognize that the right-hand sides of (3a) and (3b) are claims
about noumena (in particular, God); since knowledge of a theory of noumena
depends on antecedent cognition of the concepts involved, it would seem that we
must be in a position to know that <God> is really possibly instantiated before we can
be in a position to know (3a) or (3b) (or their right-hand sides).
Recall that the context of Kant’s objection is his attempt to explain how a priori
cognition (knowledge of the real possibility) of noumena is possible, a problem, he
claims, “in my long metaphysical labors, I, as well as others, had failed to consider”
(Ak. 10:131). In sensu stricto, though, this is clearly false, and Kant would have
known it to be false; we have just seen Crusius’s and Malebranche’s explanations of
how a priori cognition of noumena (the relation of intellectual concepts to their
objects, in the language of the Herz letter) is possible and the conditions under which
we actually have such cognition, (3a) and (3b). Kant’s claim to innovation is
strengthened in the CPR, however, when he distinguishes between the dogmatic
and the critical procedure in metaphysics: “dogmatism is therefore the dogmatic
procedure of pure reason, without an antecedent critique of its own capacity”
(Bxxxv). I take ‘antecedent’ to mean that the critique of our capacity of pure reason
will not presuppose any substantive metaphysical theory about noumena. It is not
enough to require of a ‘critique of pure reason’ that we refrain from claims to know a
metaphysical theory about noumena, for Crusius and Malebranche can use (3a) and
(3b) and the supporting metaphysics to generate knowledge of real noumenal
possibility without assuming that (3a) or (3b) are known to be true. Kant must be
setting up the project of a critique of pure reason as follows: suspend any positive
claims about noumena and explain our putative a priori cognition of them. And
within a project of that kind it is appropriate to exclude Crusius’s (3a) and Male-
branche’s (3b) not because they are ‘circular’ (they are not) but because they
presuppose a substantive metaphysical theory of noumena (in particular, of God).115
This vindicates Kant’s reasons for setting aside Crusius’s and Malebranche’s deus
ex machina theories within his own project: a transcendental critique into our faculty
of pure reason. But this does not undermine the coherence of Crusius’s or Male-
branche’s project, which in very general terms we might characterize as: starting with
prima facie plausible metaphysical and epistemic assumptions, find the best overall
theory of metaphysics and our epistemic access to it.116 What is more, it might cast

115
They are also circular if Crusius and Malebranche take (3a) and (3b) to be our source of justification
for claiming that the concepts that figure in those very principles (e.g., <God> and <cause>) are really
possibly instantiated.
116
Obviously, both Crusius and Malebranche have more specific philosophical/theological projects in
their writings, e.g., to defend libertarian freedom against the excessive rationalism of Leibniz and Wolff
(Crusius), or to perfect the Cartesian system and make it safe against atheism and skepticism
(Malebranche).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

REAL POSSIBILITY AND THE CRITICAL TURN 

some suspicion on the coherence of Kant’s own project. First of all, the negative result
of that critique—that there is no explanation of how a priori cognition of noumena is
possible—is unsurprising if we are required at the outset to bracket any positive
theory of noumena, for presumably any such explanation will consist in a theory of
noumena and how they are accessed by our minds (as Crusius’s and Malebranche’s
do). One might reasonably suspect that Kant has built the result of that critique
(noumenal ignorance) into its foundational assumptions. Secondly, there are reasons
to be suspicious whether Kant respects his own methodological scruples. Kant’s
transcendental theory of our cognitive capacity begins by assuming that we have a
sensible (receptive, non-spontaneous) form of intuition and distinguishing two
‘stems’ of cognition: sensibility and understanding.117 There are reasons to think
that, in knowing that sensibility and understanding are distinct faculties, we do not
merely know phenomena; after all, phenomena are possible objects of experience,
and experience itself depends upon our faculty of sensibility (intuitions) and under-
standing (concepts), so presumably those faculties themselves cannot be phenomena
(cannot be sensibly intuited). So there are grounds to wonder whether Kant’s own
answer to the transcendental question of how cognition a priori is possible does not,
in fact, begin with a (controversial) assumption about the non-phenomenal nou-
menal nature of our minds.
Fully adjudicating these issues would require addressing some of the thorniest
issues in the whole Critical philosophy, principal among them the epistemic status of
Kant’s own transcendental theory of experience, and whether it is compatible with
his denial of synthetic a priori knowledge of noumena.118 I do not intend to address
those issue here. I want to propose merely that the best way of understanding Kant’s
circularity objection to Crusius and Malebranche is as motivating a project in
transcendental philosophy (bracketing our positive metaphysical theories of nou-
mena, explain how a priori cognition of real possibility is even possible in the first
place), since it is far from clear that it can succeed as a non-question-begging
objection to their own metaphysical projects.
Ontology, for the logicists, was the science of all possible beings qua possible. In
Kant’s pre-Critical works he argued that logical possibility is not sufficient for
possibility, which he would later refer to as real possibility (to distinguish it from
mere logical possibility). However, his doctrine that all real possibilities are grounded
in God raised pressing questions in the metaphysics, epistemology, and cognitive
semantics of modality that he did not answer:

117
A16/B29–30. Kant’s concession that “they may perhaps raise from a common but to us unknown
root” does not, however, ameliorate the point; the question is what ground Kant could have for claiming
that sensibility and understanding are distinct faculties with distinct contents and distinct principles of
operation that “cannot exchange their function” (A51 f./B75 f.), if this is a claim about how they are in
themselves.
118
See e.g., Pereboom (1991).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 REAL POSSIBILITY AND THE CRITICAL TURN

(1) What is it to represent an object as really possible? That is, what is the content
of the representation of an object as really possible?
(2) What is it for an object to be really possible? What are the grounds of real
possibility?
(3) Of a really possible object we can represent as really possible, how can we
prove that it is really possible? And, what’s more, prove this a priori?
The outline of Kantian answers to these questions have now emerged, and they
depend upon the crucial distinction of all objects überhaupt into phenomena (objects
that can be given to us in sensible intuition) and noumena (objects that cannot be so
given):
(1) To represent a phenomenon as (formally) really possible is to represent it as
compatible with the formal intuitional (space and time) and intellectual (unity
of apperception) conditions of experience.
(2) For phenomena, the first grounds of (formal) real possibility are space, time,
and the unity of apperception.
(3) We prove the (formal) real possibility of phenomena a priori through a
transcendental critique of our capacity for cognition, which reveals the first
real grounds of the possibility of experience: space, time, and the unity of
apperception. We cannot prove a priori what is really possible for noumena;
consequently, a metaphysics of noumena is impossible for us.
This leaves open, however, whether there are notions of real possibility other than
formal real possibility (compatibility with the forms of experience); in the next two
chapters I will explore other Kantian notions of real possibility for phenomena. It
also leaves unanswered questions (1) and (2) about real possibility in relation to
noumena, and leaves open the possibility (3) that we may have some epistemic
warrant falling short of theoretical knowledge with respect to real possibility for
noumena. The final two chapters of this study concern Kant’s Critical theory of the
cognitive semantics, metaphysics, and epistemology of real possibility for noumena.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

7
Three Kinds of Real Possibility

7.1. Introduction
In the Critique of Pure Reason (CPR) Kant answers two questions that had arisen
originally, albeit implicitly, in Beweisgrund: how are real possibilities ‘given’ to the mind?
How can we know a priori what is really possible? However, Kant cannot retain his pre-
Critical theory of real possibility wholly unchanged within the Critical philosophy. Pre-
Critical real possibility concerns what is possible for all objects überhaupt, including
objects that cannot be sensibly intuited (noumena). Given Kant’s Critical doctrine that
we cannot cognize noumena and hence cannot cognize all objects überhaupt, his pre-
Critical modal metaphysics cannot constitute an object of knowledge (Wissen); cognition
of real possibility for noumena is impossible. The domain of real possibility shifts in the
Critical period to real possibility we can cognize: real possibility for phenomena.
In this chapter I begin to explore Kant’s Critical theory of real possibility, although,
in one form or another, it will occupy us for the rest of this study. In this and
subsequent chapters I distinguish various different kinds of real possibility in the
Critical system, but in section 2 I begin by taking a step back and considering what all
of these notions have in common that make them kinds of real possibility in the
Critical system. In the remainder of the chapter I focus on three different kinds of real
possibility for phenomena, which I call formal real possibility (section 3), empirical-
causal real possibility (section 4), and noumenal-causal real possibility (section 5).

7.2. Critical Real Possibility

Recall that Kant’s pre-Critical theory of real possibility had these elements:1
(Non-Logical) Real possibility is distinct from logical possibility. That some prop-
osition2 p is logically consistent (¬p does not entail a contradiction) does not entail

1
In Beweisgrund Kant focused on the possibility of predicates; I have formulated these claims in terms of the
possibility of propositions being true, to accommodate the wider range of things whose possibility Kant will
investigate in the Critical period: the possibility of experience, the possibility of objects in experience, the
possibility of those objects undergoing change, etc. Kant typically defines possibility in terms of the possibility
of an object—see his metaphysics lectures (MV, Ak. 28: 410; ML2, Ak. 28: 543; MM, Ak. 29: 81; MK3, Ak. 29: 960).
These are interconvertible definitions: an object is possible just in case the proposition that there is such an object
is possible. Likewise, a concept C is possible just in case the proposition that the concept is instantiated is possible.
2
By ‘proposition’ I just mean the content of a judgment; I introduce this terminology to make clear I am
not talking about the modal status of token acts of judging. Kant himself, however, defines the term
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

that p is really possible. So logical possibility and real possibility are at least
conceptually distinct; it is not a conceptual truth that if ◇L p then ◇Rp (equiva-
lently, it is not a conceptual truth that if □Rp then □Lp).
(Ground) Real possibilities have real grounds in actuality. If it is really possible
that p then there is some actual object or principle that grounds the fact that it
is really possible that p. The relation between grounds of real possibility and the
real possibilities they ground is not a logical relation; if x grounds the real
possibility that p, this is not because of the conceptual containment of p within
the concept of x.
(Worldly) Real possibility is a form of metaphysical, or world-based, possibility.3
Real possibilities in general are grounded in facts about the world that do not
depend upon how our minds are constituted, or how we experience or conceptu-
alize the world.4
In the Critical philosophy Kant deploys several different kinds of real possibility
without explicitly distinguishing them; in this chapter I will distinguish three of
them and, in the next chapter, a fourth. What makes all of these different kinds of
possibility kinds of real possibility is that they obey Non-Logicality and Groundedness
(with modifications to be noted below). However, some of these kinds of real possibility
violate Worldliness; Kant grounds some kinds of real possibility in our cognitive
faculties and in the causal powers of phenomena,5 rather than in objects that do not
depends upon a discursive intellect, noumena.6 For now this will constitute an inter-
pretive hypothesis; it will take me the rest of this study to argue for it fully.
This following is a provisional characterization of the concept of real possibility in
the Critical philosophy:
(Real Possibility) For any kind of possibility ◇xp (and its associated kind of
necessity □xp, where □xp $ ¬◇x¬p), ◇x is a kind of real possibility (and □xp is a
kind of real necessity) only if

‘proposition’ [Satz] to refer to an assertoric rather than merely problematic judgment; see Disc. (Ak. 8: 193 n.)
and Kant’s handwritten note at A74 (E XXXVIII, Ak. 23: 25).
3
This is not strictly true for all notions of real possibility in all of Kant’s pre-Critical writings. In ID
Kant already holds that objects in space and time are phenomena whose possibility depends on our
spatiotemporal forms of intuition. So, while the pre-Critical writings do contain some notion of real
possibility for non-noumenal objects, the dominant strand in Kant’s pre-Critical theorizing about modality
is noumenal.
4
Another element of Kant’s pre-Critical modal metaphysics (GARP—see Ch. 5.7) was the doctrine that
there is a unique being that grounds all real possibilities and exists with absolute necessity. I discuss the fate
of this doctrine in the Critical period in Ch. 9.
5
Phenomena are mind-dependent in this sense: their possibility is grounded in the possibility of our
experience of them. See A111 and Ch. 6.7.
6
The role of God is maintained in a particular kind of real possibility, which I will call ‘noumenal real
possibility’; see Ch. 9 for more on noumenal possibility.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

(i) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇xp (equivalently, it is

not a conceptual truth that □xp ⊃ □Lp), and
(ii) Groundedness: if ◇xp then the fact that ◇xp has a real ground in some actual
object or principle.
Every aspect of this definition requires further philosophical analysis, but let me
begin by pointing out that clause (i) means that every concept of real possibility (real
possibilityx) is a different concept than the concept of logical possibility. This does not
entail that they have different extensions (it does not entail that there are logical
possibilities that are not real possibilities of the relevant kind), only that they might.7
In some cases, though, we know a priori they do have different extensions. For
instance, we know a priori that it is logically possible that a triangle have internal
angles that sum to less than 180 degrees but this is not really possible. While there are
kinds of real possibility that are extensionally distinct from logical possibility,
extensional distinctness is not a conceptual requirement of real possibility.
The other aspect of this definition of real possibility that stands in need of
clarification is the notion of ‘ground’ involved in the second clause. The question
that I raised at the beginning of this chapter—what is the relation between pre-
Critical real possibility and Critical real possibility?—has its counterpart in the
question, what is the relation between Kant’s pre-Critical conception of real grounds,
and his Critical conception of real grounds? Fully answering that question would
require a book of its own. Here I can only sketch an answer. The most important
aspect of Kant’s pre-Critical conception of real grounds is retained within the Critical
system, namely:
(Non-Logicality) If x is the real ground of y then the concept of x does not
contain the concept of y, and the judgment that x exists does not logically entail the
judgment that y exists.8
In other words, real grounding relations do not obtain in virtue of relations of logical
entailment or conceptual containment. Kant’s Critical system involves at least two
very different kinds of real grounds: formal grounds (e.g., triangularity grounds the
possession of internal angles that sum to two right angles)9 and efficient causal
grounds (e.g., the sun shining on the stone grounds the stone’s warming). Both of
these kinds of real ground obey the Non-Logicality requirement. For each kind of
real possibility I distinguish below, I will also explain the grounds of this kind of real
possibility, as well as the nature of the relevant non-logical grounding relation.

7
Because I am treating possibility and necessity as modal operators on judgments, rather than
predicates of objects, their extension is the class of judgments to which they apply, rather than a class of
objects; e.g., the extension of <possible> is the set of judgments which are possibly the case.
8
In the metaphysics lectures, see MM (Ak. 29: 807), ML2 (Ak. 28: 549), MV (Ak. 28: 402), and MvS (Ak.
28: 486).
9
What Crusius would call ‘real existential grounds.’ See Ch. 3 and Ent. }79 (p. 136), as well as }3 of this
chapter.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

Whereas Kant’s pre-Critical view required that real possibilities be grounded in an

actually existing substance, his Critical view broadens the candidates for real grounds
of possibility to include grounds that are not substances (e.g., the forms of experience
themselves) and, indeed, in some cases are not even objects at all (e.g., the unity of
apperception). Precisely articulating the ‘actuality’ requirement on grounds would
require investigating the details of Kant’s Critical theory of actuality [Wirklichkeit]
and existence [Dasein], which I postpone until Chapter 10; in the meantime, though,
I will require merely that some actual object or principle grounds real possibilities:
the actual forms of experience, the actual natural laws, the actual past, etc.
This introduces an important distinction, however, among the kinds of real possi-
bility. We are concerned in this chapter with real possibilities for phenomena, but
phenomena have two kinds of grounds in Kant’s Critical system: ‘immanent’ ones and
‘transcendent’ ones. By ‘immanent’ grounds I mean grounds that are either them-
selves phenomena or are subjective conditions of phenomena, such as the forms of
experience. Phenomena have immanent grounds insofar they are conditioned (caus-
ally) by other phenomena and (non-causally) by subjective forms of experience. By
‘transcendent’ grounds I mean grounds that are not immanent to experience, namely,
noumena. Empirical objects have noumenal grounds in that they are the appearances
of those noumena; noumena causally affect subjects’ minds, giving rise to the sensory
matter (but not the form) of experience.10 Of the kinds of real possibility I distinguish
in this chapter, two have immanent grounds (formal possibility, empirical-causal
possibility) and one (noumenal-causal possibility) has a noumenal ground.
Figure 7.1 represents the distinctions I have sketched so far and indicates provi-
sionally where these different kinds of real possibility fit. In the penultimate chapter
of this study I will expand our focus to include kinds of real possibility for noumena
as well. The different kinds of real possibility are distinguished both by the range of
propositions to which they apply, as well as by the kinds of grounds they have.

7.3. Formal Possibility

In the previous chapter we explored Kant’s answer to the question, how can we know
a priori what is really possible? His answer to this question, on my reconstruction,
invoked a notion of possibility that I there dubbed ‘formal’ possibility. My discussion
of that concept was, to borrow a Kantian trope, a reconstruction of it ‘from above,’
from the philosophical problem it is supposed to solve. In this section, to continue
the Kantian trope, I will reconstruct it ‘from below,’ from the basic elements of Kant’s
theory of experience, and show that the two reconstructions arrive at the same point.
In the second half of the section I explore the notion of grounding involved in formal
possibility.

10
For noumena as grounds of the matter of phenomena see A190/B235, A387, A494/B522, A614/B642,
Prol. (Ak. 4: 289, 314, 318), CPrR (Ak. 4: 451), and esp. Disc. (Ak. 8: 215).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

Concepts of Possibility

Logical Possibility Real Possibility

Non-Logicality
Groundedness

Real Possibility for Phenomena

Real Possibility for Noumena
(Chapter 9)

Where the actual ground of real possibility is Where the actual ground of real
‘immanent’: possibility is ‘transcendent’:
Formal possibility (Section 3, Chapter Noumenal-causal possibility
6.7) (Section 5)
Empirical-causal possibility (Section 4)
Nomic possibility (Chapter 8)

Fig. 7.1. Kinds of real possibility in Kant’s Critical system

Kant defines experience at B147 as “empirical cognition”: thinking a sensibly given

object under a concept.11 This is a ‘semantic’ characterization of experience in terms of
the kind of representation it is and the kind of content it has. Experience is also,
according to Kant, the joint product of affection by mind-independent objects—
noumena or ‘things in themselves’—and processing by the subject’s mind (synthesis).12
This characterizes experience in terms of its causal origin. Noumena affect the subject’s
mind, creating low-level representations, sensations.13 The subject’s cognitive faculties
process these sensations by synthesizing them into representations of spatiotemporal
objects, experience.14
Some of the representations generated by this cognitive process depend upon the
particular sensations had by the subject. For instance, the determinate shape and size
of my desk, as represented in my intuition of it, depends upon the sensations I have.
Other representations do not depend in this way upon the particular sensory content.

11
“To think of an object and to cognize an object are thus not the same. For two components belong to
cognition; first, the concept through which an object is thought at all (the category), and second, the
intuition, through which it is given [ . . . ] [the categories] serve only for the possibility of empirical
cognition. This, however, is called experience [ . . . ]” (B146–8). This is a quite minimal notion of experi-
ence. In this section I build up from this minimal sense of experience to a stronger sense (the sense in which
Kant usually uses the term Erfahrung): a unity of perceptions that are connected according to laws. See
further discussion in text.
12
B1.
13
Kant asserts the noumenal affection of our sensibility by things in themselves at A190/B235, A387,
A494/B522, Prol. (Ak. 4: 289, 314, 318), GMM (Ak. 4: 451), and esp. Disc. (Ak. 8: 215).
14
B1–2.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

Regardless of the sensations I have, these representations will have the same content.
A representation is said to be a priori just in case its content does not depend upon
the particular sensations had by the subject.15 An a posteriori representation is a
representation whose content does depend upon particular sensations. For instance,
regardless of the sensory content I receive, I intuit outer objects (i.e. objects that are
not my inner psychological states) in space. Similarly, regardless of the sensory
content I receive, I intuit space as Euclidean, according to Kant. While the deter-
minate spatial properties of objects depend upon the particular sensory content
I receive and are therefore a posteriori, the representation of space and its Euclidean
character are a priori.16 The forms of experience (forms of intuition and forms of
understanding) are a priori representations: determinable intuitional and conceptual
structures of which any particular experience is a determinate filling-out.17 The
Euclidean structure of space is an a priori form of intuition; it is shared by all
intuitions of outer objects, and it does not depend upon the particular course of
my experience.18
Experiencing an object, by definition, requires thinking of it under a concept and,
Kant argues, this requires thinking of objects under the categories. They are a priori
concepts, because they provide a determinable conceptual structure for all experience
of objects. For instance, the category <substance–accident> requires that we experi-
ence each object as either a substance or a modification of a substance, and the
category <cause–effect> requires that we experience alterations in substances
(changes in their accidents) as governed by deterministic causal laws. The conceptual
forms of experience (categories) do not dictate which objects we experience as
substances, and which as modifications, nor which causal laws govern objects,19
just as the intuitional forms (space, time) do not dictate the determinate sizes, shapes,
locations, and durations of objects. Each particular experience is a combination of

15
This conception of the a priori is equivalent to what Kitcher calls ‘a prioriO’ in Kitcher (1993), 15–16.
16
This concept of a priority is not defined in terms of our experience-independent justification for
making a judgment about objects (e.g., that they are in space), nor in terms of deriving something from its
ground. On this conception of the a priori, see Disc. (Ak. 8: 221) where Kant distinguishes his conception of
the a priori from innatism. A priori representations are not innate, for they require experience for their
activation. What is innate is the capacity to represent objects as, for instance, spatial, regardless of the
course of experience. See also Kant’s discussion of the origin of the categories at B167. Cf. the discussion of
both passages in Longuenesse (1997), 221.
17
See A50–1/B75: “Thus pure intuition contains merely the form under which something is intuited,
and the pure concept only the form of thinking in general. Only pure intuitions or concepts alone are
possible a priori, empirical ones only a posteriori.”
18
The representation of space is determinable in the sense that it does not determine the spatial
properties of particular objects. However, the structure of space is determinate. My a priori representation
of space represents space as Euclidean. Thus, in Kantian terms, our form of intuition of outer objects is
Euclidean. Similar remarks apply for time: the form of time dictates the structure of time, but not the
determinate temporal properties of objects.
19
Kant claims that which particular causal laws govern empirical objects is not wholly determined by
the forms of experience (concepts and intuitions) alone at A125, B167, and in the CJ (Ak. 5: 180 and 184).
See Ch. 8 for more discussion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

determinable a priori intuitional and conceptual forms (space, time, the categories)
and more determinate a posteriori content or ‘matter’ (size, shape, location, particu-
lar causal laws).
The idea that experience has a priori forms corresponds naturally to the concept of
what is necessary because of the forms (formally necessary), or, more precisely:
(Formal necessity) It is formally necessary that p if and only if the fact that p is
wholly grounded in facts about the actual intuitional form (space and time) and
conceptual form (categories) of experience
where ‘p’ ranges over synthetic propositions about phenomena.20 The word ‘actual’ is
there to remind us that it is at least conceivable that there could be discursive
cognizers with different intuitional forms than ours; formal possibility is defined
in terms of our actual forms of intuition, space and time, not the idea of forms
of intuition in general (or the requirement that there be some form of intuition or
other). This definition of formal necessity, according to the interdefinability of
possibility and necessity, corresponds to the following definition of formal
possibility:
(Formal possibility) It is formally possible that p if and only if it is not the case
that facts about the actual intuitional form (space and time) and conceptual form
(categories) of experience wholly ground the fact that ¬p.
For instance, it is formally possible that we experience unicorns; the fact that we do
not experience such objects is partly grounded in how we are affected.
Kant’s standard formula for expressing the (formal) necessity of space, time, the
categories, the principles of experience, etc. is to say that they are ‘conditions’
[Bedingungen] or ‘grounds’ [Gründe] of the possibility of experience,21 and readers
may wonder about the relation of that famous Kantian formula to my definitions of
formal necessity and possibility. Before answering that question directly, I want to
make two observations about this formula. First, it is not clear in what sense of
‘possibility’ these forms are conditions of possibility; they are not, for instance,
conditions of the logical possibility of experience. Nor is it sufficient to say that
they are conditions of the formal possibility of experience, for Kant defines what I am
calling ‘formal’ possibility in terms of the forms of experience themselves: “whatever
agrees with the formal conditions of experience (in accordance with intuition and
concepts) is possible” (A218/B265 f.). Secondly, not all principles that are necessary
in the relevant sense are conditions of the possibility of experience, for some of them
are consequences of those conditions. For instance, no possible object of experience
can violate a theorem of geometry, but it is implausible that every theorem is a

20
Since in this chapter I am focusing only on kinds of modality that apply to phenomena, I will not
always make this restriction explicit.
21
E.g. A94/B126, A95, B161, A158/B197, and A237/B296; cf. Prol. (Ak. 4: 297, 319, 351).
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

condition of experience. The basic Euclidean structure of space is a condition of the

possibility of experience, while a theorem is a necessary consequence of that condi-
tion; both apply to every possible object of experience, but this is because of the
Euclidean structure of space.
However, I do not think that the necessity of the forms of experience is defined in
terms of some antecedently intelligible notion of possibility (of which they are
conditions); I take these forms to be constitutive of the relevant notion of (formal)
possibility and (formal) necessity. When Kant says that the forms are ‘conditions’ or
‘grounds’ of the possibility of experience he means: they are the grounds that are
constitutive of the formal possibility of experience in the first place. With respect to
the second issue, my definition of formal necessity has built into it a distinction
between the forms of experience (which ground formal necessity and possibility) and
the consequences of those forms. So we can say that the basic structure of space is a
ground of the formal possibility of outer objects, the theorems of geometry are
grounded in that structure, and both are equally formally necessary (they apply to
all formally possible objects). This repeats a familiar structure from Kant’s pre-
Critical theory of absolute necessity (see Chapter 5.2): there is a hyperintensional
relation of grounding among the formal necessities.
In some texts Kant talks about formal possibility in terms of compatibility with our
forms of experience, for instance, “whatever agrees with [mit . . . übereinkommt] the
formal conditions of experience (in accordance with concepts and intuitions) is
possible” (A218/B265 f.).22 I take ‘agreement with’ the forms of experience to
mean ‘compatibility with’ the forms of experience. If A218/B265 f. and other passages
that characterize possibility in terms of compatibility with the forms of experience are
discussing what I am calling formal (real) possibility, this raises the question: what is

22
Abaci (2013) and Chignell (2014b) dissent from this reading of the definition of possibility at A218/
B265 f. In the body of the Postulates themselves Kant discusses a series of concepts “the possibility of which
is entirely groundless, because it cannot be grounded in experience and its known laws” (A223/B270),
namely: a “substance persistently present in space yet without filling it [ . . . ], or a special power of our mind
to intuit the future (not merely, say, to deduce it), or, finally, a capacity of our mind to stand in community
with the thoughts of other people (no matter how distant they may be)” (A222–3/B270; cf. A290–1/B347).
Abaci and Chignell argue that Kant does not mean that the assertion that these objects are formally
possible is groundless, but that we have no reason to think they are actually instantiated (Abaci) or
compatible with our background knowledge of natural laws (Chignell). In each case, though, there is a
serious question as to whether these are concepts of objects that can be given in experience. To take the first
case, Kant argues in MFNS that matter present in space without filling it through dynamic force is
incompatible with our forms of experience (Ak. 4: 496–7). Likewise, in the second case, it is questionable
whether intuition of the future is compatible with the forms of experience, given that our sensible faculty
must be affected by its object, and ‘backwards causation’ is ruled out by the Second and Third Analogy.
In the third case, it is questionable whether the unity of apperception allows me to attach the ‘I think’
to a thought of another person; arguably, what I can attach the ‘I think’ to determines the range of what
thoughts are mine, so apperceptive access to the thoughts of another person is impossible (they would
thereby be my thoughts as well). In each case there are reasons for Kant to assert that the possibility of these
objects is “entirely groundless because it is not grounded in experience and its known a priori formal laws.”
See also A144/B184, MvS (Ak. 28: 494), ML2 (Ak. 28: 556), MM (Ak. 29: 815, 821 f.), Refl. 4081 (Ak. 17:
732), and Refl. 6360 (Ak. 18: 688).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

the relevant relation of compatibility? The definition in terms of compatibility is

equivalent to the definition in terms of grounding if the compatibility relation is
defined as follows:
(Defn.) p is compatible with our forms of experience if and only if it is not the case
that facts about our forms of experience wholly ground the fact that ¬p.
This expresses an intuitive notion of compatibility; for instance, it is compatible with
the forms of experience that I experience motions that violate Newton’s laws just in
case the forms of experience alone do not make it the case that I do not experience
such objects. My definitions of formal possibility and necessity in terms of grounding
explain and unify two standard Kantian formulas for necessity and possibility,
respectively: “conditions of the possibility of experience” (necessity) and “agreement
(compatibility) with the forms of experience” (possibility).23 I take this to be indirect
evidence of their correctness.
Experience is thinking of sensibly given objects under concepts, and thus has two
kinds of conditions: sensible conditions and intellectual conditions. In the intellectual
case, I have focused on the categories, but Kant’s ultimate view is that the categories
are derivative of a more fundamental intellectual condition, the unity of appercep-
tion.24 I take this to mean that any intellectual ground of the possibility of experience
is ultimately grounded in the unity of apperception. For instance, the categories are
intellectual grounds of experience of objects because they are grounded in the unity
of apperception. So we can reformulate our definition of possibility as follows:
(Formal possibility) It is formally possible that p if and only if it is not the case
that facts about the actual intuitional form (space and time) and intellectual form
(unity of apperception) wholly ground the fact that ¬p.
In the previous chapter I reconstructed ‘from above’ a notion of formal possibility:
(Formal ) It is formally possible that we represent a certain (conceptual or
intuitional) content just in case it is compatible with the forms of intuition
(space and time) and the highest principle of the understanding (the synthetic
unity of apperception) that we represent a sensibly given object using that (con-
ceptual or intuitional) content.25

23
I depart here from Robert Hanna, who attributes to Kant the familiar ‘possible worlds’ analysis of
possibility and necessity, and then defines what an ‘experientially possible world’ is in terms of the
conceptual and intuitional form of experience (Hanna (2001), 239–45). However, an experientially possible
world is going to be one that does not have features that are incompatible (in the sense defined above) with
our forms of experience, so we can define formal possibility directly in terms of those forms of experience,
and skip the intermediate step of constructing a set of possible worlds for the modal operators-cum-
quantifiers to range over. Furthermore, a possible worlds analysis will not help in accounting for the
hyperintensional grounding relation among formal necessities (see Ch. 5.2), because, by definition, both
relata of the grounding relation will be true in all ‘experientially possible worlds.’
24
A111, B136. See Ch. 6.6.
25
See Ch. 6.7.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

If we understand compatibility here in terms of grounding, via the definition of

compatibility above, I think it is clear that this notion of formal possibility is
subsumed by the more general definition of formal possibility I have just constructed
‘from below’ (that we represent a certain content is just a value of p).
Since all objects of possible experience are appearances (since all sensibly given
objects are appearances),26 if F is a content that figures in the content of experience
(e.g., the property of being a substance), we can infer from claims of the form:
(1) It is formally possible that we experience objects as being F
to conclusions of the form:
(2) ∴ It is formally possible that objects of experience are F.
This inference would be invalid if the objects of experience were things in them-
selves (noumena), rather than appearances (phenomena). This entailment holds
because facts of the form of (2) are grounded in facts of the form of (1); if it is
formally possible for objects to have property F, this is in virtue of the fact that it
is formally possible to experience objects as having property F. Since the forms of
experience (space and time, the unity of apperception) are the grounds of formal
possibility, they are the grounds of facts of the form of (1). It follows that the forms
of experience are grounds of the formal possibility of objects of experience (phenom-
ena) themselves, as Kant states in the A Deduction:
The a priori conditions of a possible experience in general are at the same time conditions of the
possibility of the objects of experience. Now I assert that the categories that have just been
adduced are nothing other than the conditions of thinking in a possible experience, just as space
and time contain the conditions of the intuition for the very same thing. They are therefore
fundamental concepts for thinking objects in general for the appearances, and they therefore
have a priori objective validity, which was just what we really wanted to know. (A111)27

This means that the forms of experience (Kant here focuses on the categories, but the
same point applies to space and time as well) apply to all possible objects of experience.
The forms of experience alone make it the case that all objects of experience obey these
forms. It follows, by definition, that it is formally necessary that the objects of
experience obey the forms of experience (e.g., they instantiate the categories).
It is clear that formal possibility (and necessity) satisfy the two conditions on real
possibility (and necessity) from section 2:
(Real PossibilityF) Formal possibility ◇Fp (and its associated kind of necessity
□Fp, where □Fp $ ¬◇F¬p) is a kind of real possibility (and □Fp is a kind of real
necessity)28 only if

26 27
A30/B45. Cf. A158/B197.
28
Abaci (2013, 25 n. 8) objects to my assertion that Kant has a notion of formal necessity, not made
explicit in the Postulates itself, but derived from formal possibility in the standard fashion. While I admit
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

(i) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇Fp (equivalently, it is

not a conceptual truth that □Fp ⊃ □Lp), and
(ii) Groundedness: if ◇Fp then the fact that ◇Fp has a real ground in some
actual object or principle.
Formal possibility is not conceptually equivalent to logical possibility because it is not
even extensionally equivalent: there are logically possibilities (e.g., that we experience
non-spatial outer objects) that are not formally possible. Formal possibility is
grounded in actuality because, if it is formally possible that p, then this is so in virtue
of a fact about our actual forms of experience (they do not make it the case that ¬p).
If we, per impossibile formaliter, had different forms of intuition, the facts about what
is formally possible might be different. It is not grounded in an actual object, because
the forms of intuition, much less the unity of apperception, are not objects.
Readers might wonder what notion of ‘ground’ is involved in the definition of
formal possibility and necessity. It should be clear that formal necessity and possi-
bility cannot be defined in terms of logical grounding (entailment). The relation
between the axioms and theorems of geometry parallels the relation between the
forms of outer experience and the formally necessary principles of experience that
follow from them. If we analyzed formal possibility and necessity in terms of logical
grounding, it would follow that geometrical axioms logically entail geometrical
theorems, i.e. that all formally impossible propositions are logically incompatible
with the axioms. But Kant holds that geometric reasoning is irreducibly synthetic
because it relies on construction in pure intuition.29 Consequently, the relation
between forms of experience and what is formally necessary cannot be assimilated
to the relation of logical entailment; the formal necessities do not follow logically
from the forms of experience. Consequently, formal possibility satisfies the require-
ment that kinds of real possibility have real grounds in actuality.
What, then, is the nature of this real grounding relation? It is very hard to extract a
fully satisfying answer to that question from Kant’s writings. However, certain
broad features of Kant’s Critical theory of real grounds of possibility can be noted.
As noted in section 2, the Critical Kant continues to distinguish logical from real
grounds,30 of which causal grounds are a species.31 Further, he follows Wolff,
Baumgarten, and Crusius32 in distinguishing between grounds of becoming (rationes

that Kant could have been more explicit about making all of these modal distinctions, he clearly needs a
notion of formal necessity to make sense of the modal status of the a priori formal principles of experience
themselves. This cannot be empirical necessity (see below) because empirical necessity only applies to
alterations in substances and is only knowable a posterioriJ.
29
For more on Kant’s views on construction and the irreducibly synthetic character of geometric
reasoning, see Friedman (1992a), 80–7, Parsons (1992), 78–9, and Shabel (2007), 97–107.
30
MM (Ak. 29: 807), ML2 (Ak. 28: 549), MV (Ak. 28: 402), and MvS (Ak. 28: 486).
31
MV (Ak. 28: 403)and MvS (Ak. 28: 486).
32
Ont. } 874, Meta. } 311, and Ent. } 34. Thanks to Smit (2009), 200–2 for drawing my attention to this
important distinction.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

fiendi) and grounds of being or possibility (rationes essendi).33 A ground of becoming

(ratio fiendi), as its name suggests, is the ground of a change, either an alteration in,
or the generation or corruption of, a substance (its arising or perishing). Kant defines
a ground of being (ratio essendi) as follows: “the ratio essendi is the ground of that
which pertains [zukömmt] to a thing considered according to its possibility” (MM,
Ak. 29: 809). The ratio essendi of a being is what makes it possible, while its ratio
fiendi accounts for its actuality. The forms of experience are real rationes essendi of
objects of possible experience but this does not tell us what this real grounding
relation is.
While I do not think we can find a reductive analysis of the real grounding relation
between the forms of experience and their consequences, I think we can make some
progress towards determining the structure of this relation. I will begin by focusing
on Kant’s example of a ratio essendi in the Mrongovius lectures: “the three sides
[Seiten] in the triangle are the ground of the three corners [Winkel]” (Ak. 29: 809).34
The example is potentially confusing because we refer to triangles as
‘triangles’ [Dreiecke] but we typically define them as three-sided [dreiseitig] plane
figures. Kant is assuming (not implausibly, I think) that their trilaterality is more
fundamental than their triangularity; triangles are triangular because they are trilat-
eral. This is not a logical grounding relation (that T is trilateral does not logically
entail that it is triangular), nor is it a causal relation. Furthermore, the relation is
asymmetric: trilaterality grounds triangularity, and not vice versa. Nonetheless, the
two properties are (formally) necessarily co-instantiated: necessarily, a figure is
trilateral if and only if it is triangular.35 So the relevant grounding relation is a

33
ND (Ak. 1: 392), MH (Ak. 28: 37, 54), MM (29: 809, 810), ML2 (Ak. 28: 571), MD (Ak. 28: 648), MK2
(Ak. 28: 802), and Refl. 5182 (Ak. 18: 111). Cf. Kant’s claim in the CPrR that “freedom is the ratio essendi of
the moral law” (Ak. 5: 14). Strictly speaking, Kant’s example in the quoted passage is one in which the
essence of a possible thing is the ground of that very (numerically identical) thing’s properties; in the rest of
the section I am applying that notion to cases where one thing is the ground of the possibility of a
numerically distinct thing (e.g., the form of experience and an object of experience). But we have seen that
Kant’s system requires a real grounding relation between a ground and the possibility of a numerically
distinct thing; this, I take it, is sufficient reason to generalize the notion of ratio essendi beyond the case of
grounding of properties ‘within’ one individual (e.g., a triangle).
34
The MM transcript does not use the term ‘Dreieck’ but merely a drawing of a triangle. This example is
sufficiently similar to Crusius’s example of a real-existential ground (the three sides of a triangle and their
relation ground the size of its angles—Ent. }36) that it is quite likely that Kant is borrowing both the idea of
an inefficacious ground of possibility, and the example, from Crusius (see Watkins (2005), 83–4). However,
Crusius’s discussion of real-existential grounds in Ent. }36 does not cast much light on what exactly this
relation is; his discussion in }79 is even less helpful for understanding Kant, for his examples are concrete
beings (a wedge and a lever). I think it is also descended from the notion of a ‘formal cause’ in Descartes,
which he takes from Aristotle (see CSM II: 169/AT VII: 238–42 where he cites Posterior Analytics, bk. II,
ch. 11).
35
In his lectures on metaphysics Kant attempts to reconcile the asymmetry of grounding with the fact
that the positing of grounds and consequences mutually (though not logically) entail one another: if you
posit a ground you posit a consequence, and if you posit a consequence you posit a ground (MV, Ak. 28:
486; MM, Ak. 29: 808). Kant’s solution is to say that the positing of a consequence does not involve positing
a determinate or specific ground but merely some ground in general (MV, Ak. 28: 408; ML2, Ak. 28: 549;
MD, Ak. 28: 625, 628), but all of his cases involve causal preemption. It stands to reason, then, in the case of
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

(i) non-causal, (ii) non-logical, (iii) asymmetric relation between (iv) (formally)
necessarily co-instantiated properties. If we were to attempt to analyze this notion
we might say that trilaterality grounds triangularity because we can prove the latter
from the former; however, we can also prove the former from the latter, violating
the asymmetry condition. We might supplement this with the requirement that the
ground (e.g., trilaterality) explains the consequence (e.g., triangularity) and not vice
versa. But this is just to appeal to an unanalyzed notion of one fact asymmetrically
(non-logically, non-causally) explaining another, which is the notion of grounding
in another guise. Clearly, a reductive analysis will be hard to come by, and I will
not further pursue one here.36
In keeping with the Critical shift away from ontology (which studies the highest
grounds of all possible beings) to transcendental philosophy (which studies the
highest grounds of possible objects of representation), Kant internalizes the ground-
ing relation within experience. As I will argue in the rest of this section, there are
really two basic grounding relations involved in Kant’s definition of the formal
modalities: first, a grounding relation between the forms of experience and its
contents (which explains the formal possibility of representing objects as having
certain properties), and secondly, a grounding relation between those contents and
the properties of objects (which explains the formal possibility of objects having those
properties). I will refer to these as form-content grounding and content-property
grounding, respectively, and will discuss them in that order. There will also be
various derivative relations of real grounding, for instance, causal grounding between
phenomenal substances and their accidents; these relations are derivative, because
the obtaining of them depends upon the grounding relation between experience and
its objects (content-property).
The first grounding relation holds between the form of experience and the content
of experience. I will start with the case of (specifically outer) intuition. Our spatial
form of outer intuition grounds the fact that any outer intuition has spatial content
(represents its object in space).37 This grounding relation has the same structure from
earlier: our spatial form of intuition (i) non-causally, (ii) non-logically makes it the
case that objects of outer intuition are represented in space, and this relation is (iii)
asymmetric because the spatial content of intuition does not explain why we have the
form of intuition we do. Finally, (iv) both our form of intuition and the space in
which we intuit objects are formally necessary. But this grounding relation is not

non-causal grounding that the positing of the ground and the consequence mutually necessarily entail one
another.
36
Asymmetric non-logical, non-causal explanatory relations between mutually necessarily entailing
truths (e.g., that Socrates exists and that {Socrates} exists) have recently become a major topic of
metaphysical research, and even the eighteenth-century name for this relation has been retained: ground-
ing. See the essays in Correia and Schnieder (2012), as well as Bliss and Trogdon (2014) and Raven
(forthcoming), for a survey of the recent discussions.
37
A26/B42, A42/B59, A48/B65.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

limited to the contents of intuition; objects of (formally) possible intuition have the
spatial properties they are (formally) necessarily intuited as having. So the next
grounding relation holds between the contents of intuition and the properties of the
objects of intuition; objects of (formally) possible intuition have spatial properties
because of the content of intuition (which is itself grounded in the form of intuition).
The same structure repeats itself: the content of possible intuition is a (i) non-logical,
(ii) non-causal, (iii) asymmetric ground of the possible spatial properties of objects,
even though (iv) both are formally necessary.
This applies at the level of the pure form of space (and time) and objects that can
be presented (constructed) in pure intuition, but we want to understand the structure
of the grounding relation within experience generally. To bring out this further
structure it will help to think about the structure of experience itself. The most
minimal level of experience,38 according to Kant, is merely intuiting and conceptu-
alizing sensory manifolds, without relating those manifolds to intersubjectively
accessible ‘public’ objects.39 I will refer to this as perception40 to distinguish it from
more robust conceptions of experience (see later discussion). Perception of sensory
manifolds requires thinking them under the mathematical categories of quantity and
quality, but not (yet) the categories of relation (substance, cause, community).
Thinking them under the categories of quantity requires representing them as
spatiotemporal wholes composed of parts (extensive magnitudes),41 while thinking
them under the categories of quality requires representing their sensory qualities as
continuously gradable (intensive magnitudes).42 This means that the grounding
relations that obtain between space and objects of pure intuition holds for perception
and its sensory manifolds: the forms of experience (space and time, categories of
quantity and quality) ground the fact that they are perceived as extensive and
intensive magnitudes, and this content grounds the fact that they have the properties
of extensive and intensive magnitudes. The first relation holds between the form of
experience (in this case, the forms operative in perception) and its content, and the

38
There is a more minimal notion of (s-)cognition available: thinking about the pure manifolds of space
and time. It would be inappropriate to call this experience because it is a priori. In the main body of the text
I start with perception of sensory manifolds.
39
This is how I interpret Kant’s distinction between ‘mathematical’ (quantity, quality) and ‘dynamical’
(relation, modality) categories (B110, A160/B199, A176/B218): the former are involved in any cognition
(thinking an intuited object under a concept) of sensory manifolds as such, while the latter are required to
think about sensory manifolds as ‘appearances’ (in the empirical sense—A30/B45, A45–6/B63) of objects
in an intersubjective space that do not depend upon perceptions of them (the ‘empirical thing in itself ’—
see previous passages). To cognize sensory manifolds as such we need to think them under categories of
quantity (which brings them under the Axioms of Intuition) and quality (which brings them under the
Anticipations of Perception); to cognize sensory manifolds as appearances of intersubjective objects we
need to think them under categories of relation (which brings them under the Analogies of Experience).
My thinking about these issues has benefited greatly from conversations with Clinton Tolley and from his
(forthcoming).
40
“Perception is empirical consciousness, i.e. one in which there is at the same time sensation” (B207).
41 42
A162/B203. A168/B210.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

second relation holds between the content of experience (perception) and the
properties of its objects.43
This describes the structure of the most rudimentary notion of experience, mere
perception of sensory manifolds. Let us now consider experience of objects in some
intersubjectively accessible ‘public’ space, objects which we think of as existing
independently of our perceptions of them and as being accessible to other cognitive
subjects.44 Kant refers to this as ‘experience’ in the System of Principles to distinguish
it from mere ‘perception’ of sensory manifolds.45 To experience these objects, Kant
argues, we must think of them using the relational categories: “inherence and
subsistence, causality and dependence, community [Gemeinschaft] (reciprocal action
[Wechselwirkung] between agent and patient)” (A90/B106).46 Although these are
often referred to by Kant’s own parenthetical explanations in the CPR (“substance

43
The titles of the Axioms of Intuition (“all intuitions are extensive magnitudes”—B202) and the
Anticipations of Perception (“in all appearances the real, which is an object of the sensation, has intensive
magnitude, i.e. a degree”—B207) announce the outcome of this line of reasoning (the properties of intuited
manifolds), while the body of those sections gives the argument for the first grounding relation (form-
content): the forms of intuition plus the categories of quantity (Axioms of Intuition) and the categories of
quality (Anticipations of Perception) ground the fact that intuition presents manifolds with extensive and
intensive magnitudes. Kant takes the second grounding relation (content-property) to be clear to the reader
by that point in the CPR, for he announced it earlier as the supreme principle of synthetic a priori judgment
(A158/B197).
44
This is my gloss on what Kant means by distinguishing intuition of an object from cognition of its
existence (A160/B190) and why the Analogies of Experience are requirements on cognition of existence
specifically (A176, B219, A182, B225, B233, B257, A215–16/B262–3). Existence cannot here mean ‘abso-
lute positing’ (see Ch. 1) for we can absolutely posit mere sensory manifolds (there is such-and-such a
sensory manifold) without using the categories of relation. Likewise, ‘existence’ here cannot mean causal
efficacy for that would render at least the Second Analogy tautologous. Instead, I propose, ‘existence’ of an
object here means: its being intersubjectively available to multiple cognizers and possessing properties
independently of any particular discursive cognizer’s perceptions of it. To use another Kantian distinction,
the Axioms of Intuition and Anticipations of Perception state the necessary conditions on representing
appearances in the empirical sense while the Analogies state the necessary conditions on representing the
thing in itself in the empirical sense. See A29–30/B45, A44–5/B60, and what I take to be the closely related
notions of ‘necessary universal validity’ and the distinction between judgments of perception and judg-
ments of experience in Prol. }}18–19 (Ak. 4: 298–301).
45
This corresponds to one of Kant’s dominant uses of ‘Erfahrung’: a synthetic unity of perceptions
(A110, A124 f., A156/B195, B161, B218, A213/B260, A183/B226) that can, consequently, represent ‘existing’
objects (see previous note). I have been using ‘experience’ in a more minimal sense to refer to ‘empirical
cognition,’ thinking a sensibly given object under a concept (B146, B165), in order to reserve ‘cognition’ for
the specifically modal notion of representing an object one can know to be possible (Bxxvi n.).
46
Given the topic of this book, readers might wonder where the modal categories belong, which Kant
includes, with the categories of relation, among the dynamical categories (A162/B201). I think Kant is
oversimplifying matters by including the modal categories among the dynamical categories. Even in the
case of cognition of pure manifolds, we can distinguish the possible, that which agrees with the form of
intuition and the categories of quantity, from that which does not so agree; in the case of perception of
sensory manifolds we can distinguish, additionally, between what is actually present to the senses from
what is not. I think that at each ‘stage’ of cognition (intuition, perception, experience) we can apply modal
categories. Only at the final level (experience in the full sense) do we apply the modal categories according
to the principles given in the Postulates; Kant is correct, then, in counting the Postulates as he formulates
them as dynamical principles rather than mathematical principles. In the note he added to the B edition to
explain the mathematical/dynamical distinction (B201n.), Kant does not mention modal categories.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

and accident, cause and effect, reciprocal action”), or by the shortened names he gives
them in the Prolegomena (substance, cause, community),47 it is clear from his initial
naming of them in the CPR that they are concepts of relations between objects. At the
form-content level, the forms of experience (space, time, and the categories of
relation) (i) non-logically, (ii) non-causally, (iii) asymmetrically explain why the
content of experience represents its objects as standing in relations of inherence
(accidents to substances),48 causation (substances, objects which do not stand in
inherence relations to other objects, to accidents of substances), and community
(substances to substances that cause one another’s accidents), while (iv) it is formally
necessary both that we have these forms of experience and that experience has this
content. At the content-property level, though, this grounds relations among objects
of possible experience: the content of experience (i) non-logically, (ii) non-causally,
(iii) asymmetrically explains why objects of experience stand in the relations of
inherence, causation, and community.49
Notice, though, that these three relations among objects of experience are them-
selves real grounding relations: a substance is a ratio essendi of the possibility of its
accidents, for an accident would not be possible without the very substance it inheres
in; and substances are rationes fiendi of one another’s accidents (grounds of alter-
ation). These are real relations because the ground of a substance’s accidents is not its
concept, but the other substances with which it interacts, and substances are not
logical grounds of the accidents they cause to inhere in one another. In Chapter 3 we
saw Kant, in Negative Magnitudes, grappling with the problem of how we understand
logically irreducible relations of real grounding; in that work he settled on the—
admittedly not very satisfying—view that we simply possess a primitive stock of
unanalyzable concepts of kinds of real grounding relations.50 Kant’s pre-Critical
project was not to logically analyze the real grounding relation (this, by definition,
is impossible) but to acquire ‘insight’ [Einsicht] into such relations, in Kant’s tech-
nical sense of that term: a prioriG knowledge.51 So Kant’s pre-Critical problem was to
explain relations of real grounding among finite things, that is, to acquire knowledge
of the ground that makes real grounding relations among finite things possible in
the first place.52 His pre-Critical solution to this problem (see Chapter 5.5) was to

47
Ak. 4: 303.
48
Kant defines an accident as something that inheres in something else, and a substance as that which
does not inhere in anything else (nothing can inhere in itself). See A187/B230, Ak. 8: 225 n., Ak. 28: 562–3,
Ak. 28: 639, and Ak. 28: 1104 f.
49
The titles of the three Analogies announce the relations that obtain among objects of experience,
while the body of the text of each Analogy is an argument for the corresponding form-content grounding
claim. Kant takes the second grounding relation (content-property grounding) to be clear to the reader by
that point in the CPR, for he announced it earlier as the supreme principle of synthetic a priori judgment
(A158/B197).
50
NG, Ak. 2:204.
51
See LPö (Ak. 24: 539), LDW (Ak. 24: 730), Refl. 1866 (Ak. 16: 141), Refl. 1955 (Ak. 16: 169), and Refl.
2394 (Ak. 16: 342), CPrR (Ak. 5: 27, 46, 47), and CJ (Ak. 5: 83).
52
For a clear statement of the problem, see NG (Ak. 2: 203–4) and MH (Ak. 28: 12, 102).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

appeal to the common source of the existence of a real ground and its consequence,
God; finite things can be the real grounds of one another’s alterations because
they share a common real ground of their existence.53 Although this may merely
push the question back one step (how can we have insight into the grounding
relationship between God and finite substances?),54 there is a deeper reason why
Kant must reject it in the Critical period: it violates his restriction on cognition of
noumena (God).
But now we have seen Kant’s solution to at least part of this problem within the
CPR: we can have insight (knowledge from the ground) into the real grounding
relations inherence and causation among phenomena because we know that their
ground (the ground of the possibility of these relations obtaining) is the grounding
relation between the content of experience and the objects of experience. We also
have insight into why this grounding relation obtains: objects of experience are
phenomena, not noumena, so the content of experience grounds their properties.
But our insight goes yet further: we have insight into the possibility of experience
having this content because we know the a priori ground of its possibility, the
forms of experience, and we have a priori insight into how the forms of experience
make it possible for experience to have this content, synthesis of sensory manifolds
according to categorial schemata (see Chapter 6.7). This explains one aspect of a
typical Kantian formulation in the CPR; he does not merely describe the form of
experience as grounds or conditions of possible experience but as a priori
grounds.55 Read in its full historical context, though, this does not merely mean
that the forms of experience can be known a prioriJ (although they can) but that
they are the antecedently determining rationes essendi (grounds of possibility) from
which we can have a prioriG knowledge of possible experience and, thus, of all
possible objects of experience.56 Kant has solved several of the problems that arose
for him in the 1760s regarding our ability to have insight (knowledge from the
ground of possibility) into real grounding relations by internalizing the grounding
relation, and its relata, within experience itself.57

53
ND (Ak. 1: 413), MH (Ak. 28: 51–2, 132), ID (2: 407–8).
54
Kant implicitly acknowledges that his own view (see Ch. 5) explains the possibility of finite substances
standing in real grounding relations only through an appeal to the unexplained real grounding relation
between God’s will and those very substances: “the will of God is something. The world which exists is
something completely different” (NM, Ak. 2: 202).
55
A30/B46, A116, B150, A136/B175, A156–7/B196–7, B202, A177/B220, A268/B324.
56
See Chs. 1.3 and 3.2 for more on eighteenth-century theories of grounding.
57
Abaci (2013) attributes to Kant a view on which his modal metaphysics (within experience) is
oddly divorced from his modal epistemology: the grounds of real possibility are the forms of
experience but we cannot know that an object is really possible just by proving a priori its
compatibility with the forms of experience (we must actually experience the object or prove what
I call its ‘formal necessity’). My interpretation delivers a more unified account of Kant’s modal
epistemology and metaphysics.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

7.4. Empirical-Causal Possibility

We have already seen that the “Postulates of Empirical Thinking in General”, the
section of the System of Principles in the CPR devoted to the modal categories
(possibility, existence, and necessity), contains a definition of formal possibility.58
However, the Postulates section also contains the following discussion of necessity:
From this it follows that the criterion of necessity lies solely in the law of possible experience
that everything that happens is determined a priori through its cause in appearance. Hence we
cognize only the necessity of effects in nature, the causes of which are given to us, and the mark
of necessity in existence does not reach beyond the field of possible experience, and even in this
it does not hold of the existence of things, as substances, since these can never be regarded as
empirical effects, or as something that happens and arises. Necessity therefore concerns only
the relations of appearances in accordance with the dynamical law of causality, and the
possibility grounded upon it of inferring a priori from some given existence (a cause) to
another existence (the effect). Everything that happens is hypothetically necessary; that is a
principle that subjects alteration in the world to a law, i.e. a rule of necessary existence, without
which not even nature itself would obtain. (A227–8/B280)

That Kant here writes that we “cognize the necessity of effects in nature” shows that
he does not have formal necessity in mind, for particular events are not formally
necessary. In this section I develop an alternate conception of necessity, which I call
empirical-causal necessity, and argue that, in this passage and in others, Kant invokes
this kind of necessity. However, he does not explicitly distinguish this conception of
empirical-causal necessity from formal necessity. The distinctions I am drawing
between kinds of modality are made implicitly, not explicitly, by Kant.
Empirical substances are in time, so they undergo alteration [Veränderung], which
Kant defines as successive change [Wechsel] of accidents.59 In the Second Analogy of
Experience, Kant argues that the form of time and the category of cause–effect ground
(see section 3) the fact that every empirical alteration has a determining cause,60 a

58
A218/B265.
59
A188/B232. My discussion of Kant on causation is informed by the meticulous analyses of Watkins
(2005).
60
A determining cause is to be distinguished from a merely sufficient cause. A sufficient cause is
opposed to a merely partial or insufficient cause, which is insufficient, by itself, to bring about its effect.
A determining cause, on the other hand, operates according to a rule: any substance of the same kind,
under the same conditions, would cause the same effect. The crucial distinction between determining and
merely sufficient causes derives from Crusius (De Usu }}I, III, XX); a determining ground is one that, under
the same conditions, would always produce the same effect. Crusius introduces this distinction to save
freedom from the necessitarian consequences of the Leibnizian and Wolffian versions of the PSR: the free
actions of rational agents have sufficient, but not determining grounds, according to Crusius. Kant praises
this distinction in ND (Ak. 1: 398) but denies that free actions lack determining grounds, a position he will
later reverse (see section 5). It is clear, both from the schema of cause–effect (A144/B183), and from the
second Analogy itself, that Kant takes himself to have proved that empirical alterations have determining
and not merely sufficient grounds: “in accordance with such a rule there must therefore lie in that which in
general precedes an occurrence the condition for a rule, in accordance with which this occurrence always
and necessarily follows” (A193/B238). Cf. Hogan (2005) and Hogan (2009a).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

cause from which the effect follows according to a law.61 This means that it is formally
necessary that every empirical event has a determining cause and that cause–effect
relations are governed by laws. In other words, the principle of causation is formally
necessary, but which causal laws govern the actual world, and which empirical events
occur, are formally contingent.62 Let p be the proposition that alteration E occurs at
time t. Consider the following definition:
(Empirical-causal possibility) It is empirically-causally possible that p if and only
if it is compatible with actual natural laws, and the past history of the empirical
world up until time t, that p.63
For instance, it is empirically-causally possible (empirically possible, for short) that
I refrain from making a lying promise at time t just in case it is compatible with the
entire history of the empirical world up to that moment, and the natural laws, that
I so refrain. As a merely heuristic device, we might introduce the notion of ‘possible
worlds’ and say: it is empirically possible that E occur at t if and only if there is a
possible empirical world (series of empirical alterations) that is qualitatively indis-
tinguishable from the actual world before t, and has the same natural laws, in which E
occurs at t.64 According to Kant, it is formally necessary that every event has a
determining cause. That causes are determining, for Kant, means that the laws that
govern the operation of these causes are deterministic. This means that if C is the
cause of E, and C occurs, and the background conditions are held fixed, it is
incompatible with the natural laws that E not occur. To invoke the heuristic of
possible worlds once again, we might define a ‘divergent world’ as one that is
qualitatively identical to the actual world up to time t, in which the actual laws
obtain, but is qualitatively different from the actual world after time t. The natural
laws governing the actual world are deterministic if and only if there are no divergent
possible worlds. The history of the empirical world up to a moment, plus the natural
laws, determine a unique future history. From the definition of empirical-causal

61
Some scholars have questioned whether the Analogies of Experience are supposed to demonstrate
that there are causal laws, or merely that every alteration has a cause. I am going to assume that Kant took
himself to have established the stronger conclusion; see section 4, and the first footnote in Ch. 8, for more
discussion.
62
See A766/B795: “Thus if wax that was previously firm melts, I can cognize a priori that something
must have preceded (e.g., the warmth of the sun) on which this has followed in accordance with a constant
law, though without experience, to be sure, I could determinately cognize neither the cause from the effect
nor the effect from the cause a priori and without instruction from experience.”
63
In Stang (2011) I called this empirical possibility. There is a slight complication here, though: on
Kant’s view ‘the entire history of the empirical world up to time t’ does not refer to any object of possible
experience so there is no such object (see the resolution of the first antinomy, A517–23/B545–51). See later
discussion for how to deal with this problem.
64
By introducing this heuristic device, I am not claiming that Kant accepts a ‘possible worlds’ analysis
of modality, something I have consistently rejected. On the possible worlds analysis of determinism, see
Lewis (1983), 31–3.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

possibility above,65 we can define a companion notion of empirical-causal necessity

(letting p be as before):
(Empirical-causal necessity) It is empirically-causally necessary that p if and only
if it is incompatible with actual natural laws, and the past history of the empirical
world up until time t, that ¬p.
That natural laws are deterministic entails that every empirical event is empirically-
causally necessary (empirically necessary, for short).66 No non-actual event is empir-
ically possible, because the prior history of the empirical world and the natural laws
ensure that it does not occur. Consequently, the actual, the empirically possible, and
the empirically necessary are co-extensive.
I have defined empirical possibility and necessity in terms of the relation of being
compatible with the past and the laws. As with the corresponding definition of formal
possibility, the question arises: what does ‘compatibility’ mean here? I propose that
we define compatibility in terms of grounding:
(Defn.) p is compatible with actual natural laws and the past history of the
empirical world up to time t if and only if the facts about the actual natural laws
and the past history of the empirical world up until t and the laws do not wholly
ground the fact that ¬p.
This, in turn, allows us to define empirical-causal possibility and necessity in terms of
grounding:
(Empirical-causal possibility) It is empirically-causally possible that p if and only
if it is not the case that the fact that ¬p is wholly grounded in facts about actual
natural laws, and the past history of the empirical world up until time t.
(Empirical-causal necessity) It is empirically-causally necessary that p if and only
if the fact that p is wholly grounded in facts about actual natural laws, and the past
history of the empirical world up until time t.
The same problem that arose in the case of defining formal possibility in terms of
grounding—precisely delineating the relevant notion of ground—arises again in the
case of empirical possibility and necessity, but it is more tractable in this case. The
laws describe the causal powers of phenomenal substances,67 so the ultimate grounds
of empirical-causal possibility are those powers, which are both rationes fiendi of the
existence of alterations in phenomenal substances and rationes essendi of the
empirical-causal possibility of those very alterations. In section 3 I discussed at length

65
Combined with the principle that □p $ ¬◇¬p; see MV (Ak. 28: 418), MvS (Ak. 28: 498), ML2 (Ak.
28: 556, 557), and MD (Ak. 28: 633).
66
As I have set up the definitions, it would be more precise to say: for any event E, the proposition that
E occurs is empirically necessary. However, I am treating the two formulations as identical, for ease of
exposition.
67
See Ch. 8.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

how the content of possible experience grounds the possibility of causal relations
among phenomena, but now we have a relation of possibility-grounding among
phenomena themselves. The powers of phenomenal substances are grounds of the
empirical-causal possibility of alterations in one another, and this grounding relation
is itself made possible by the content of experience (see section 3). So we have a
relation of real grounding between, on the one hand, powers of phenomenal sub-
stances, and, on the other hand, possible alterations in substances, and although this
relation is not logically intelligible (the concept of the former does not contain the
concept of the latter), we can have a prioriG insight into it: the ground of its
possibility is the content of experience itself.
However, there is a problem with this definition of empirical-causal possibility.
According to Kant’s transcendental idealism, alterations in empirical substances, and
temporally ordered series of them, are phenomena: objects of a (formally) possible
experience. A formally possible experience of a series of empirical alterations is a unity
synthesized from experiences of its individual elements. Since we can only ever
synthesize finitely many elements (a completed infinite series is formally impossible)
it follows that no infinite series of empirical alterations is a possible object of experience
for us. Consequently, “the past history of the empirical world up until time t” does not
refer to any possible object of experience. For any finite empirical series S there is a
‘larger’ finite empirical series S* that includes S as a proper part; in other words, finite
empirical series leading up to time t can be indefinitely ‘extended’ to include alterations
further in the past, but the “complete past before t” refers to no (formally) possible
phenomenon. Since it clearly refers to no noumenon (it is the concept of a temporal
series, and noumena are not in time) I conclude that it refers to nothing: there is no
such thing, according to Kant, as the complete past before a given time.68
Fortunately, this problem can be resolved. Kant’s deterministic conception of
natural laws entails, not that the natural laws plus the past history of the empirical
world up until time t determine a unique future, but that the natural laws plus the past
history of the empirical world for an arbitrary interval of non-zero duration before
time t determines a unique future.69 To use, once again, the heuristic of possible
empirical worlds (series of empirical alterations) from previously, a world w is said to
diverge from the actual world just in case w is qualitatively identical to the actual
world for a non-zero interval before t, but is qualitatively different after t. Kant’s
determinism entails that there are no worlds governed by the actual natural laws that
diverge from the actual world after a period of being qualitatively identical. So if we

68
This is not to deny that we have an idea of the “complete series of alterations before a given time,” a
concept of the unconditioned series of temporal conditions of a given alteration. This idea is the subject of
the first Antinomy. But this does not mean there is any object answering to this idea (a complete series of
past times).
69
This is a consequence of the continuity of the relation between cause and effect. See A207–8/B253–4.
This means that if empirical series A is the same as empirical series A* for a non-zero interval before time t,
they are the same after t.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

fix a non-zero interval of the empirical series before t, given the natural laws, only one
future is possible. This requires a slight change in the definitions of empirical-causal
possibility and necessity:
(Empirical-causal possibility) It is empirically-causally possible that p if and only
if it is not the case that the fact that ¬p is wholly grounded in facts about the actual
natural laws and a non-zero interval of the past before time t.
(Empirical-causal necessity) It is empirically-causally necessary that p if and only
if the fact that p is wholly grounded in facts about the actual natural laws and a
non-zero interval of the past before time t.
In what follows I will sometimes speak of empirical-causal possibility as being
grounded in the natural laws and “the past.” In every case this means: the natural
laws and an arbitrary non-zero interval of the past.
It should also be clear that empirical possibility and necessity are kinds of real
possibility and necessity, respectively. Recall the criteria on real modality from
section 1:
(Real PossibilityE) Empirical-causal possibility ◇Ep (and its associated kind of
necessity □Ep, where □Ep $ ¬◇E¬p) is a kind of real possibility (and □Ep is a
kind of real necessity) only if
(i) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇Ep (equivalently, it is
not a conceptual truth that □Ep ⊃ □Lp), and
(ii) Groundedness: if ◇Ep then the fact that ◇Ep has a real ground in some actual
object or principle.
Empirical possibility and necessity satisfy these criteria. It is logically possible that,
for instance, some non-actual empirical alteration occurs (e.g., no contradiction
results from the proposition that it occurs), but it is empirically-causally impossible
for any non-actual alteration to occur. Similarly, if it is empirically-causally possible
that some alteration occurs at some designated time, this alteration has real grounds
in actual objects: the actual natural laws, and a non-zero interval of the past before
that time. Similar remarks apply to empirical necessity. In section 2 I noted that one
difference between Kant’s pre-Critical conception of real possibility and his Critical
conception of real possibility is that pre-Critical real possibility is grounded in mind-
independent reality (noumena). Empirical possibility (and necessity) is grounded in
facts about phenomena: powers of phenomenal substances and the past before a
given time. As such, it is not a kind of ‘noumenal’ possibility.
When Kant writes, in the long passage from the Postulates section quoted at the
beginning of this section (A227–8/B280), that “it follows that the criterion of
necessity lies solely in the law of possible experience that everything that happens
is determined a priori through its cause in appearance,” he means that it follows from
the principle that every alteration has a determining cause that every alteration is
empirically necessary. He writes that “everything that happens is hypothetically
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

necessary” because empirical necessity is a form of hypothetical necessity. Alterations

are necessary given the hypothesis of the prior history of the empirical world, and the
natural laws. He contrasts substances with accidents because, on Kant’s view, sub-
stances are permanent; they do not arise or perish, unlike alterations. Thus, restrict-
ing ourselves to causes within the empirical world, substances are not caused—their
alterations are caused—and the existence of a given set of substances is neither
empirically-causally possible nor necessary; empirical-causal modality is simply not
defined for substances themselves.70
Before concluding my discussion of ‘immanent’ modality, I want to comment
briefly on Kant’s crucial claim that “the categories have this special feature: they do
not augment the concept to which they are ascribed in the least, as a determination of
the object, but express only the relation to the faculty of cognition” (A219/B266).
I have already explained at length what it means that <existence> is not a determin-
ation or real predicate. In this chapter I have focused on thinking of formal possibility
and necessity as operators on propositions (judgment-contents), because many
questions in formal possibility are most naturally formulated this way, but we can
also define these modal concepts as predicates of objects:
(FP-objects) A phenomenon x is formally possible if and only if it is compatible
with the forms of experience that we experience x.
(EN-objects) An alteration x in a phenomenal substance is empirically-causally
necessary if and only if it is incompatible with the natural laws and a non-zero
interval of the past relative to x that x not occur.
But it follows from this that these object-level predicates are not determinations of
the domain of objects for which they are defined: it is (formally) impossible that there
is a phenomenon that is not formally possible, and it is (formally) impossible that
there is an alteration that is not empirically necessary.71 To say of a phenomenon that
it is formally possible is not to further determine that object (ascribe to it a predicate
it might lack) but to express the compatibility of the experience of it with the forms of
experience. To say of an alteration that it is empirically necessary is not to determine
it (ascribe to it a predicate it might lack) but to express its relation to the past and the
laws, which, given your forms of experience, make it necessary.

7.5. Noumenal-Causal Possibility

We have already seen that the Postulates of Empirical Thought, the section of the
CPR devoted to the modal categories (possibility, actuality, and necessity), contains

70
This is not the only passage in which Kant deploys this conception of empirical necessity. See MV
(Ak. 28: 417) and MM (Ak. 29: 814), as well as Refl. 4298 (Ak. 17: 499) and 5177 (Ak. 18: 109).
71
In the technical terminology of Ch. 1.6, <formally possible> does not determine <phenomenon> and
<empirically necessary> does not determine <alteration in phenomenal substance>.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

discussions of formal possibility and of empirical-causal possibility, although they are

not explicitly distinguished. The conclusion of the Postulates, however, is harder to
characterize. Kant raises a pair of questions: “whether the field of possibility is greater
than the field that contains everything actual, and whether the latter is in turn greater
than the set of that which is necessary” (A230/B282). Given the interdefinability of
possibility and necessity, though, these are really one question: are there non-actual
possibilities?72 Kant’s subsequent discussion of non-actual possibilities is quite intri-
cate, so I will analyze this text in some detail.73
The first case Kant discusses is the possibility of a non-actual experience:
[a] [The two questions from above] mean, roughly, to ask whether all things, as appearances,
belong together in the sum total [Inbegriff] and the context [Kontext] of a single experience, of
which each given perception [Wahrnehmung] is a part and which [der] therefore could not be
combined with any other appearances, or whether my perceptions could belong to more than
one possible experience (in their general connection). (A230–1/B283)74

“A single experience, of which each given perception is a part” is a reference to Kant’s

strongest conception of experience, experience as the totality of all perceptions
combined according to natural laws, which he characterizes as follows in the
A Deduction: “there is only one experience, in which all perceptions are represented
as in thoroughgoing and lawlike connection [ . . . ] if one speaks of different experi-
ences, they are only so many perceptions insofar as they belong to one and the same
universal experience” (A110).75 I will refer to this concept of a “single all-
encompassing experience” (A582/B610) as ‘universal’ experience. Universal experi-
ence, as I understand it, is the idealized limit of a process of combining individual
perceptions according to a priori principles and natural laws; this is a rough charac-
terization, but for our purposes it will have to suffice.76 Since alternate individual
experiences are possible, according to A110, in virtue of belong to a ‘universal’
experience, Kant focuses on the question of whether a ‘universal’ experience, other
than the actual one, is possible. Possible experiences are playing a role analogous to
possible worlds here, and just as ‘possible worlds’ accounts of possibility standardly
require that worlds be maximal (if p is consistent with world w, then p is true in w)
Kant is concerned with maximal experience, or ‘universal’ experience.

72
If possibility and necessity are interdefinable in the standard way—necessarily p if and only if not
possibly not-p—the questions is the field of possibility greater than the field of the actual? (i.e. are there non-
actual possibilities?) and is the field of the actual greater than that of the necessary? (i.e. is anything
contingent?) receive the same answer.
73
Cf. Stang (2011), 454–8, and Abaci (2013), 4–7.
74
Note that, in the second relative clause (“which [der] therefore could not be combined with any other
appearances [ . . . ]”), the masculine relative pronoun ‘der’ must take sum total (Inbegriff) or context
(Kontext) as its referent, not perception (Wahrnehmung), which is not clear in Guyer–Wood (I have
modified their translation accordingly).
75
Cf. A492–3/B520–1 and A582/B610.
76
I say more about ‘universal experience’ in Stang (2012) and (forthcoming).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

This passage can be read as raising either of two modal questions:

(1) Could my actual perceptions be part of more than one possible universal
experience?
(2) Is there more than one possible universal experience?
On the first reading, in the final clause of the sentence (underlined above) Kant is
referring to the de re possibility that my actual perceptions be part of a non-actual
single, universal experience. On the second reading, Kant is referring to the de dicto
possibility that I have some perceptions or other that are part of a non-actual single
universal experience. To continue with the possible worlds analogy, these are analo-
gous to asking, of some proper part of the actual world, (1*) would this exist in
another possible world? or (2*) are there any possible worlds other than the actual
one? While this passage alone does not settle the interpretive question, the continu-
ation of Kant’s discussion in the rest of the paragraph does. He goes on to discuss the
possibility of a non-actual form of experience, and then the possibility of a non-actual
matter of experience; I take this to mean that the original question posed at the
beginning of the paragraph is about the possibility of a non-actual experience, (2).
The most urgent question raised by this passage, though, is what sense of possibility
is involved in these questions, and my argument, that it is a kind of possibility
grounded in noumena, will still apply if the original question raised at A230–1/B283
is (1) rather than (2).
Kant goes on to discuss two modal issues, the first of which is the possibility of
non-actual forms of experience:
[b] Even were they possible, we could still not conceive of and make comprehensible [erdenken
und faßlich machen] other forms of intuition (than space and time) or other forms of
understanding (than the discursive form of thinking, or that of cognition through concepts);
and even if we could, they would still not belong to experience, as the sole cognition in which
objects are given to us. (A230 f./B283)

Kant remains agnostic about the possibility of forms of intuition other than space
and time, and about the possibility of a non-discursive (intellectual—see Chapter 10)
understanding.77 We cannot cognize the real possibility of such non-actual forms of
experience because the actual forms of experience ground the only space of real
possibilities we can cognize: formal possibility and empirical-causal possibility. Since
(by definition) no non-actual form is either an object in that space or a form
constitutive of such an object, we cannot cognize their real possibility. Immediately
after this, he writes:
[c] Whether other perceptions than those which in general belong to our entire possible
experience, and therefore an entirely different field of matter, can obtain cannot be decided

77
Kant expresses agnosticism about the possibility of a non-spatiotemporal form of intuition elsewhere.
See A42/B59, A231/B283 and Prol. }57 (Ak. 4: 351).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

by the understanding, which has to do only with the synthesis of that which is given. (A231/
B283).

Coming, as it does, immediately after Kant’s discussion of non-actual forms in [b],

this is most plausibly read as a discussion of whether a non-actual matter of
experience is possible. Retrospectively, it sheds light on [a]. Kant’s question there
was (2), and he then goes on to separate this into two subordinate questions: [b] is a
universal experience with a non-actual form possible?; and [c] is a universal experi-
ence with a non-actual matter possible?
We can conclude our line-by-line reading of the Postulates discussion of non-
actual possibilities with this passage:78
[d] All that can be added to my understanding is something beyond agreement with the formal
conditions of experience, namely connection with some perception or other; but whatever is
connected with this in accordance with empirical laws is actual, even if it is not immediately
perceived. However, that another series of appearances in thoroughgoing connection with that
which is given to me in perception, thus more than a single all-encompassing experience, is
possible, cannot be inferred from that which is given, and even less without anything being
given at all. (A231/B284)

I take the first sentence to mean that what is left undetermined by the forms of
experience is the particular objects of experience, which are either perceived directly
or inferred from their causal connection with directly perceived objects.79 However,
Kant reminds us, “whatever is connected with” particular perceived objects agrees
with empirical laws.80 In other words, whatever formally contingent empirical
objects we experience, these objects are governed by causal laws. Consequently,
their alterations are empirically necessary because they follow from previous alter-
ations by natural laws. So no non-actual empirical alterations are empirically pos-
sible. In the second sentence Kant expresses agnosticism about the possibility of
“more than a single all-encompassing experience” which I read as a reference back to
“the sum total [Inbegriff] and the context [Kontext] of a single experience” in [a].
Kant is returning to the original question with which he began in [a] (is a universal
experience other than the actual one possible?) and claiming that what is given to us
(actual empirical objects) does not allow us to infer such a possibility.

78
I have so far ignored Kant’s comment that these are “appropriate [artige] questions, and are to be
resolved synthetically [von synthetischer Auflösung], though they also fall under the jurisdiction of reason”
(A230/B282). I take this to mean that the questions I am investigating in the body of the text (whether a
non-actual matter or form of experience is possible) are not settled by analysis of our concepts <possible>
and <actual> but must be referred to the faculty of reason, which seeks conditions (grounds) for the
conditioned; in this case, reason seeks a ground of non-actual possibilities outside the realm of possible
experience, in noumena. Cf. Ch. 9.6.
79
Cf. the postulate of actuality, A218/B266 and A225/B272.
80
It is possible to read this passage as concerning the possibility of non-actual empirical laws. However,
the same questions arise: non-actual empirical laws are clearly formally possible and empirically impos-
sible, so why does Kant express agnosticism? Cf. Guyer (1998), 304–8.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

I have been analyzing these passages in detail to determine exactly what question
about non-actual possibility Kant is investigating, but the deeper issue is: what kind
of possibility is involved here? As Kant himself points out, in [d], the matter of a non-
actual universal experience is formally possible, but the alterations that would be part
of that matter are empirically-causally impossible. So if Kant has either formal or
empirical-causal possibility in mind, it is unclear why he expresses agnosticism, since
his own theory straightforwardly delivers a determinate answer. The same point
holds with respect to the question of the possibility of a non-actual form of experience
in [b]; non-actual forms of experience are trivially formally impossible, and causal-
empirical possibility is not defined for them (they are not alterations of phenomenal
substances). Consequently, Kant must have a distinct kind of modality in mind in
these passages. Both formal possibility and empirical-causal possibility have what
I have called ‘immanent’ grounds, in the sense that they are either objects of
experience (empirical possibility) or the subjective forms that condition those objects
(formal possibility). However, phenomena and their subjective forms are not the only
domain in Kant’s metaphysics; there are also the noumena. Although we cannot
cognize noumena (know what is really possible for them) this does not necessarily
preclude us from raising questions about noumena and whether they ground non-
actual possibilities.81
When Kant expresses agnosticism about whether the matter or form of a non-
actual experience are possible, he might be expressing agnosticism about whether
there is a noumenal ground of the possibility of a non-actual matter and form for
experience. While it is somewhat harder to say what role noumena might play in the
possibility of a non-actual form of experience, it is relatively straightforward to
understand the role they might play in the possibility of non-actual matter. The
matter of experience is given to us through affection of our sensibility by noumena, as
Kant makes abundantly clear in On a Discovery:
Having raised the question ‘what gives sensibility its matter, namely sensations?’ [Herr
Eberhard] believes himself to have pronounced against the Critique when he says “we may
choose what we will—we nevertheless never arrive at things-in-themselves.” Now, that, of
course, is the constant contention of the Critique; save that it posits this ground of the matter
of sensory representations not once again in things, as objects of the senses, but in something
super-sensible, which grounds the latter, and of which we can have no cognition. It says that
objects as things-in-themselves give the matter to empirical intuitions (they contain the ground
by which to determine the faculty of representation in accordance with its sensibility), but they
are not the matter thereof. (Disc., Ak. 8: 215)

Not only here, but in numerous other passages in the CPR and other Critical writings,
Kant maintains that things in themselves (noumena) affect our sensibility, producing

81
On the restriction of formal and empirical possibility to empirical objects, and the question of
whether there is a kind of modality that applies beyond the bounds of experience, see Schneeberger
(1952), 32–3.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

the matter of experience.82 This, in turn, allows us to understand how noumena

might ground the possibility of a non-actual matter for experience: by having the
causal power to produce in us a matter distinct from the one they actually produce.
I will provisionally label this concept of possibility noumenal-causal possibility and
define it as follows: where it is formally contingent that p,83 it is noumenally-causally
possible that p if and only if noumena have the causal power to so affect our
sensibility that the resulting experience is one of which p is true. Obviously, this is
only defined for values of p that are about experience and its objects. Since causal
powers are real grounds, this kind of noumenal possibility is a kind of real possibility.
But this is not the only place where the notion of noumenal-causal possibility
arises in Kant’s system. His theory of freedom, one of the foundations of the entire
Critical system, rests on a notion of possibility grounded in the causal powers of
noumena (of agents’ wills considered as noumena). According to Kant, ought implies
can, so if I ought not to perform a given act, then I can omit that act. If the moral law
(which has the form of an imperative or ‘ought’ in relation to finite wills, like ours,84
which are afflicted with sensible inclination) applies to me then it must be possible
that I omit the actions I ought to omit. But every act I commit is an empirical
alteration (the changing of accidents in empirical substances, e.g., my body) and is
therefore a necessary consequence of prior history and natural laws. In the case of a
morally wrong act, how can we reconcile the empirical-causal necessity of that act
with the possibility that I omit it?85 Kant thinks that if we restrict our account to the
empirical world, freedom, and therefore the moral law, is impossible:
For if appearances are things in themselves, then freedom cannot be saved. Then nature is the
completely determining cause, sufficient in itself, of every occurrence, and the condition for an
occurrence is always contained only in the series of appearances that, along with their effect,
are necessary under the law of nature. (A536/B564)

If appearances were things in themselves, i.e. if we could not appeal to anything other
than appearances and their properties, then freedom would be destroyed. Freedom

82
In the third chapter of Adickes (1924), Erich Adickes assembles an impressive array of textual
evidence for this claim. See esp. A190/B235, A387, A494/B522, Prol. (Ak. 4: 289, 314, 318), and GMM
(Ak. 4: 451).
83
If we want to extend the notion of noumenal-causal possibility to include the noumenal-causal
possibility of non-actual forms of experience, we would need to remove this clause.
84
GMM, Ak. 4: 413.
85
Kant confronts precisely this question in the CPrR: “If I say of a human being who commits a theft
that this deed is, in accordance with the natural law of causality, a necessary result of determining grounds
in preceding time, then it was impossible that it could have been left undone; how, then, can appraisal in
accordance with the moral law make any change in it and suppose that it could have been omitted because
the law says that it ought to have been omitted? That is, how can that man be called quite free at the same
point of time and in regard to the same action in which and in regard to which he is nevertheless subject to
an unavoidable natural necessity?” (Ak. 5: 95–6). Kant is trying to reconcile the empirical necessity of all of
my acts with the possibility of my omitting them. He thinks that my acts are normatively subject to the
moral law only if they are free, and freedom requires that it be possible for me to act otherwise.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

would be destroyed because each of my actions would be empirically necessary, and

there would be no possibility of my omitting any of my acts.86
Kant resolves the conflict between our freedom and the empirical necessity of our
actions by appeal to a non-empirical domain: noumena and their causal powers. Self-
conscious rational subjects, for Kant, are noumena (things in themselves)87 that
appear to themselves and other subjects as objects in space and time, subject to
causal laws. Therefore, as a rational subject, I can be considered both as a noumenon
(outside of space and time) and as the appearance of that noumenon (in space and
time, subject to causal determination). In CPrR Kant writes:
Every action [of a subject] . . . is to be regarded in the consciousness of his intelligible existence
as nothing but the consequence and never as the determining ground of his causality as a
noumenon. So considered, a rational being can now rightly say of every unlawful action he
performed that he could have omitted it even though as appearance it is sufficiently deter-
mined in the past, and so far, is inevitably necessary; for this action, with all the past which
determines it, belongs to a single phenomenon of his character, which he gives to himself and
in accordance with which he imputes to himself, as a cause independent of all sensibility, the
causality of those appearances. (CPrR, Ak. 5: 97–8)

Kant resolves the conflict between freedom and necessity by appealing to the causal
powers of the subject, considered as a noumenon, “a cause independent of all
sensibility.” When Kant writes that “a rational being can now rightly say of every
unlawful action he performed that he could have omitted it,” I take him to be
claiming that for any unlawful action an agent performs, the omission of that action
is possible because the agent, considered as a noumenon, has the causal power to
choose in such a way that the empirical world does not include that action.88
Kant renders compatible the empirical-causal necessity of my actions (as events in
time) with the possibility of my omitting them by claiming that, considered as a
noumenon, I could have caused those actions not to occur. I will call this kind of
possibility noumenal-causal possibility, and define it as follows:
(Noumenal-causal possibility) It is noumenally-causally possible that p if and only if
(i) it is formally contingent that p,89 and (ii) there is some noumenon with the causal
power to make it the case that p, where p is a synthetic proposition about phenomena.

86
Kant assumes throughout that the mere formal possibility of my omitting some act is not sufficient
for freedom. He appears to assume that freedom requires that I have the causal power to act otherwise than
I do. The mere compatibility of acting otherwise with the forms of experience entails nothing about my
having the power to act otherwise. Similarly, he passes over logical possibility: the mere logical consistency
of the proposition that I act otherwise is not sufficient for my being able to do otherwise.
87
See the end of Ch. 6.7 for an explanation of how I am using these terms.
88
Exactly how the very same event can be both empirically necessary and noumenally contingent
involves Kant’s notions of ‘empirical character’ and ‘intelligible character’ and lies outside the scope of this
chapter. Cf. Wood (1984), 102–12, Watkins (2005), 325–39, and Pereboom (2007), 550–9.
89
Noumenal-causal possibility is only defined for the matter of experience, which includes the actions
of rational agents. If we want to inquire into the possibility of non-actual forms of experience, as Kant does
at A230–1/B282–3, we will need another notion of possibility grounded in noumena.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THREE KINDS OF REAL POSSIBILITY

Noumenal-causal possibility, like empirical-causal possibility, is grounded in the

causal powers of actual things. Unlike empirical-causal possibility, though, it is
grounded in the causal powers of things in themselves, noumena. In this chapter
I am applying these conceptions of possibility—formal possibility, empirical-causal
possibility, and noumenal-causal possibility—to propositions about empirical
objects, e.g., the proposition that I omit some action I actually commit. In
Chapter 9 I will consider whether we can extend this notion of noumenal
possibility—possibility grounded in noumena—to the modal status of propositions
about noumena themselves, that is, whether noumena can be grounds of possibilities
for other noumena.
It should also be clear that noumenal-causal possibility and necessity satisfy the
two conditions on real possibility and necessity, respectively, from section 2:
(Real PossibilityN) Noumenal possibility ◇Np is a kind of real possibility only if
(i) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇Np, and
(ii) Groundedness: if ◇Np then the fact that ◇Np has a real ground in some actual
object or principle.90
In the case of formal and empirical possibility Kant can prove that they are exten-
sionally distinct from logical possibility: there are logical possibilities that are not
formally possible (e.g., that 2+2=5) and there are logical possibilities that are not
empirically possible (e.g., that some actual alteration not occur). Within theoretical
philosophy he cannot argue in this fashion in the case of noumenal-causal possibility
because we do not have sufficient theoretical knowledge of what is (and is not)
noumenally-causally possible.91 However, he can rest this claim on the fact that no
analysis of our concepts reveals <logically possible> and <noumenally-causally pos-
sible> to be co-extensive; we cannot determine whether these concepts apply to all
and only the same propositions merely by analyzing them. Consequently, they are
different concepts; noumenal-causal modality satisfies the Non-Logicality criterion.
Furthermore, it is clear that noumenal-causal modality satisfies the Groundedness
criterion: facts about what is noumenally-causally possible (and necessary) are
grounded in the causal powers of actual noumena, and causal powers are real
grounds.
Whereas Kant’s pre-Critical conception of real metaphysics incorporates what
I called the ‘Worldliness’ criterion—that real possibility is grounded in the nature of
mind-independent reality, what the Critical Kant calls things in themselves (nou-
mena)—Kant’s Critical modal metaphysics does not. It includes kinds of real possi-
bility that are not ‘worldly’ because they are grounded in subjective forms of

90
Using the interdefinability of possibility and necessity it is easy to define a notion of noumenal-causal
necessity; for the sake of brevity I omit this.
91
Kant’s account of practical reason, and how it underwrites cognition, and even, potentially, know-
ledge, lies outside the scope of this study. For more on this notion, see Schafer (unpublished).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THREE KINDS OF REAL POSSIBILITY 

experience (formal possibility) or in the properties of mind-dependent objects

(empirical possibility). But clearly, noumenal-causal possibility and necessity are
‘worldly’: they are grounded in the causal powers of mind-independent objects,
things in themselves (noumena). Consequently, noumenal-causal possibility and
other kinds of real possibility grounded in noumena have some claim to be the
closest Critical analogue of pre-Critical real possibility, a theme I explore in greater
depth in Chapter 9.
Some readers might object that my definitions of noumenal-causal modality
attribute causal and modal properties to noumena. This might be taken to violate
Kant’s doctrine that categories like <cause–effect>, <possible>, and <necessary>
cannot be used to cognize noumena. My response is that I am not yet attributing
modal properties to noumena.92 I am attributing modal properties to empirical
objects on the basis of the causal powers of noumena.93 As I argued in Chapter 6,
Kant distinguishes between thinking [Denken], cognition [Erkenntnis], and know-
ledge [Wissen]. All that is required for thinking of an object falling under a concept is
that there be no logical inconsistency in the concept. Since there is nothing logically
contradictory about noumena instantiating categories, we can think of them as
having causal powers.94 Since Kant expresses agnosticism about the possibility of a
non-actual matter and form of experience, it is sufficient that we can think of
noumena grounding these possibilities.95 We can think about whether non-spatial
and non-temporal intuition and a non-actual matter of experience are possibilities
grounded in noumena, but we cannot cognize this as a real possibility.

92
However, Kant does attribute modal properties to things in themselves, and distinguishes, for
instance, the possibility of a thing in itself from the possibility of an object of experience. See Refl. 5184,
5723, and 5177. See Ch. 9 for a more extensive discussion of modality for things in themselves.
93
In a variety of passages from the Critical period, Kant describes things in themselves in causal terms.
See especially his discussion of the forces of things in themselves in his response to Mendelssohn’s
Morgenstunden at Ak. 8: 153–4. See also A42/B49, A44/B61, B72, A190/B235, A358 f., and A372. Cf.
Adickes (1924), 28–37.
94
See “On the ground of the distinction of all objects in general into phenomena and noumena” in the
2nd edition (B294–315).
95
In a crucial passage of the CPrR Kant considers his warrant for applying the category of causality to
the supersensible world: “But how is it with the application of this category of causality [ . . . ] to things that
are not objects of possible experience but lie beyond its boundaries? For I was able to deduce the objective
reality of these concepts only with respect to objects of possible experience. But what gives them a place in
the pure understanding, from which they are referred to objects in general (whether sensible or not) is just
this: that I saved them only in case I proved that objects may nevertheless be thought through them
although not determined a priori. If anything is still wanting, it is the condition for the application of these
categories and especially that of causality to objects, namely intuition; where this is not given, application
with a view to theoretical cognition of an object as noumenon is made impossible” (Ak. 5: 54). We can think
supersensible objects under categories, but we cannot cognize or know them to fall under the categories.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

8
Nomic Necessity

8.1. Introduction
In the Critique of Pure Reason (CPR), Kant argues that the a priori forms of
experience—space, time, and the categories, especially the category <cause–effect>–
make it the case that alterations in phenomena are governed by causal laws that are
universal (they have no exceptions) and deterministic (the cause necessitates the
effect).1 The forms of experience make it the case that empirical objects are law-
governed, but they do not determine which particular laws govern them. In the
terminology of the previous chapter this means that the principle of causation itself
(that every event has a sufficient cause, and causation is governed by universal,
deterministic laws) is formally necessary, while the particular causal laws are formally
contingent. However, Kant also holds that these causal laws are necessary. The
combination of these two views raises an interpretive puzzle: in what sense are
particular causal laws necessary for Kant? Answering that question will be the aim
of this chapter.
Before continuing, it is important to be clear about what this question means.
I have attributed two different modal claims to Kant:
(1) Necessarily, there are universal laws governing nature.
(2) For any law L that governs nature, necessarily L.
The sense of necessity involved in the first claim is ‘formal necessity,’ as defined in
the previous chapter. Here I am interested in the second claim and the sense of
necessity involved in it. I will call it ‘nomic necessity.’2

1
Some commentators have questioned whether the forms of experience ground the fact that there are
universal and necessary causal laws; e.g., Paton (1936); Bird (1962); Beck (1978); Buchdahl (1965), (1969),
and (1992); and Allison (1983), 228–34 and (1994), 298. Some commentators claim that the second
Analogy proves only that there must be ‘rules’ (e.g., A-type events always cause B-type events) but not laws;
for the basis of the distinction, see A113, A126, Refl. 5414, and the discussion in Gloy (1976), 19 f. I do not
have space in this book to argue for the traditional view that the forms of experience require that there be
laws; see, however, A108, A113, A114, A127–8, B165, A159/B198, A216/B263, and Prol. } 36, as well as
Watkins (2005), 286–91. I am not claiming that the second Analogy successfully proves that there are causal
laws, but only that this is what Kant takes himself to have shown there.
2
My interpretation agrees, in broad outline, with Kreines (2009) and Watkins (2005). While Kreines
focuses mainly on arguing against the ‘best systems’ account of nomic necessity, and arguing for the
philosophical appeal of a view on which laws state explanatory grounds, I focus mainly on exploring the
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

In section 2 I present the textual case that nomic necessity is not formal necessity.
I also consider several other candidate conceptions of nomic necessity and argue that
none of them is correct. This clears the ground for my own interpretation, that Kant
holds what is now called an ‘essentialist’ view of laws: laws are grounded in the
essences of empirical natural kinds.3 This entails the following definition of nomic
necessity:
(Nomic necessity) It is nomically necessary that p if and only if p and the fact
that p is grounded in the real essences of empirical natural kinds (e.g., matter,
water, etc.).
I begin my argument in section 3 with an account of the Kantian distinction between
real and logical essence. Then, in section 4, I offer a close reading of the Preface to
Kant’s 1786 work Metaphysical Foundations of Natural Science (MFNS), and argue
that he identifies the essences of empirical objects with their ‘formal natures’ and that
laws, on Kant’s view, are grounded in the formal natures (essences) of empirical
objects. This entails my account of nomic necessity. I then explain why this ‘essen-
tialist’ account of laws and nomic necessity is crucial to Kant’s philosophical project
in MFNS. In sections 5 and 6 I diagnose a tension between my ‘essentialist’ concep-
tion of laws (and their necessity), and the a priori ‘constructive’ procedure of MFNS:
how can a priori constructions inform us of the essential properties of matter? In
section 7 I explain how this problem can be solved. In section 8 I return to the texts
that jump-started my interpretation in section 2 and explain them in light of that
interpretation.
First, though, a note about how I am using the term ‘law.’ In some texts, Kant
defines ‘law’ quite broadly; in some places he even calls the formally necessary
principle that all alterations in empirical substances have causes the ‘law’ of causal-
ity.4 In other texts, though, Kant seems to use ‘law’ more narrowly, so that transcen-
dental principles of this kind would not count as ‘laws.’5 In this chapter I investigate
the modal status of the truths ‘intermediate’ between contingent empirical general-
izations (e.g., all dogs have fleas) and formally necessary principles grounded in the

modal status of such laws, and its relation to other kinds of modality in Kant’s system. Watkins (2005)
argues, on the basis of a detailed exegesis of the Analogies of Experience, that Kant conceives of causation
as the exercise of a power in a substance, where that power is part of the substance’s essence (2005), 286–91;
this is very much in line with my ‘causal essentialist’ reading of Kant on laws, although I do not have the
space in this book to ground that claim in a detailed interpretation of the Analogies.
3
See Shoemaker (1980) and (1998), Ellis and Lierse (1994), and Bird (2005). Hawthorne (2001)
contains a very helpful overview of this literature.
4
See esp. the title of the second Analogy in the B edition (B232); also A188/B234, A227–8/B280, A460/
B488, A536/B564, and A636/B664.
5
E.g., B165, A127, and CJ (Ak. 5: 180 and 184). Kant also refers to the metaphysical principles of matter
as ‘laws’; see the ‘Metaphysical foundations of mechanics’ chapter of MFNS where Kant refers to the ‘laws’
of mechanics (Ak. 4: 541, 543, 544). On the distinction between metaphysical and transcendental principles
see the Preface to MFNS (Ak. 4: 469–70), CJ (Ak. 5: 181), and Prol. }15 (Ak. 4: 295).
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

 NOMIC NECESSITY

forms of experience. In some contexts Kant might not call all of these truths laws, but
that is not a serious problem for my project. In section 5 I explore the various kinds of
laws in Kant’s system.

8.2. What Nomic Necessity Is Not

In a variety of texts from the Critical period, Kant makes clear that which causal laws
obtain is not determined by the a priori forms of experience alone. For instance:
The pure faculty of understanding does not suffice, however, to prescribe to the appearances
through mere categories a priori laws beyond those on which rests a nature in general, as
lawfulness of appearances in space and time. Particular laws, because they concern empirically
determined appearances, cannot be completely derived from the categories, although they all
stand under them. Experience must be added in order to know particular laws at all; but about
experience in general, and about what can be cognized as an object of experience, only those a
priori laws offer instruction. (B165; cf. A127)

But there is such a manifold of forms in nature, as it were so many modifications of the
universal transcendental concepts of nature that are left undetermined by those laws that the
pure understanding gives a priori, since these pertain only to the possibility of a nature (as
object of the senses) in general, that there must nevertheless also be laws for it which, as
empirical, may seem to be contingent in accordance with the insight of our understanding, but
which, if they are to be called laws (as is also required by the concept of a nature), must be
regarded as necessary on a principle of the unity of the manifold, even if that principle is
unknown to us. (CJ, Ak. 5: 180)
The understanding is of course in possession a priori of universal laws of nature, without which
nature could not be an object of experience at all; but still it requires in addition a certain order
of nature in its particular rules, which can only be known to it empirically and which from its
point of view are contingent. These rules, without which there would be no progress from
the general analogy of a possible experience in general to the particular, it must think as laws
(i.e. as necessary), because otherwise they would not constitute an order of nature, even though
it does not and never can cognize their necessity. (CJ, Ak. 5: 184)

There is a lot going on in these passages. I want to focus, though, on one claim that
Kant makes in all of them: that the forms of experience underdetermine which
particular causal laws govern empirical objects. Kant is not merely claiming that
for all we know a priori, particular causal laws are not completely determined
by the forms of experience. While that reading may be compatible sensu stricto
with the first passage (from the CPR), it is incompatible with the second two texts.
In the second passage he claims that these laws are “left underdetermined by the
laws that the pure understanding gives a priori,” and in the third text he claims that
they are contingent “from the point of view” of the understanding (a claim he also
makes in the second passage). I take this to mean that these laws are contingent
given the ‘laws’ of the understanding (principles of experience) alone: the
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

principles of experience themselves (in concert with the a priori forms of intuition)
do not necessitate that empirical objects are governed by the laws that actually
obtain. I interpret this to mean that these causal laws are not formally necessary;
because they are not wholly grounded in the a priori forms of experience, the laws
that actually obtain are formally contingent. However, in the underlined portions
Kant also claims that these particular causal laws are necessary. This is a claim
Kant makes in a wide variety of other texts.6 This constitutes, I take it, clear
evidence that, according to Kant’s Critical view, laws are necessary but formally
contingent. Consequently, the necessity of laws (nomic necessity) is not formal
necessity.
There are two ways this claim—that nomic necessity is not formal necessity—can
be taken, namely:
(1) No law is formally necessary.
(2) Not all laws are formally necessary. The necessity that applies to laws as such is
not formal necessity.
I take it that the texts above support attributing (2), at least, to Kant and I will
proceed on that hypothesis. If ‘law’ is taken in the broadest sense, (1) is clearly false.
Kant sometimes refers to a priori formally necessary principles (like the causal
principle) as ‘laws.’7 So I interpret the texts quoted earlier, in which Kant claims
that “particular laws [ . . . ] cannot be completely derived from the categories,” as
making the generic claim that laws are not in general grounded in the forms of
experience; it is compatible with this that some laws, e.g., the metaphysical principles
of matter, are formally necessary, though I will argue that they are not.
What is nomic necessity, if it is not formal necessity? First of all, I think it should
be clear that nomic necessity is not logical necessity. Causal laws are not logically
necessary; the negation of some law that actually obtains does not entail a contra-
diction. Nomic necessity, whatever it is, is a kind of real necessity.8 Nor can nomic
necessity be empirical-causal necessity, because empirical-causal necessity applies
only to alterations in empirical substances, not to the laws that govern those
alterations.9 Nomic necessity, obviously, cannot be noumenal-causal necessity, or
any other kind of necessity grounded in noumena, because we cannot know what is
noumenally-causally necessary and what is noumenally-causally contingent, but we

6
E.g., MFNS (Ak. 4: 468–9), CJ (Ak. 5: 183), and Refl., 5414 (Ak. 28: 176). See also A91–2/B123–4,
A159/B198, and A200/B246. Contra Guyer (1990b), 234, laws are already claimed to be necessary in the
first Critique; on this point, see Walker’s reply to Guyer in the same volume (Walker 1990, 252).
7
See the texts quoted previously.
8
See section 7 for a more complete argument that nomic necessity is a kind of real necessity.
9
Guyer (1990b), 234 also makes the point that (in my terminology) neither formal necessity nor
empirical necessity are nomic necessity.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

can know of many truths (e.g., the truth that a particular event occurs) that they are
nomically contingent.10
Several commentators have attempted to account for the necessity of laws in terms
of systematicity.11 One prominent representative is Paul Guyer. On Guyer’s reading,
laws are necessary in virtue of being parts of hierarchically ordered systems in which
generic higher-order laws subsume and explain specific lower-level laws. 12 However,
systematicity alone cannot explain nomic necessity, for two reasons. First of all, one of
Kant’s main points against Hume is that regularity does not suffice for necessity.13
There is no reason why Hume cannot acknowledge systems of regularities hier-
archically ordered from more general to more specific. If we accept Guyer’s
proposal then Kant is left without a response to the Humean who claims to have
accounted for the necessity of natural laws just as well as Kant has. Secondly,
Guyer’s proposal leaves us without an account of the necessity of the highest-level
laws, for instance, the law of universal gravitation. There is a sense in which a
particular regularity in nature (e.g., the true orbits of the planets) is understood as
non-accidental, and in that sense necessary, when it is brought under a higher-level
law (e.g., the law of universal gravitation). However, the problem we are concerned
with is the necessity of precisely those higher-level laws that, on Guyer’s picture,
‘inject’ the necessity into the whole structure. Guyer might reply that it is the a
priori, formally necessary principles of experience that ‘inject’ this necessity. But
the texts quoted in section 2 show that even the highest-level and most general
natural laws (e.g., the inverse-square law of gravitation) are not necessary in the
same way those principles are, so Guyer’s story leaves us without an explanation of
how these highest-level laws are in some sense necessary but not formally necessary
like the principles of experience. Guyer needs some necessity to attach to the
highest-level natural laws, but does not tell us what that necessity is.
Finally, we might try to understand nomic necessity through the logical function of
assertoric judgment. Kant derives the table of categories from the table of the logical
functions of judgment.14 The modal functions of judgment are problematic, asser-
toric, and apodictic, which correspond, respectively, to the modal categories: possi-
bility, actuality, and necessity. One might, therefore, think that the way to understand
the necessity of judgments (including judgments that state laws) is by understanding
the logical function of apodictic judgment. In the following passage, Kant appears to
equate the logical functions of judgment with their logical-deductive role and
connects this in turn to the modal categories:

10
Which does not mean that we generally know natural laws. See Kreines (2009) for a discussion of the
limits of our knowledge of natural laws.
11
E.g., Kitcher (1986), 209; Buchdahl (1969), 342; and Brittan (1978), 185. Kreines (2009) incisively
critiques the ‘best system’ reading of Kant’s theory of laws.
12
Guyer (1990a), 287–8; see also Guyer (1990b), 238.
13 14
Prol. (Ak. 4: 258). See section 6.7.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

The assertoric proposition speaks of logical actuality or truth, as, say, in a hypothetical
syllogism the antecedent in the major premise is problematic, but that in the minor premise
is assertoric and indicates that the proposition is already bound to the understanding accord-
ing to its laws; the apodictic proposition thinks of the assertoric one as determined through
these laws of the understanding itself, and as thus asserting a priori, and in this way expresses
logical necessity. (A76/B101)

As I understand this passage, Kant is ascribing logical functions to judgments in

virtue of their role in a syllogistic form of reasoning, e.g.,
(1) p ⊃ q
(2) p
(3) ∴ q
The middle premise, (2), Kant claims, is assertoric and therefore represents its
content as ‘actual.’ The very same content, p, is the antecedent in the hypothetical
first premise, and thus is represented as merely possible; (1) represents p as possible
and states the consequence of its actuality, q. Kant describes (3) as apodictic because,
in the context of the syllogism of which it is the conclusion, it represents q not merely
as true but as true in virtue of a rule, (1), thus, non-accidentally true. Consequently,
Kant claims, q is represented as necessary.15
But this should caution us in trying to model nomic necessity on apodictic
judgment and Kant’s association of that form of judgment with necessity. Natural
laws correspond, not to the conclusions of syllogisms, but to their major premises,
the rule (1) from which the conclusion (3) is deduced by subsuming (2) under the
condition (antecedent) of (1). Put in less Kantian terms, by associating apodictic
judgment with the category of necessity, Kant is giving an account of the necessity
of judgments in terms of their roles in logical-deductive structures, but it is an
account on which necessity consists in being the conclusion of a syllogism. This is
completely unsuited to accounting for nomic necessity, because the laws in ques-
tion are precisely the rules from which particular states of affairs (e.g., the true
motions of bodies) are derived as conclusions. The necessity of laws is not the
necessity of conclusions, but the necessity of the premises from which these
conclusions are derived. In a similar spirit, Michael Friedman offers a sophisticated
and intricate reading of Kant’s use of the modal categories in the Phenomenology
section of MFNS.16 For reasons of brevity, though, I defer my critical engagement
with that part of Friedman’s reading to an Appendix that can be found on my website
(see the Preface); I discuss Friedman’s reading of the ‘constructive’ procedure of MFNS
in section 5 below.

15
See Leech (2012) for a nuanced analysis of Kant’s views on the modal functions of judgment;
Rosenkoetter (2013) argues for a quite different reading of the modal functions.
16
Friedman (1992a), 159–64 and Friedman (2013), 531–61.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

8.3. Essences: Real and Logical

In the next section I will argue for my ‘essentialist’ interpretation of nomic necessity.
But first we must understand what essences are for Kant. That is the task of this
section.
Throughout his lectures, Kant distinguishes between (i) the logical essence of a
concept and the real essence of the object of that concept and between (ii) nominal
definitions and real definitions. The former distinction typically appears in his
lectures on metaphysics;17 the latter, in his lectures on logic.18 However, these
distinctions are correlative: a nominal definition reveals a logical essence and a real
definition reveals a real essence.
The best place to start understanding the distinction between logical and real
essence is this long passage from the L2 lectures on metaphysics:
The concept of essence belongs properly in logic. Essence is either a logical essence or a real
essence. A logical essence is the first ground of all logical predicates of a thing; a real essence is
the first ground of all determinations of a thing. For an essence is either logical or real. We posit
a logical essence through the analysis of the concept. The first ground of all predicates thus lies
in a concept; but that is not yet a real essence. E.g., that bodies attract belongs to the essence of
things, although it does not lie in the concept of the body. Accordingly, the logical essence is
the first inner ground of all that which is contained in the concept. But a real essence is the first
inner ground of all that which belongs to the thing itself [was der Sache selbst zukommt]—If I
have the logical essence, I still do not yet have the real essence. In metaphysics, essence should
never be understood as logical essence, for this belongs in logic. [ . . . ] Predicates belonging to
the essence, but only as a consequence, are called attributes; what on the other hand belongs to
essence as a ground is called an essential property. [ . . . ] The real essence is not the essence of
the concept, but rather of the thing [sondern der Sache]. E.g., the predicate of impenetrability
belongs to the existence of body. Now I observe through experience much that belongs to its
existence; e.g., extension in space, resistance against other bodies, etc. Now the inner ground of
all this is the nature of the thing. We can infer the inner principle only from the properties
known to us; therefore the real essence of things is inscrutable to us, although we cognize many
essential aspects. We become acquainted with the powers of things bit by bit in experience.
(ML2, Ak. 28: 553)

Kant makes a series of claims about logical and real essences in this passage, which
I will unpack and analyze.
First, he distinguishes between the logical essence of a concept and the real essence
of a thing (“a real essence is the first inner ground of all that which belongs to the
thing itself ”). He identifies the logical essence of a concept as the complete set of

17
In the metaphysics lectures, MH (Ak. 28: 49), MV (Ak. 28: 411), MvS (Ak. 28: 492), ML2 (Ak. 28:
553), and MD (28: 629).
18
In the logic lectures, LB (Ak. 24: 113–18, 268–73), WL (Ak. 24: 838–40, 913–25), LDW (Ak. 28:
727–9, 756–60), JL (Ak. 9: 61, 140–5), LP (Ak. 24: 408–9, 456–9), LBu (Ak. 24: 634–5, 656–60), and LPö
(Ak. 24: 535, 573–5).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

marks contained in the concept. The logical essence of a concept is specified in its
nominal definition; e.g., the nominal definition of <gold> might be: <yellow>, <mal-
leable>, and <metal>.19 Kant thinks that nominal definitions are relatively trivial for
empirical scientific purposes. It may be important to make sure we all mean the same
thing by our words—e.g., that we are all using ‘gold’ to denote yellow, malleable
metal—but this does not by itself tell us anything substantive about gold, the stuff our
concept refers to.20
A real essence, by contrast, is the essence of the thing picked out by a concept. The
real essence of a thing is given by its real definition, and real definitions are
definitions of things [Sach-Erklärungen],21 not of concepts. Real definitions are not
of individual things, however, but the kinds that these concepts pick out; for instance,
the real definition of gold specifies the real essence of the kind gold rather than <gold>
(the concept) or an individual sample of gold. Kant denies that we can give real
definitions of individual things.22 Kant also denies that we can know the complete
real definition, hence the complete real essence, of kinds of empirical object. At this
point, though, we are interested in what real essences are, according to Kant, not
whether we know them.
Kant’s definition of real essence in the L2 text (“the first inner ground of all that
which belongs to the thing itself ”) is idiosyncratic relative to other texts, because it
does not make explicit the modal dimension of real essence. More typical is his
definition in the Mrongovius lectures: “the first inner ground of all that belongs to the
possibility of a thing is its [real] essence” (Ak. 29: 820).23 But even this more typical
formulation poses an interpretive challenge. It suggests that the real essence of x is the
ground of the real possibility that there is x, but, as we have seen, the grounds of the
real possibility of x are going to be something other than x itself (e.g., the forms of
experience, the laws and the past, etc.). In a pre-Critical context we might even
expect, on the basis of this definition alone, that the real essence of any finite being is
God! However, Kant thinks of the real essence of x as the ‘inner character’ of x that
explains its manifest character and its relation to other things, so the real essence is
not some entity distinct from x (e.g., God). For instance, in the example from the L2
lectures above, Kant describes the real essence of matter as including attractive
force. So Kant thinks of the ‘first inner ground of all that belongs to x itself ’ and
‘the first inner ground of all that belongs to the possibility of x’ as somehow being
properties or characteristics of x, the inner characteristics of x that make it the
possible being it is.24 My interpretation of this is: where K is a kind (the appropriate

19
Prol. (Ak. 4: 267).
20
LB (Ak. 24: 116, 271), LDW (Ak. 24: 757), and WL (Ak. 24: 839, 918).
21
JL }106 (Ak. 9: 143); cf. WL (Ak. 24: 839) and LDW (Ak. 24: 757). 22
LB (Ak. 24: 118, 268).
23
See LDW (Ak. 24: 760), LP (Ak. 24: 456), JL, }106 (Ak. 9: 143, 144), WL (Ak. 24: 919, 920), Refl. 2916
(Ak. 16: 574–5), and Refl. 2995 (Ak. 16: 607).
24
In some texts Kant defines a real definition of a thing as providing marks sufficient to distinguish it
from all other possible things; in other words, the real essence (what is specified in a real definition) makes
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

x is an instance of kind K Fig. 8.1. The real essence of K

grounds

x possesses [P1 + P2 + P3+ . . .]

(the real essence of K)

target of a real definition) and x is a possible instance of that kind, the real essence of
K is the complex of properties possessed by x that ground x’s being an instance of
kind K. In other words, the real essence of water is the complex of properties that
make it the case, of any possible sample of water, that it is a sample of water (see
Fig. 8.1).
What does ‘grounding’ mean here? Clearly, possessing the essential properties of
matter does not cause an object to be matter. Nor is it logical grounding. If possession
of the real essence of kind K logically entails being an instance of K there is no
distinction between logical and real essence; the point of Kant’s distinction between
logical and real essence is that possession of the marks sufficient to logically entail
membership in a kind (the logical essence) is not the same as its real essence. To
return to a theme from the previous chapter (7.3), I think this is a non-logical non-
causal asymmetric real grounding relation between mutually necessarily entailing
propositions (necessarily, something is matter if and only if it has the essence of
matter); to use Kant’s terminology, the possession by x of the properties constituting
its real essence is the ratio essendi of its being an instance of kind K. This is only to
put a name and a structure on this relation; it is not yet even an attempt at a reductive
analysis. Over the course of the chapter I will attempt to see what we can say about
this real grounding relation and what connections it might have to the real grounding
relations from Chapter 7.
Another important feature of Kant’s theory of real essence is his distinction
between the properties that compose the essence (essential properties) and the
properties that are grounded in the essence but not contained in it (attributes).25
As Kant says in the quoted passage: “predicates belonging to the essence, but only as a
consequence, are called attributes; what on the other hand belongs to essence as a
ground is called an essential property.” The essential property/attribute distinction is
important because not all laws will be ‘parts’ of essences. In general, that a law holds

a possible thing the possible thing it is; see A241–2 n., LB (Ak. 24: 658) and Refl. 2992 (Ak. 16: 606).
However, there are other passages in which Kant appears to equate that ‘differentiation’ model of real
essence with the ‘ground of possibility’ model (from the main text), or introduces one within the space of a
page of the other, which strongly suggests he thinks of them as equivalent; see e.g., LDW (Ak. 4: 760), LP
(Ak. 24: 473), and WL (Ak. 24: 919, 920). I take this to mean that ‘the first inner ground of all that belongs
to the possibility of x’ is the set of marks that make x the possible being it is.
25
The terminology comes from Baumgarten, who distinguishes essential predicates (essentialia),
predicates entailed by essential predicates (attributa) and accidents (Meta. }}39, 41, 50); Kant takes over
these distinctions almost verbatim in his metaphysics lectures: MV (Ak. 28: 411), MvS (Ak. 28: 492–3), MP
(Ak. 28: 552–3), MD (Ak. 28: 629), and MM (Ak. 29: 820).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

of a certain kind will be an attribute of that kind, not part of the essence. For instance,
take some very complicated mathematical consequence of the inverse-square law of
gravitation for systems of n bodies (a theorem of Newtonian dynamics); it is
implausible that this law is literally a part of the essence of matter, but more plausible
that it is grounded in that essence. This raises, of course, the question of the
grounding relation between an essence and its attributes; I return to this issue in
section 7.
Some empirical concepts are what I will call empirical kind concepts: they apply
to objects in virtue of those objects belonging to the same kind (e.g., gold, water,
etc.). This is liable to provoke the objection that it is anachronistic to attribute to
Kant the view that there are empirical kinds at all, or that the function of empirical
concepts is to pick them out. Consider, however, the following passage from the
logic lectures:
Can concepts of experience be defined analytically? When the inquirer into nature defines
water, e.g., as a fluid body without taste or color, one readily sees how precarious the definition
is. He who is not already acquainted with water will not thereby become acquainted with it. It is
simply not necessary to define concepts of experience per analysin [nominal definition—NS],
however. For why do I need such a definition? For when I say the word water, others
understand me [ . . . ] For in the concept water there lies so little that I immediately go outside
the concept and have to collect new marks through experience, i.e. I have to define the concept
through exposition synthetically [partial real definition—NS], and not analytically. (WL, Ak.
24: 918–21)

Kant is claiming here that the nominal definition of empirical concepts like <water>
is relatively insignificant for scientific purposes. The concept contains only some
commonly observable marks, so that we can verify that we are all talking about the
same thing (e.g., “the clear drinkable stuff in lakes and rivers”). The goal of scientific
inquiry is to determine what all samples of water as such have in common, what
makes them water, that is, the real essence of water. This requires that we use the
concept <water> not to pick out everything that has the marks contained in its logical
essence (because the nominal definition is dispensable, as Kant here claims) but to
pick out objects in virtue of belonging to the kind water, that is, everything that has
the marks contained in its real essence of water. Not all empirical concepts are like
this. It is implausible, for instance, that weeds share a real essence; all there is to being
a weed is our nominal definition of the concept <weed> because this is a concept
arbitrarily invented by us. Concepts like <water>, that are used to pick out a kind in
virtue of a shared real essence, I will call ‘empirical kind concepts’; Kant calls them
‘given’ empirical concepts.26 Kant does not think we can ever discover the complete
real essence of any empirical kind, so we can never give the complete real definition
of an empirical natural kind. But he does think that in natural science we can

26
LB (Ak. 24: 272), WL (Ak. 24: 838–40), LDW (Ak. 24: 727–9), and JL (Ak. 9: 142).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

determine some of the essential properties and attributes, as he makes clear in the L2
passage quoted earlier: “therefore the real essence of things is inscrutable to us,
although we cognize many essential aspects.”

8.4. Essences, Natures, and Laws

If we want to understand Kant’s views about essences and laws, one natural place to
turn is MFNS. Kant begins the Preface to that work by distinguishing the ‘formal’ and
‘material’ meanings of the word ‘nature’:
If the word ‘nature’ is taken simply in its formal meaning, where it means the first inner principle
of all that belongs to the existence of a thing*, then there can be as many different natural sciences
as there are specifically different things, each of which must contains its own proper [eigenthüm-
liches] inner principle of the determinations belonging to its existence. But ‘nature’ is also taken
otherwise in its material meaning, not as a constitution, but as the sum total of all things, insofar as
they can be objects of our senses, and thus of experience. [ . . . ] (MFNS, Ak. 4: 467)

The formal sense of ‘nature,’ I take it, is the sense involved in claims of the form ‘it is
in the nature of ç to c’ (e.g., it is in the nature of gold to melt at 1,064 C), while, in
the material sense, ‘Nature’ refers collectively to all objects of sensible intuition (both
inner and outer).27 For ease of reference, I will capitalize ‘Nature’ when it refers to
nature in the material sense. The asterisk in this passage indicates a footnote, where
Kant clarifies his meaning:
Essence is the first inner principle of all that belongs to the possibility of a thing. Therefore, one
can attribute only an essence to geometrical figures, but not a nature (since in their concept
nothing is thought that would express an existence). (MFNS, Ak. 4: 468 n.)28

‘Existence’ here means being causally efficacious, either as a substance or the acci-
dents inhering in a substance due to its interactions with other substances. This is the
sense of ‘existence’ in which the Analogies of Experience express conditions on the
possibility of representing existing objects.29 As Kant claims here, there are objects
that do not exist in this technical sense of ‘existing,’ which we can think of as
equivalent to being concrete in the contemporary sense. I will refer to this notion of
existence as existence*. In the final chapter of this study I will explain Kant’s Critical
theory of existence, existence*, and actuality, but for now I just want to note that, by
admitting that existence* (concreteness), is a ‘real predicate’ (one that applies to a
proper subset of the objects there are), Kant is not opening space for an ontological

27
Cf. the discussion of the formal sense of nature in the footnote to A418/B446, Prol. }14–17 (Ak. 4:
294–7), and GMM (Ak. 4: 437). The formal meaning of ‘nature’ is clearly a descendant of Aristotle’s
definition of nature; see Physics 2.1, 192b20–23.
28
Cf. the discussion of nature and essence in the metaphysics lectures: MH (Ak. 28: 49), MV (Ak. 28:
411), MvS (Ak. 28: 492), ML2 (Ak. 28: 553), and MD (Ak. 28: 629).
29
See section 6.6, and A160/B190, A176, B219, A182, B225, B233, B257, and A215–16/B262–3.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

argument; he can freely admit that God exists* is an analytic judgment (if there is a
God, he is concrete), while denying that there is a God (that there is an object
instantiating the concept <existent* God>) is analytic. The point of the quoted
passage is that only existing* objects have what Kant calls ‘natures.’
In the first sentence of the Preface, quoted above, Kant claims there can be as many
different sciences as there are “specifically different things” that have a nature; I take
this to refer to generic kinds rather than individual things, kinds constituted by things
that share a common nature. Since only existing things have a nature, we might
interpret this notion of (formal) nature as follows:
(Nature*) If ç is a kind of existing* thing, then property c is part of the nature of
çs if and only if necessarily all çs have c.
But this is inadequate, because Kant asserts an explanatory connection between the
nature and “what belongs to” the existence of the kind: the nature is “the first inner
principle of all that belongs to the existence of a thing.” The nature is a “first principle”
because it explains the properties that things of that kind necessarily have. I propose
instead:
(Nature) If ç is a kind of existing* thing, the nature of ç is the complex of
properties the possession of which by any possible instance of ç grounds its being
ç and existing*.
The nature of a kind is the set of properties such that instances of the kind are instances
of the kind in virtue of having those properties. Since natures are defined for existing*
things, possessing the properties that compose a nature grounds not only membership
in the kind, but existence (causal effectiveness) as well. The nature of a thing is “the first
inner principle of all that belongs to the existence* of a thing” because its nature is the
set of properties in virtue of which it exists* as the kind of thing it is.
Kant begins his footnote about formal nature by distinguishing it from (real)
essence—non-existing* mathematical objects have essences, but not natures—so we
might wonder about the relation between these notions. It is clear that anything that is
part of the nature of ç (if ç is a kind of existing thing) is also part of its real essence: if a
set of properties grounds being ç and existing*, then (trivially) it grounds being ç. What
is more, the essence of ç is identical to the nature of ç. If ç is a kind of existing* thing
(concrete thing) then it is essentially a kind of existing* thing; existing* is either part of
the essence of ç (essential property) or it is grounded in that essence (attribute). If some
sample possesses the properties that constitute the real essence of ç then it possesses
properties sufficient to ground its existence*; so the real essence of ç grounds not only
its being ç but its existing* as well.30 Consequently, the real essence of ç is its nature.

30
Note, though, that this is the narrow conception of existence (=concreteness), not the ‘absolute
positing’ (quantificational, non-nuclear) conception of existence, so to say that ç essentially exists* does
not entail that there is an ontological argument for there being any ç (or any of its instances).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

(Nature=essence) If ç has a nature (is a kind of existing* thing) then the nature of
ç = the real essence of ç.
For instance, take ç to be a kind of physical substance, like molybdenum. Clearly,
molybdenum is a kind of thing that exists*. The properties in virtue of which a
sample of molybdenum is a sample of molybdenum, are also properties that
ground its being causally efficacious (existing*). So the essence of molybdenum
is the nature of molybdenum.31 This interpretation is confirmed by Kant’s
metaphysics lectures, where he repeatedly equates nature and real essence (for
existing* things) e.g., “this real essence is nature” (MM, Ak. 29: 821).32 In section
3 I defined empirical kind concepts as concepts that apply to objects in virtue of
belonging to a kind with a common essence; since essences for existing* things are
natures, we can also call these ‘empirical natural kind concepts’: concepts that
pick out objects in virtue of sharing a (formal) nature.33 Since I will almost
exclusively be discussing existing* things I will not always make this explicit.
Likewise, I will often refer to real essence as essence simpliciter because logical
essences will not play much of a role in my argument; when I mean logical essence
I will make this explicit.
I have analyzed the notions of essence and nature in such detail because Kant’s
definitions of ‘law’ in MFNS depend on the formal concept of nature:

Since the word Nature [in the material sense—NS] already carries with it the concept of laws,
and the latter carries with it the concept of the necessity of all determinations of a thing
belonging to its existence [ . . . ] (MFNS, Ak. 4: 468)
For laws, that is, principles of the necessity of that which belongs to the existence of a thing
[ . . . ] (MFNS, Ak.4: 469)

Kant’s definitions of law in these passages (especially in the second) is so close to his
definition of (formal) nature that we might be tempted to identify the nature of a
thing with the laws concerning it, or even to claim that the nature of a thing is
grounded in laws: what it is to be molybdenum, for instance, is to obey such-and-
such laws (e.g., to have a melting point of 2,623 C). However, this would leave us
without an analysis of what it is to be a law, and there is a more faithful reading of the
texts available. There is a subtle difference between Kant’s definition of formal nature
in the very first sentence of MFNS (“the first inner principle of all that belongs to the

31
Failure to recognize this point, I believe, is what leads Plaass to claim, implausibly, that ‘nature in the
formal sense’ as defined in MFNS is distinct from ‘nature in the formal sense’ as defined in other texts (cited
by Plaass himself) where Kant identifies nature with real essence (Plaass (1965), 217 n. 2).
32
MH (Ak. 28: 49), MV (Ak. 28: 411), MvS (Ak. 28: 492), ML2 (Ak. 28: 553), and MD (Ak. 28: 629). See
also the May 12, 1789 letter to Reinhold (Ak. 11: 37).
33
The qualification empirical natural kind concept is important, because it may be that there are a priori
natural kind concepts: substances might form a kind in virtue of the common real essence of substance. But
I want to remain neutral on that.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

existence of a thing”) and these definitions of law: the nature of a thing is the first
principle of what belongs to its existence while these two passages define laws as
necessary conditions on the existence of a thing.34 I take this to mean that the nature
of a thing grounds or explains the laws that are true of it, i.e.
(LawNAT) If it is a law that ç has c then (a) it is true that all samples of ç have c,
(b) ç is a natural kind (<ç> is a natural kind concept), and (c) every sample of ç
has c in virtue of the nature of ç.
Consequently, there can be laws for each kind, and the laws of the more generic kinds
(grounded in the generic nature of the species) will be generalizations that subsume
the laws of the more specific kinds (grounded in the specific difference in nature that
makes that species the species it is). For each such kind and its associated laws there
can be in principle be a science of that kind.35 This is the meaning of Kant’s claim
that “there can be as many different natural sciences as there are specifically different
things, each of which must contain its own peculiar inner principle of the determin-
ations belonging to its existence” (Ak. 4: 467). Given the identity of essences and
natures for existing* things, it follows that:
(LawESS) If it is a law that ç has c then (a) it is true that all samples of ç have c,
(b) ç is a natural kind (<ç> is a natural kind concept), and (c) every sample of ç
has c in virtue of the essence of ç.36
Every law states a truth grounded in the essence of a natural kind. If this view is
correct, there is a natural sense in which all laws as such are necessary: they obtain in
virtue of the essences of natural kinds. Every law is nomically necessary if we define
nomic necessity as follows:
(Nomic necessity) It is nomically necessary that p if and only if p is grounded in
the essences of actual empirical natural kinds.

34
Cf. his remark about nature in the material sense: “this signifies a derivation of the manifold
belonging to the existence of things from their inner principle” (MFNS, Ak. 4: 468).
35
For any nature, there are facts grounded in that nature, e.g., that objects with that nature have the
properties contained in that nature. So, trivially, there will be laws: the law that all objects with that nature
have those properties. However, as Allison (1994), 298 points out, this leaves open whether the laws are
‘one off ’—for there may be only one object with a given nature. In other words, this leaves open whether
there is a different law for every empirical object, and whether there is any systematic unity among the laws.
On my reading, it is the role of reflecting judgment and the regulative use of theoretical reason to justify us
in holding that nature constitutes a system, in particular, that the natural kinds constitute a system and this
‘one-off ’ situation does not obtain. However, interpreting Kant’s idea of nature as a system lies outside the
scope of this book.
36
I have stated these as necessary conditions on lawhood, rather than necessary and sufficient
conditions (or even definitions of lawhood) to make room for the possibility that there may be more to
the Kant’s conception of lawhood, such as the organization of laws into a system of Nature (of natures). Cf.
Appendix to the Transcendental Dialectic, A643/B671–A668/B696.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

My interpretation offers a unified account of Kant’s conception of essence, nature,

law, and nomic necessity.37
Some readers might suspect that this Aristotelian metaphysics of essence and
nature attributes to Kant more metaphysics than he needs in MFNS and more
metaphysics than he is allowed, given his epistemology. On the contrary, the meta-
physics of essences and natures is crucial to his project in MFNS: to provide a
foundation for the rational science of physics (of outer Nature). As he writes in the
Preface,
Any whole of cognition that is systematic can, for this reason, already be called science, and if
the connection of cognition in this system is an interconnection of grounds and consequences,
even rational science. If, however, the grounds or principles themselves are still in the end
merely empirical, as in chemistry, for example, and the laws from which the given facts are
explained through reason are mere laws of experience, then they carry with them no con-
sciousness of their necessity (they are not apodictically certain), and thus the whole of
cognition does not deserve the name of a science in the strict sense [ . . . ] (MFNS, Ak. 4: 468)

A science is a body of cognitions organized according to levels of increasing generality:

a hierarchy of particular truths subsumed under more general laws, which are
themselves subsumed under more general laws, up to and including one or more
fundamental laws that are not subsumed under anything more general. In such a
system, either the lower-level laws ground the higher-level laws, or vice versa.38 In a
rational science, the higher-level laws and principles (partly) ground the lower-level
ones and the lower-level laws (partly) ground the facts at the bottom-level.39 In a non-
rational science, which does not properly deserve the title ‘science’ at all but is merely
an “historical doctrine of nature” (Ak. 4: 468), the higher-level principles are simply
generalizations about the bottom-level facts and, as such, are grounded in those facts.
For instance, what Kant calls ‘natural description’ of objects is not a rational science
because its highest-level principles are simply classifications grounded in similarities

37
According to Watkins (1997), it was commonly held in the eighteenth-century German rationalist
tradition that the laws of motion are grounded in the essence of matter; I am arguing that Kant held this
view for laws in general and the natural kinds they concern.
38
This is my gloss on Kant’s claim that the principles that “order” a science are “either principles of
empirical or rational connection of cognitions into a whole” (MFNS, Ak. 4: 468). Kant does not say that the
principles of the science are either empirical or rational; he says the principles according to which the
science is organized are either empirical or rational. A science organized according to empirical principles,
I think, is a science in which general principles are a posteriori grounds of principles of lower generality; the
truths of lower generality ground the truth of the general principles, but the general principles are a ground
of knowing the more particular, less general truths. For instance, if all doves are white, this is made true by
all of the individual white doves; but the fact that all doves are white is a ground of knowing, of each
individual dove, that it is white. By contrast, in a rational science of ornithology the claimed ground of the
whiteness of doves would be the nature of doves.
39
The qualification ‘partly’ is crucial here because the nature of the genus (e.g., matter) does not wholly
ground the laws of its species (e.g., water), for each species has a specific difference in its nature that
grounds specific laws for that kind, which do not apply to matter as such (e.g., the law that water boils at
100 Celsius).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

among individual objects. Because, by definition, Nature is a domain of objects

governed by necessary laws, any science of Nature as such must be a rational science;
a mere ‘natural description’ of Nature would not be a science of Nature as Nature (as a
law-governed totality). Kant further requires of a rational science that its most
fundamental, highest-level laws are known to be necessarily true. A putative science
whose most fundamental laws are known merely through empirical generalization
(as, Kant claims, is the case with chemistry as it currently stands)40 is not really a
rational science, for we cannot know through experience whether those generaliza-
tions are necessary.41 Consequently, Kant’s characterization of laws as “principles of
the necessity of that which belongs to the existence of a thing” is crucial to his
understanding of a rational science of Nature.42
Furthermore, Kant regards these essentialist claims as scientifically necessary.
Consider his only explicit criticism of Newton in MFNS:
It is commonly supposed that Newton did not find it necessary for his system to assume an
immediate attraction of matter, but with the most rigorous abstinence of pure mathematics
allowed the physicists full freedom to explain the possibility of attraction as they might see
fit [ . . . ] But how could he ground the proposition that the universal attraction of bodies [ . . . ] is
proportional to the quantity of their matter, if he did not assume that all matter as matter,
therefore, and through its essential property, exerts this moving force? [ . . . ] [H]e could by no
means say that the attractive force of two planets, those of Jupiter and Saturn, for example,
manifested at equal distance of their satellites (whose mass is unknown), are proportional to the
quantity of matter of these heavenly bodies, if he did not assume that they attracted other matter
merely as matter, and thus according to a universal property of matter. (MFNS, Ak. 4: 514)

He refers here to the derivation of the relative masses of Jupiter, Saturn, and the Earth
in the Corollaries to Principia, Proposition VII, which relies on the assumption that
every body exerts an attractive force on every other body (not just on its satellites).43
Kant claims that Newton would not be justified in claiming that all bodies as such
attract all other bodies without the assumption that such a force is an essential
property of matter. But why is that? Why can’t Newton assume that it is a fact,
perhaps even a law, that all bodies attract one another without assuming that this is
essential to matter? My account delivers a straightforward answer: if this fact is not
grounded in the essence of matter then it is merely a generalization about matter and
hence is not a law and hence not part of a rational science of matter that explains the
motions of bodies (rather than merely recording true generalizations about them).

40
MFNS, Ak. 4: 468. Kant also calls chemistry a “systematic art” (Ak. 4: 469; cf. 470 f.), by which I take
him to mean that chemistry tells us how to produce various items (e.g., salts) but does not explain why
these procedures succeed (it is not a prioriG).
41
“They carry with them no consciousness of their necessity (they are not apodictically certain), and
thus the whole of cognition does not deserve the name of a science in the strict sense” (MFNS, Ak. 4: 468).
42
On this point, I differ from Konstantin Pollok who downplays the importance of the formal concept
of nature for the project of MFNS; see Pollok (2001), 46.
43
I draw here on the discussion of these issues in Friedman (1992a), 153–9.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

For Newton’s theory to be genuinely explanatory and hence scientific (in the strict
sense) it needs to assert that the force it attributes to bodies is essential to them.
Consequently, Kant’s account of how that science is possible in MFNS must be able
to rationally ground that essentialist assertion.

8.5. Essentialism and Constructivism

in Metaphysical Foundations
Up to this point I have discussed ‘laws’ in general, but Kant’s system includes several
different kinds of laws corresponding to different sciences. Figure 8.2 summarizes the
relations among the most important Kantian sciences and their respective laws.
Transcendental philosophy is the science of the principles that govern any possible
object of cognition; the critique of our cognitive faculty that prepares the way for the
complete system of transcendental philosophy is the Critique of Pure Reason itself.44
Objects of cognition divide into two kinds, objects of inner sense and objects of outer
sense, corresponding to two different regions of Nature and two different sciences:
the science of inner Nature (psychology) and the science of outer Nature (physics).
Kant argues in the Preface to MFNS that there can be no science of psychology, so the
only possible science of Nature is physics.45 The metaphysical principles of physics
(outer nature) are more determinate than transcendental principles restricted to
outer objects, because metaphysics concerns an empirical concept, albeit the most
general empirical concept of outer objects whatsoever: <matter>.46 All outer objects
fall under the concept <matter> and the metaphysical principles of physics are the a
priori principles that make possible experience of matter. For instance, the transcen-
dental principle that all alterations have a cause has a more determinate metaphysical
correlate: all alterations in matter must have an external cause (matter cannot move
itself).47 Another important metaphysical principle of matter, derived by Kant in the
Dynamics chapter of MFNS, is that all matter has attractive and repulsive forces.
What I am calling ‘rational physics’ contains the a priori consequences of these
metaphysical principles, for instance, the principle that there is some point at which
the attractive and repulsive forces of a body cancel one another. More determinate
than these principles of rational physics are the laws of empirical physics, the a
posteriori laws that apply to all matter as such. For instance, if the inverse-square law
of gravitation is a posteriori, then it belongs here, because it is a more determinate
form of the a priori metaphysical principle of universal attraction and does not follow

44 45
Axxi, Bxxii. MFNS, Ak. 4: 470.
46
MFNS, Ak. 4: 469–70. Cf. CJ, Ak. 5: 181 on the distinction between metaphysical and transcendental
principles.
47
MFNS, Ak. 4: 543; this is precisely Kant’s example of a metaphysical specification of a transcendental
principle in the CJ, Ak. 5: 181. For critical discussion, see Westphal (2004), 205–27.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

Transcendental Philosophy
Object = Nature (all objects of possible
experience, inner and outer)
Formally
A Priori
Pure (no empirical concepts) Necessary

A. Metaphysics of Outer Nature

Object = Outer Nature
A Priori
Depends upon the highest empirical concept of
outer sense: matter
A Priori rules for constructing experience of true
motions from perceptions of apparent motions
A Priori

B. Rational Physics
A Priori consequences of A laws
Subsumes the more determinate empirical laws
of matter as such (empirical physics)

C. Empirical Physics
A Posteriori laws of all matter as such
E.g., inverse-square law of gravitation

D. Laws of Kinds of
Matter
Different laws for
different kinds of K1 K2 . . . . K Nomically
matter (e.g., acids, Necessary
alkali, etc.)
E.g., chemical laws
Science properly so-
called if and only if K K
laws are constructions
of empirical concepts
or a priori
consequences thereof.
Kinds are divided into
K K
species and genera .
that form a system. .
.
Fig. 8.2. The Kantian system of sciences
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

a priori from the principles of rational physics (by assumption).48 Finally, below
empirical physics we have various sciences of more determinate kinds of matter
(since all outer objects are matter, all kinds of outer objects are kinds of matter); if
chemistry can be a science (in the strict sense) and not a mere ‘natural description,’ it
would belong here.49 Kant refers to the principles of each of these sciences as laws; in
some contexts, he even refers to the principles of transcendental philosophy (e.g., the
causal principle) as laws. My focus in this chapter has been on A-D laws, the laws that
are more determinate than the formally necessary principles of transcendental
philosophy. It is a commitment of my interpretation that the laws of each of these
sciences is grounded in the nature/essence of their respective domains: A-C laws are
grounded in the essence of matter, while D-laws are grounded in the essences of the
relevant empirical natural kind (e.g., various chemical kinds like acid or salt).
However, this division of Kantian sciences poses a potential problem for my
essentialist interpretation: there is a tension between (what I claim are) Kant’s
essentialist conclusions and the methodology he uses to prove them in MFNS. Put
roughly, Kant’s method in MFNS is to describe the a priori rules we apply in
constructing experience of the true motions of bodies from perceptions of their
apparent motions and thus derive the A-laws and some B-laws (though not all of
rational physics). I will call this the ‘constructivist’ method of MFNS. But how can
Kant, merely by elucidating the rules of a constructive procedure, determine anything
about the real essence of matter, much less determine that real essence a priori?
This poses, in an acute way, the relation of my interpretation to that of Michael
Friedman, who develops a compelling interpretation of Kant’s a priori method for
establishing A-laws and (some) B-laws in his (1992a). What is more, Friedman’s
reading delivers a unified view about the modal status of A- and B-laws; in my
terminology, they are not formally necessary sensu stricto, but (as I explain below) the
consequents of formally necessary conditionals. However, I think I can retain within
my larger essentialist view these merits of Friedman’s interpretation. In particular,
I will argue that my view delivers (i) a unified account of the necessity of all laws,
which (ii) explains why some laws are a priori knowable (A- and B-laws), and (iii)
delivers a better understanding of the relation between formal necessity and nomic
necessity.50 First, I will explain what I mean by the ‘constructivist’ method of MFNS
(on which I largely agree with Friedman 1992a), and then I will explain how this can
be embedded within my essentialist interpretation of laws.

48
The epistemic status of the inverse-square law is problematic. In several texts, Kant appears to flirt
with the idea of an a priori proof of it (e.g., Prol. }38 and MFNS, Ak. 4: 518–21), but backs off from
endorsing such a proof. See Friedman (1992a), ch. 4 for a detailed and, to my mind, convincing case that
Kant is not endorsing a purely a priori derivation of the inverse-square law in these passages.
49
Kant holds open the possibility that chemistry might become a science (in the strict sense) if chemical
concepts can be “constructed” (Ak. 4: 471); I discuss this further in }8.
50
In the Appendix on Friedman (see my website) I argue that Friedman does not deliver (i).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

In the Preface to MFNS Kant describes the method of proof he will deploy in the
rest of the book:
But in order to make possible the application of mathematics to the doctrine of body, which
only through this can become natural science, principles for the construction of the concepts
that belong to the possibility of matter in general must first be introduced. (MFNS, Ak. 4: 472)

The key term here is ‘construction,’ but Kant uses that crucial term in two incom-
patible ways in MFNS. In some passages, he claims that no empirical concept
(e.g., <matter>), nor any concept that depends upon an empirical concept (e.g., the
concept <matter in motion>), can be constructed, because construction only
occurs in pure intuition and pure intuition is independent of any experience;51 but
in others he talks about the construction of empirical concepts like <matter> and
<motion>.52 I think the solution rests in recognizing that in MFNS ‘construction’ can
have two senses, which I will call ‘mathematical construction’ and ‘metaphysical
construction.’53
Mathematical construction. Mathematical construction is construction in pure
intuition. To construct a concept is to prove that an object that instantiates that
concept can be presented in intuition. Because mathematical construction is pure
and a priori, it does not depend upon the sensory content of experience; whatever can
be mathematically constructed, can be mathematically constructed regardless of how
we are affected. Consequently, no empirical concept can be mathematically con-
structed, for our possession of empirical concepts depends upon the matter of
experience. Most importantly, in the context of MFNS, the concept <matter> cannot
be mathematically constructed. Nor can any concept of a force be constructed.54
In general, no concept of an existing* thing is mathematically constructible, because
our ability to represent existing* things depends upon the sensory content of
experience.55
Metaphysical construction. Metaphysical construction, unlike mathematical con-
struction, is always construction of an empirical concept. To construct an empirical
concept is to elicit the a priori rules we employ in transforming perception of the
object of that concept into experience of that object. To recall the discussion of

51
Cf. MFNS, Ak. 4: 469, and 487–8.
52
Cf. MFNS, Ak. 4: 470, 472, 517, 518, 534, and 549.
53
Kant himself suggests this distinction when he writes: “for this purpose I have considered it necessary
[to isolate] the former [empirical natural science] from the pure part of natural science (physica generalis),
where metaphysical and mathematical constructions are customarily run together” (MFNS, Ak. 4: 473). It
is extensively developed in Plaass (1965), 272–81; my analysis owes much to his. Buchdahl (1992), 231–5,
also distinguishes two kinds of constructions. However, I would not rest my case for the distinction on this
passage, but on the sense it allows us to make of MFNS as a whole.
54
See Westphal (1995), }II. I have also learned a great deal from Westphal (2004)’s meticulous analysis
of the arguments of MFNS and their relation to the CPR.
55
“For laws, that is, principles of the necessity of that which belongs to the existence of a thing, are
concerned with a concept that cannot be constructed, since existence cannot be presented a priori in any
intuition” (MFNS, Ak. 4: 469). “Existence” here refers, I take it, to existence*.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

Chapter 7.3, the content of a perception records merely how the object appears to the
subject having the perception.56 Experience of an object, by contrast, is objectively
valid. The problem of the (metaphysical) construction of matter is the problem of
determining how we transform our merely subjectively valid perceptions of matter
into objectively valid experience of matter.57 I introduced the topic of construction in
MFNS in order to understand Kant’s claim in the Preface that “principles for the
construction of the concepts that belong to the possibility of matter in general must
first be introduced.” This is a case of metaphysical construction: Kant means that
principles for the construction of the experience of bodies from perceptions of bodies
must be introduced, in order for there to be a rational science of physics.58
My explication of the notion of metaphysical construction rests on the distinction
between perception and experience. Experience is a whole of perceptions, synthesized
by a priori rules.59 The perceptions that constitute experience also include the
perceptions of other subjects. Consider the important distinction drawn in the
Prolegomena between judgments of perception and judgments of experience:
All of our judgments are at first mere judgments of perception; they hold only for us, i.e., for
our subject, and only afterwards do we give them a new relation, namely, to an object, and
intend that the judgment should also be valid at all times for us and for everyone else; for if a
judgment agrees with an object, then all judgments of the same object must also agree with one
another, and hence the objective validity of a judgment of experience signifies nothing other
than its necessary universal validity. (Prol., }18, Ak. 4: 288)

I interpret this as a distinction between judgments of how things seem to me

(judgments of perception), which may not be valid for other subjects, and inter-
subjectively valid judgments about how empirical objects actually are (judgments of
experience). Kant defines judgments of experience here in terms of their “necessary
universal validity,” which I take to mean: intersubjective validity. When I make a
judgment of experience, I am making a valid claim to the agreement of all other

56
However, the perception/experience distinction functions differently in MFNS than it does in the
CPR. In the CPR it is the distinction between empirical cognition of mere sensory manifolds (perception)
and combining them into representations of publicly available existing* objects (experience—see A176,
B219, A182, B225, B233, B257, and A215–16/B262–3). The objects of MFNS (matter and its motions) are
already publicly available existing* objects, so the MFNS distinction is a distinction within experience in the
CPR sense: experience of relative motion (perception, in MFNS) and experience of true motion (experi-
ence, in MFNS).
57
Here I hew closely to the interpretation of Friedman (1992a).
58
Cf. MFNS (Ak. 4: 510) where Kant says that the impenetrability of matter makes it possible for us to
“transform” our perceptions of outer sense into the concept of an experience (Erfahrungsbegriff ) of “a
matter as an object in general” [eine Materie als Gegenstand überhaupt]. I take this to mean: (represen-
tation of) repulsive force makes it possible to transform our perceptions of matter into experience of
determinate material objects with determinate boundaries, bodies. See later discussion for the distinction
between Materie and Körper in MFNS. Thanks to an anonymous referee for OUP for pressing me to clarify
my interpretation on this point.
59
A110, B161, A176/B218–19, A183/B226, B218–19, A179/B222–3, A221/B269, A225/B272–3, A227/
B279, A230–1/B282–4, A495/B523, and A764/B792.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

cognitive subjects like me: those with discursive spatiotemporal intellects. Kant
identifies this with making a judgment ‘with objective validity,’ which I take to
mean: a judgment about how the object is represented in the unified experience
composed of perceptions of all subjects. A judgment of experience makes a claim to
intersubjective validity because it represents its content as part of the unified experi-
ence constituted by the lawful coherence of all subjects’ perceptions.
To bring this discussion back to MFNS, Kant there distinguishes the appearance of
motion from the experience of motion (Ak. 4: 554). The appearance of motion is
merely the motions that bodies are perceived to have relative to some other body we
designate as fixed. A judgment of perception of motion records the appearance of
motion, or, equivalently, the content of a perception of motion. The possible true (as
opposed to apparent) motions of bodies are the motions it is possible to experience; as
Kant states in the Phenomenology section of MFNS, “matter is the movable insofar as
it, as such a thing, can be an object of experience” (Ak. 4: 554). To make a judgment
of experience of motion is to make a judgment about the true motions of bodies, or,
equivalently, to make a judgment about how their motions are experienced rather
than perceived.
Constructing the concept of the motion of bodies, therefore, is determining the a
priori rules that make it possible for us to transform perceptions of motion into
experience of motion.60 It is equivalent to answering the question, how is it (for-
mally) possible for us to make judgments of experience about the motions of bodies?
In what sense are these rules necessary for constructing the experience of motion?
They are at least partly grounded in our forms of experience. For example, because
we are only ever given empirical spaces, sets of bodies we take to be fixed and relative
to which we define the apparent motions of bodies, we can never experience the
motion of bodies relative to absolute space. For any pair of ‘empirical spaces’ (in
contemporary terms, reference frames) in uniform rectilinear motion relative to one
another, it is impossible for us to experience one of them as stationary with respect to
absolute space. It is impossible for us to make a judgment of experience that the one
empirical space is absolutely at rest while the other is absolutely in motion because
the alternate judgment (or the judgment that both are in uniform rectilinear motion
relative to absolute space) is equally supported by the empirical evidence and our a
priori constructive rules.61 This familiar point about the ‘equivalence of hypotheses’,
when embedded in Kant’s epistemology, entails that absolute space is not an object of
possible experience at all, hence not even an object at all (absolute space cannot be a
noumenon, because “space represents no property at all of things in themselves, nor
any relation of them to each other”—A26/B42), but a mere idea of reason.62
But these a priori laws are not themselves, strictly speaking, formally necessary.
For example, take the laws that Kant derives in the Dynamics chapter of MFNS: all

60
This means that MFNS is a ‘transcendental’ argument. Cf. Watkins (1998a).
61 62
MFNS, Ak. 4: 555. MFNS, Ak. 4: 559.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

matter exerts an attractive and a repulsive force on all other matter. This law cannot
be formally necessary, for if it were, it would be grounded wholly in our forms of
experience, which apply to all outer objects as such. If the laws are grounded in our
forms of experience and apply to all objects of experience as such, this means, in
terms of the distinction drawn earlier, that they are transcendental principles not
metaphysical principles. This would mean that they could be derived from the forms
of experience without using any empirical concept, including the most general
empirical concept in physics, <matter>. This is clearly not Kant’s view for it would
collapse the distinction between transcendental philosophy (CPR) and metaphysics
of nature (MFNS). The a priori metaphysical principles of matter must be at last
partly grounded in the essence of matter.
But the same argument will apply to any law about matter, or any empirical
natural kind, for every empirical natural kind is a kind of matter. No law about
any empirical natural kind is formally necessary. What are formally necessary are
various conditional principles like:
(1) If matter is an object of possible experience, it has its motions relative to a
family of inertial reference frames (it is not absolutely at rest or in motion).
(2) If matter is an object of possible experience, it has attractive and repulsive
forces.
The A- and B-laws (the metaphysical principles of matter and the laws of rational
physics) are the consequents of such conditionals, that is:
(1c) Matter has its motions relative to a family of inertial reference frames (it is
not absolutely at rest or in motion).
(2c) Matter has attractive and repulsive forces.
The tension between Kant’s constructive method and his essentialist conclusion
comes down to this: what warrants Kant in claiming that the consequents of these
conditionals ((1c) and (2c)) are essential to matter?
To take a concrete instance, Kant argues that:
(FC) It is formally necessary that if matter is an object of possible experience it has
attractive and repulsive forces. [Formal necessitation of conditional (2).]
Kant’s argument, in brief, is that we can only experience matter if we experience it as
extended, and this requires experiencing matter that has a determinate location. This
in turn requires experiencing matter that occupies space by repelling other matter
that attempts to move into its space; the other ways that matter could occupy
[einnehmen] space (through absolute hardness, or by mere presence in space)
would violate our forms of experience (absolute hardness)63 or make it impossible

63
MFNS, Ak. 4: 502.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

for us to experience determinate location (mere presence in space).64 If matter has

only repulsive force it will become arbitrarily dispersed as time elapses (its quantity in
a bounded space will become arbitrarily close to zero after a sufficient duration), and
for any finite amount of time, more than that amount of time has dispersed (there is
no first point in time, though there is no ‘completed’ infinity of past times);65 so if
matter had only repulsive forces, for any finite volume, it would be dispersed to a
density arbitrarily close to zero within that volume, making it impossible for us to
perceive and hence to experience matter (for to perceive matter we need to perceive it
within some finite volume).66
But Kant does not merely draw the conclusion that it is formally necessary that if
we can experience matter then it has attractive and repulsive forces; he concludes that
these forces are essential to matter:
A property on which the inner possibility of a thing rests, as a condition, is an essential element
thereof. Hence repulsive force belongs to the essence of matter just as much as attractive force,
and neither can be separated from the other in the concept of matter. (MFNS, Ak. 4: 511)

In case there was any doubt that he has ‘real essence’ in mind here, the attractive force
of matter is precisely his example, in the L2 metaphysics lectures, of a really essential
property of a thing not contained in the logical essence of its concept: “the first
ground of all predicates thus lies in a concept; but that is not yet a real essence.
E.g., that bodies attract belongs to the essence of things, although it does not lie in
the concept of body” (ML2, Ak. 28: 553).67 Let us just grant Kant that his constructive
procedure can prove (FC). What warrants him in concluding that:
(EC) Matter essentially possesses attractive and repulsive forces.

64
What I have in mind by ‘mere presence’ in space is the view that there is no constitutive connection
between matter’s location (the space it occupies) and a force to repel the motion of other matter into that
location; matter is just primitively located. Kant dismisses such a view because it would render it impossible
for us to experience the determinate locations of matter: if matter can be located in a space without exerting
force in that space then any body could be located anywhere in space, for all we know by experience.
65
Cf. Ch. 7.4 for a brief discussion of Kant’s view that there is no such object as the complete past before
a given time.
66
This is my (highly condensed) reconstruction of the proofs of Propositions 1, 2, and 5 in the
Dynamics chapter of MFNS (Ak. 4: 497–500, 508–9).
67
My interpretation differs sharply from that of Hanna (2006), 140–90. Hanna does not distinguish
Kant’s different notions of real modality. He assumes that if a scientific essentialist claim like ‘gold is the
element with atomic number 78’ is necessarily true, it is (in my terminology) formally necessary. However,
on Kant’s view, this claim is formally contingent but nomically necessary: if gold is in fact the element with
atomic number 78, then this is part of its essence, and, although we cannot experience gold that has a
different atomic number, this fact is grounded in the essence of gold, not in our forms of experience.
Hanna’s argument in the last section of that chapter—that there are experienceable worlds that involve
gold but have different laws, in which gold might have a different atomic number—is beside the point. That
shows the formal contingency of the fact that gold has atomic number 78; it does not address its modal
status with respect to nomic necessity. Given that, as I have argued, laws, for Kant, are grounded in real
essences of empirical natural kinds, in a world with different laws different empirical natural kinds are
instantiated.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

The relation between (FC) and (EC) is this: (FC) is a formally necessitated condi-
tional; (EC) is the claim that the consequent of that conditional is essential to matter.
That consequent—matter has attractive and repulsive forces—is the A-law in ques-
tion. So my question can be restated: how do we go from the formal necessitation of
the conditional principle to the claim that the A-law (the consequent of that
conditional principle) is essential to matter?

8.6. From Forms to Essences

At the end of the last section I raised a problem whose general form is this: suppose
we grant Kant that his constructive procedure can prove that:
(FC*) It is formally necessary that (if matter is an object of possible experience,
matter is c).
What warrants him in concluding that:
(EC*) It is essential to matter to be c.
I argued in section 3 that the real essence of matter is the complex of properties in virtue
of which, for anything that is an instance of matter, that thing is matter (see Fig. 8.3).
Let us assume, for the sake of argument, that being an object of a possible outer
experience (an outer phenomenon) is part of the real essence of matter (matter is
essentially the object of a possible experience), as shown in Figure 8.4.
In Chapter 7.3 I discussed the crucial claim of Kant’s transcendental idealism: “the
a priori conditions of a possible experience in general are at the same time conditions
of the possibility of the objects of experience” (A111). I argued there that this means,
where x is any possible object of experience, that conformity to the forms of
experience is part of what makes x a possible object of experience; in the terminology
of this chapter, conforming to the forms of experience is part of the real essence of
being an object of possible experience. This is expressed in Figure 8.5.

x is matter Fig. 8.3. Real essence of matter

grounds

x possesses [P1 + P2 + P3+ . . .]

(the real essence of matter)

x is matter Fig. 8.4. Real essence of matter

grounds

x possesses [possible outer object + . . . ]

OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

Fig. 8.5. Conformity to forms of x possesses [outer object]

experience
grounds

x possesses [conformity to the forms of

experience + . . .]

Fig. 8.6. Conformity to forms of x has property ψ

experience
x is matter
grounds
x possesses [outer object + P2 + . . .]
grounds

x possesses [conformity to the forms of

experience + . . .]

What does ‘conformity to the forms of experience’ mean? To return to a theme

from Chapter 7.3, the forms of experience ground a certain content in experience,
and this content grounds the properties of objects of possible experience; that x
conforms to the forms of experience means that x has certain formally necessary
properties, and it has them in virtue of the forms of experience. The possession by x
of those properties is grounded in the forms of experience.
If we put these pieces together, x is matter partly in virtue of being a possible object
of outer experience (Fig. 8.4), which it is partly in virtue of conforming to the forms
of experience (Fig. 8.5). This is shown in Figure 8.6.
Now the conclusion of Kant’s constructive argument is (FC*): in virtue of the
forms of experience, if matter is the object of a possible experience it has property c.
But I have argued that being an object of possible experience, and hence conforming
to the forms of experience, is part of the essence of matter, part of what makes x
matter. So the essential properties that ground x being matter also suffice to ground it
being c, in other words, being c is grounded in the essence of matter, or, more
simply, matter is essentially c.
To make this slightly more concrete, consider any possible object that is matter
(any possible body).68 It is matter partly in virtue of being a possible object of outer
experience, and it is a possible object of outer experience in virtue of conforming to
the forms of experience. So it is essentially a possible object of experience conforming
to the forms of experience (though that is not its whole essence). Now we know from

68
The relationship between the concept <body> and the concept <matter> in MFNS is not entirely
clear; Kant writes that “a body, in the physical sense, is a matter between determinate boundaries” (Ak. 4:
524). I take this to mean that body is a count noun, while matter is a mass term. All bodies instantiate the
concept <matter> because they are made of matter, but <body> adds more determinate content than is
contained in the concept <matter>, namely, that bodies have determinate boundaries.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

(FC*) that, in virtue of the forms of experience, any matter we can experience has
property c (e.g., has attractive and repulsive forces). So it has property c (e.g., has
attractive and repulsive forces) in virtue of properties that are essential to it. So it is
essential to matter to have property c.69
To return to and, finally, answer our initial question: how can we know a priori
that it is nomically necessary that matter has attractive and repulsive forces? Con-
formity to the forms of experience is part of the essence of matter (it is part of the
essence of any kind of outer object), and that matter has these forces is a formally
necessary consequence of our being able to experience it, and both of these claims are
a priori knowable: the first is a consequence of the a priori principle that all kinds of
outer objects are essentially kinds of objects of possible experience, and the second
follows from the a priori argument sketched above.70 The apparent tension between
Kant’s constructive procedure in MFNS and his essentialist conclusions is thereby
resolved.71

8.7. Nomic Possibility as Real Possibility

In section 4 I argued for a definition of nomic necessity:
(Nomic necessity) It is nomically necessary that p if and only if p is grounded in
the essences of actual empirical natural kinds.
Although I did not pause to unpack this notion there, it is important to note that the
nomic necessities are the truths grounded in the totality of facts about the essences of
empirical natural kinds. For instance, if it is not essential to kind ç to have property
c, but it is essential to kind ç* that it cause samples of kind ç to have property c, we
may not want to say that it is a law that ç has c (because this fact is not grounded in
the essence of ç itself) but we will want to say that it is nomically necessary, that it
‘holds with the necessity of law’ (given the empirical natural kinds there actually are).
We will also want to include among the nomic necessities, the logical consequences
of nomic necessities. For instance, if it is nomically necessary that p and nomically
necessary that q, it is, intuitively, nomically necessary that p&q even if we would not
want to count it as a law per se; this holds with the necessity of law, for if it were false,
at least one law would be violated. I propose, therefore, the following emendation to
our definition of nomic necessity from earlier:

69
Does this argument entail that it is formally necessary that matter is c? No, because it is not grounded
in the forms of experience alone that matter is an object of possible experience; that is partly grounded in
the essence of matter, which is not exhausted by ‘possible outer object’ (see Fig. 8.4).
70
Note that I am not claiming it is part of the essence of x itself to be matter; although I do think that
Kant accepts de re necessities like this, I am remaining neutral on that issue here.
71
Buchdahl (1992), 233, notes that MFNS takes its orientation from the construction of the concept of
matter rather than from claims about its real essence, but fails to note that Kant takes the former to justify
the latter.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

(Nomic necessity) It is nomically necessary that p if and only if p is a member of

the set S, where S contains all of the facts about the essences of the actual empirical
natural kinds there are, and S is closed under logical entailment (i.e. if p is in S, and
p logically entails q, then q is in S) and under grounding (i.e. if p is in S, and p
grounds q, then q is in S).
Given the interdefinability of necessity and possibility, this corresponds to the
following definition of nomic possibility:

(Nomic possibility*) It is nomically possible that p if and only if ¬p is not a

member of the set S, where S is as above.

This is not exactly right, though, for this definition will entail that, if there are no
empirical natural kinds, then everything is nomically possible. But that, intuitively, is
wrong; if there are no natural kinds then there is no such thing as nomic possibility.
So I will define nomic possibility instead as:

(Nomic possibility*) It is nomically possible that p if and only if ¬p is not a

member of the set S, where S is as above and S is non-empty.

For the most part, though, I will ignore these complications and regard nomic
possibilities as propositions whose negations are not grounded in the essences of
actual empirical natural kinds.
It should be clear that nomic possibility, so defined, satisfies the two conditions on
real possibility from Chapter 7.2:

(Real Possibility) Nomic possibility ◇Nop (and its associated kind of necessity
□Nop, where □Nop $ ¬◇No¬p) is a kind of real possibility (and □Nop is a kind of
real necessity) only if
(i) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇Nop (equivalently, it is
not a conceptual truth that □Nop ⊃ □Lp), and
(ii) Groundedness: if ◇Nop then the fact that ◇Nop has a real ground in some
actual object or principle.

For obvious reasons, it is not a conceptual truth that every nomic necessity is
logically necessary; as Kant points out, attractive force is not contained in the concept
<matter> but it is essential to matter. It is likewise clear that the grounding relation
that holds between the essences of empirical natural kinds and laws is not, in general,
a logical one. In section 6 I argued that the synthetic, logically contingent a priori
formal principles of experience, provide, in some cases, the relation between the real
essence of matter and its nomically necessary properties (e.g., attractive force); the
real essence of matter grounds the fact that matter has attractive force, but it does not
logically entail it. Nor is the relation a causal one. The essence of matter does not
cause these laws to obtain; the essence of matter grounds the fact that matter has
attractive and repulsive forces with certain properties, and the operations of these
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

forces cause motions. Kant needs this grounding relation to draw the distinction
between essential properties sensu stricto (properties that are parts of real essences)
and attributes (properties grounded in real essences).72 As I pointed out earlier, it is
not plausible that all laws state properties contained in an essence; it not plausible, for
instance, that every complex mathematical theorem of Newtonian dynamics is
contained in the essence of matter.
One constraint this grounding relation must meet is that the attributes be ‘just as
necessary’ as the essential properties in which they are grounded. Readers might well
wonder what it could mean for something to be ‘just as necessary as’, or more or less
necessary, than something else.73 Intuitively, the idea is this: we do not want it to be
the case that the attributes are grounded in the essence but might not have been. This
is not a problem if the grounding relation is logical grounding (logical relations hold
with logical necessity, and logical necessities are true under any counterfactual
supposition), but it would be if the grounding relation here were causal grounding
(the essence might not have its actual effect). Since the grounding relation involved,
I have argued, is non-logical non-causal grounding, we need an account of why this
problem does not arise. The natural answer is that the grounding relation is itself
essential to the natural kind in question: if it is essential to ç to stand in a grounding
relation to c then there is no worry that, in some counterfactual situation (some
other possible world), ç might lack c.
This would be the case if the constructive procedure described above were true of
all laws as such. Assume that all laws have the form ç is c, where ç is an empirical
natural kind and c is a formally necessary condition on experiencing objects of kind
ç. Since the forms of experience are essential to ç (ç is a kind of matter, and the forms
of experience are essential to matter), it follows, by the argument of the previous
paragraph, that the grounding relation between ç and c is essential to matter. What
is more, this would allow us to give a unified account of the grounding relation
involved in nomic necessity. In Chapter 7.3 I explored the relation between two basic
grounding relations in Kant’s theory of experience: the content-content grounding
(between the forms of experience and the content of experience) and content-
property grounding (between the content of experience and the properties of phe-
nomena). We might be able to analyze the grounding relation involved in nomic
necessity as the product of these two grounding relations. If, as we have assumed, the
forms of experience make it the case that if we experience ç we experience it as
having c, then there is a content-content grounding relation between these contents:
experiencing the content ç grounds experience of the content c. Likewise, there is a
content-property grounding relation: since the objects in question are phenomena,
our experience of them as c grounds their possessing that property (e.g., having their

72
See the lectures on metaphysics: MV (Ak. 28: 411), MvS (Ak. 28: 492–3), ML2 (Ak. 28: 552–3), MD
(Ak. 28: 629), and MM (Ak. 29: 820).
73
For a sophisticated account of ‘degrees’ of necessity, see Lange (2009).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

motions relative to a family of equivalent inertial frames). Combining these two

grounding relations, experiencing objects as ç grounds their possessing c; since çs
are phenomena, they are essentially objects of possible experience, so their possession
of c is grounded in the essence of ç. This would bring a satisfying degree of unity to
Kant’s theory of real grounds, but it would not constitute a reductive analysis of this
real grounding relation.
I have been investigating the possibility that the real grounding relation in the
definition of nomic necessity could, in general, be modeled on the real grounding
relations involved in formal necessity, effectively generalizing this constructive pro-
cedure to all laws as such. There is reason, though, to be skeptical that this is Kant’s
view. First of all, in MFNS Kant adopts a skeptical attitude to the possibility that
chemistry could ever be a rational science, which means: the possibility that the
constructive procedure could ever be applied to chemical kinds.74 He does not
express skepticism that there are chemical laws, only that they could be discovered
a priori through a constructive-mathematical operation like the one by which we
discover the a priori laws of motion. This suggests that Kant thinks that the laws of
nature may outstrip, in principle, all possible a priori constructions. At a more
general level, Kant is committed to the idea that Nature forms a system of higher-
and lower-level empirical (kind) concepts and laws75 and this commitment appears
independent of his commitment to the a priori constructive procedure for discover-
ing A- and B-laws in MFNS. For reasons of space, though, I cannot discuss the real
grounding relation in any more detail here.76

8.8. Back to the Beginning

I began this chapter by quoting certain texts in which Kant claims that it is formally
necessary that there are laws, but not formally necessary which laws there are. But we
are now in a position to understand some additional claims Kant makes in these
passages, which might otherwise have been obscure. In the very same passages in the
third Critique in which Kant claims that it is formally necessary that there are laws,
he typically inserts a note of caution. For instance:
[Laws] must be regarded as necessary on a principle of the unity of the manifold, even if that
principle is unknown to us. (CJ, Ak. 5: 180)
[ . . . ] given the constitution and the limits of our faculties of cognition we have no insight at all
into [gar nicht einsehen] this necessity [of laws]. (CJ, Ak. 5: 183)
[The understanding] does not and never can cognize [laws’] necessity. (CJ, Ak. 5: 184)

74
MFNS, Ak. 4: 471.
75
See the Appendix to the Transcendental Dialectic (A642/B670–A668/B696) and the two introduc-
tions to the CJ.
76
Readers interested in further discussion can see the article “Further issues in grounding and nomic
necessity” on my website (see the Notes on the Text).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOMIC NECESSITY

Now we can understand Kant’s claims in these passages: even if we cognize a law we
never have complete insight into why it obtains because we never know the complete
real essence in virtue of which it obtains. 77 As Kant puts it in the L2 lectures: “the real
essence of things is inscrutable to us, although we cognize many essential aspects”
(ML2, Ak. 28: 553).78 This will be especially true of C- and D-laws, which are known a
posteriori if known at all (assuming the constructive procedure does not generalize):
even if it is a law that p and we know that p, we will typically not know what it is about
the relevant natural kind in virtue of which p obtains. We will lack insight into the
necessity of this law.
I have focused mainly on A- and B-laws in this chapter. But in conclusion I want
to make a few remarks about C- and D-laws and how they fit into my interpretation.
C-laws are defined as laws that are not a priori consequences of A- or B-laws. If there
are such laws, they are not a priori because they are not grounded in the a priori
cognizable components of the essence of matter; if the a priori constructive proced-
ure generalizes to all laws (all laws can be proved from the constructive procedure
necessary to experience objects of the relevant kind), then there are no C-laws.79 The
clearest candidate for a C-law is the law that the attractive force of matter varies
inversely with the square of the distance; Kant’s attitude towards the inverse-square
law is somewhat complicated, but if he regards it as a posteriori then it is a C-law. If
the inverse-square law is a posteriori then it is a C-law grounded in the real essence of
matter that cannot be derived a priori from A- and B-laws.
Kant’s views about D-laws are somewhat more complicated, because they are tied
up with a complex set of doctrines about the systematicity of Nature, the regulative
use of reason, and reflecting judgment; a complete articulation of his views on these
topics would require extensive discussion of the Appendix to the Transcendental
Dialectic in the CPR and (at least) the two Introductions to the CJ. All I can do here is
indicate how my essentialist interpretation of nomic necessity fits into those issues.
D-laws, remember, are laws about empirical natural kinds more specific than matter.
I am unsure whether it is formally necessary that there be such laws; the formally
necessary principle that all alterations in substances be governed by universal laws
may be satisfied if all the laws applied to matter as such (A-, B-, and C-laws).
However, in scientific inquiry we are bound by the rational demand to find system-
atic unity in nature, guided by the idea of a system of nature in which the highest

77
Gibson (2011) claims we have no ‘insight’ into the natures of empirical objects (15). If ‘insight’
[Einsicht] is taken in its strict technical sense of a prioriG knowledge, this is correct, for insight into essences
would require knowing the whole essence and how it grounds essential properties. However, this is
compatible with our knowing many of the essential properties of generic objects (e.g., matter), a point
unappreciated by Gibson. I also think Gibson is wrong to claim that for Kant we can only know de dicto
necessities of the form necessarily all phenomena are F but no de re necessities, e.g., if x is a phenomenon,
then x is necessarily F; however, I do not have the space here to argue for that claim.
78
Cf. MM, Ak. 29: 821 (“we can cognize much that belongs [to the real essence], but not all”).
79
There might still be D-laws: laws grounded in the formally necessary conditions for experiencing
specific kinds of matter (e.g., chemical laws).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOMIC NECESSITY 

empirical concept, matter, is divided into species, which I take to be lower-level

natural kinds, which themselves have further species, etc. It follows from my essen-
tialist interpretation of nomic necessity that each of these species will have a specific
difference from its genus, something in its essence that distinguishes it from the
essence of the genus (e.g., metals will essentially be a specific kind of matter), and
these specific differences will ground lower-level laws that apply to those more
specific natural kinds. So the idea of a system of species (natural kinds) is equivalent
to the idea of a system of laws. However, our forms of experience do not determine
which such laws will obtain. In fact, our forms of experience leave it open whether the
system of laws is so vast and complicated that we could never comprehend even a
single one of them; it further leaves open whether the manifest similarities between
objects correspond to the underlying system of natural kinds in any way discoverable
by us. Consequently, empirical inquiry, whose goal is to infer the underlying system
of natural kinds and laws from their manifest properties, is only possible if we reflect
upon nature as the product of an intelligent cause that has ordered its kinds and laws
in such a fashion as to be comprehensible to us. This does not give us the right to
claim that there is such an intelligent cause, but rationality requires and enables us to
investigate Nature as though there is one.
Since one of the themes of this book is the continuity of Kant’s theory of modality
from Beweisgrund through the CPR, I want to mention briefly, before concluding this
chapter, that this essentialist view of laws, which I have argued Kant holds in the
Critical period, was also his view in Beweisgrund:
On this basis it can be seen that the laws of the motion of matter are absolutely necessary.
That is to say, if the possibility of matter is presupposed it would be self-contradictory to
suppose it operating in accordance with other laws. This is a logical necessity of the highest
kind. (OPG, Ak. 2: 100)80

We should not let his pre-Critical claim that the possibility (essence) of matter
logically entails its laws obscure the more important point. Kant in both the pre-
Critical writings and the Critical period takes the essence of matter to ground the
laws of its motions, although in the Critical period he takes this to be a real, rather
than a logical, grounding relation.

80
See also OPG, Ak. 2: 96, 99, and 131. See Insole (2011) for critical discussion.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

9
The Unity of Kant’s Modal
Metaphysics

9.1. Introduction
Up to this point I have explored Kant’s rejection of the logicist metaphysics of
Leibniz, Wolff, and Baumgarten, his own pre-Critical theory of the grounding of
real possibility in God, his Critical reasons for denying that any such transcendentally
realist metaphysical theory could constitute knowledge, as well as his own distinction
among various kinds of modality in the Critique of Pure Reason (CPR) and other
works written in the 1780s (most importantly, Metaphysical Foundations of Natural
Science). This may give the inaccurate impression that Kant’s Critical modal meta-
physics is a mere aggregate of different modal notions, and, furthermore, that it lacks
any substantive connection to his pre-Critical theory, other than his consistent
rejection of logicism. In this chapter I argue, on the contrary, that Kant’s modal
metaphysics is tightly unified, both historically and systematically. First, it is system-
atically unified because all of these different kinds of real possibility exhibit a
common scheme in virtue of which they are kinds of real possibility. Some of
this was sketched earlier, in Chapter 7; in this chapter I go into more detail about
Kant’s unified conception of real possibility in the Critical system. Second, it is
historically unified because the content of Kant’s pre-Critical theory of God as the
ground of real possibility is retained within the Critical philosophy, while its epi-
stemic status is changed; what was once the conclusion of a philosophical demon-
stration is now an article of rationally justified belief [Glaube] that falls short of
cognition [Erkenntnis] or knowledge [Wissen].
The jumping-off point for my argument in this chapter is this passage in the Pölitz
transcripts of Kant’s rational theology lectures from the mid-1780s:1
On this point rests the only possible ground of proof for my demonstration of God’s existence,
which was discussed in detail in a work I published some years ago. Here it was shown that of
all possible proofs, the one which affords us the most satisfaction is the argument that if we
remove an original being, we at the same time remove the substratum of the possibility of all

1
Kant lectured on rational theology during the winter semesters 1783–4 and 1785–6, and it is unclear
which series of lectures is the basis of the Pölitz transcripts. See Kant (1996b), 337–8 for more.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

things.—But even this proof is not apodictically certain; for it cannot establish the objective
necessity of an original being, but establishes only the subjective necessity of assuming [anneh-
men] such a being. But this proof can in no way be refuted, because it has its ground in the nature
of human reason. For my reason makes it absolutely necessary for me to assume a being which is
the ground of everything possible, because otherwise I would be unable to know what in general
the possibility of something consists in [worin etwas möglich sey]. (Pöl.RT, Ak. 28: 1034)2

The “work published some years ago” is, of course, Beweisgrund. In this passage,
Kant appears to be saying that the 1763 argument does not constitute a proof or even
a ground of proof 3 of the existence of a first ground of real possibility. I take this to
mean that by following the steps of the 1763 argument and concluding on that basis
that there is a first ground of real possibility we do not thereby come to know that
there is such a being. Nonetheless, it is rationally necessary for us to postulate the
existence of such a being that grounds all real possibility and to postulate it for the
very reasons given in Beweisgrund. Whereas in 1763 Kant argued that, since there are
real possibilities, there must exist a ground of all real possibility, he now thinks that
the fact that something is really possible rationally requires us to seek an explanation
of that fact, and the only rationally satisfying explanation is that there is a ground of
all real possibilities, so we are rationally necessitated to accept—I want to remain
neutral for the moment on exactly what ‘pro’ attitude we are supposed to adopt
towards this proposition so I am going to use the non-Kantian term ‘accept’4—that
there is such a being.
This passage raises a number of questions, the most important of which is: what is
the nature and source of the rational necessity that we postulate such an original
being? Before we go any further, though, I want to point out that this is not the only
necessity invoked in this passage. Kant very quickly adds another necessity, the
necessity of the being thereby posited:
For in addition to the logical concept of the necessity of a thing (where something is said to be
absolutely necessary if its non-being [Nichtseyn] would be a contradiction, and consequently
impossible), we have yet another rational concept of real necessity. This is where a thing is
eo ipso necessary if its non-being [Nichtseyn] would remove all possibility. Of course in the
logical sense possibility always precedes actuality [Wirklichkeit], and here I can think the
possibility of a thing without actuality [Wirklichkeit]. Yet we have no concept of real possibility
except through existence [Existenz], and in the case of every possibility which we think realiter
we always presuppose some existence [Daseyn]; if not the actuality of the thing itself, then at
least an actuality in general [Wirklichkeit überhaupt] which contains the data for everything
possible. Hence every possibility presupposes something actually given [wirklich gegebenes],
since if everything were merely possible, then the possible itself would have no ground; so this

2
Cf. Volk.RT (Ak. 28: 1176) and Danz.RT (Ak. 28: 1259). The translation of the Pölitz text is from Kant
(1996b), with minor modifications.
3
See Chapter 5.3 for the difference between Beweis and Beweisgrund.
4
In order to distinguish between ‘accepting’ (my neutral term) and Kant’s technical term Fürwahrhal-
ten, I translate the latter as ‘holding to be true.’ I discuss Fürwahrhalten and its different grades in }5.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

ground of possibility cannot itself be merely possibly, but must be actually given [wirklich
gegeben]. But it must be noted that only the subjective necessity of such a being is thereby
established, i.e. that our speculative reason sees itself necessitated to presuppose this being if it
wants to have insight at all into why something is possible [wenn sie überhaupt einsehen will,
warum etwas möglich sey], but the objective necessity of such a thing can by no means be
demonstrated in this matter. (Pöl.RT, Ak. 28: 1036)5

Kant attributes necessity not just to the postulation of an ‘original being’ but to the
content of that postulate: it is rationally necessary that we postulate an absolutely
necessary being. He says we have a logical concept of necessary existence—a being
exists necessarily if and only if the proposition that it does not exist entails a
contradiction—which, of course, is the conception of necessary existence that formed
the basis of the ontological argument in Leibniz, Wolff, and Baumgarten. Kant rejects
this conception of necessary existence because it requires the (according to him) false
assumption that existence is a determination or ‘real predicate.’6
I want to focus on what Kant calls the ‘other’ concept of absolutely necessary
existence: a being exists absolutely necessarily just in case that being exists and, were
it not to exist, nothing would be really possible. This is the very same definition of
absolutely necessary existence Kant gave in Beweisgrund and which I analyzed in
some detail in Chapter 5. He also introduces here a concept of ‘real’ possibility and a
principle governing real possibility—if it is really possible that p then the fact that it is
really possible that p is grounded in something that actually exists—that are familiar
from that pre-Critical work and our earlier discussion of it.
Together, these passages from the Pölitz theology lectures raise a number of
interesting questions:
(1) What does Kant mean when he says “in the case of every possibility which we
think realiter we always presuppose some existence”?
(2) What is the concept of necessary existence that Kant claims we have (in the
second passage)?
(3) Why is it rationally necessary that we postulate such a necessary being?
(4) What is the epistemic status of this postulate? How and why is this epistemic
status different than the epistemic status of the conclusion of the 1763
argument?
In sections 2–4 I answer the first and second question by showing the unity of
Kant’s Critical theory of modality and how this concept of necessity fits into it. In
section 4 I begin to answer the second two questions by examining Kant’s discussion
of the modes of ‘holding to be true’ [Fürwahrhalten] in the ‘Canon of pure reason’
section of the CPR. I argue for a category of ‘holding to be true’ that I call ‘necessary

5
This crucial text is also cited and discussed in Adams (1994), 182, n. 9; and Chignell (2007b), 349.
6
See Ch. 1.6.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

theoretical beliefs.’ A necessary theoretical belief is a belief in a postulate that is

necessary to satisfy theoretical reason’s need for explanatory unity, but which cannot
be cognized by us. The postulation of an absolutely necessary being that grounds real
possibility is just such a necessary theoretical belief. It constitutes the only rationally
satisfactory explanation of why anything at all is really possible, and thus is the only
way to answer theoretical reason’s demand for an explanation of this fact. I conclude
the chapter by returning to the Pölitz passage and explaining in detail what it means.

9.2. Categories and Real Definitions

The first thing to note about the Pölitz passage quoted above is that these modal
concepts—absolutely necessary existence and real possibility—are being applied
beyond the range of experience—we are rationally necessitated to postulate a super-
sensible object that is absolutely necessary.7 This involves using the modal category
<necessity> in thinking of a supersensible object, a noumenon. But what is the
content of a modal category when applied to noumena? This is an instance of a
larger question in Kant’s philosophy: what (if anything) is the content of a category
when it is applied to noumena?8 In this section I try to answer that question.
Although we can (and must) think of all objects under the categories,9 there is an
important difference in the epistemic status of these thoughts for objects we can
intuit (phenomena) and those we cannot intuit (noumena).10 We can know a priori
that it is really possible for phenomena to fall under the categories; this, I argued in
Chapter 6, is what it means that phenomena can be cognized a priori using the
categories, or, equivalently, that the categories have a priori objective reality with
respect to phenomena. Cognition [Erkenntnis] is distinct from, but necessary for,
knowledge [Wissen]: a priori knowledge about objects in some domain (e.g., nou-
mena) using a set of concepts (e.g., the categories) requires that we can come to
know a priori the real possibility that objects in that domain instantiate those
concepts (that those concepts are cognitions for us with respect to that domain).
Since we cannot think about any objects without using the categories,11 if the

7
Some readers might object that the necessary being hypothesized at Ak. 28: 1034–6 is not super-
sensible; this being is merely the ‘totality of experience’ that grounds all possibility according to Kant at
A581–2/B609–10. However, that claim occurs in the context of Kant’s discussion of the idea that God is
necessary as the ground of the complete determination of all objects. The Beweisgrund argument does not
concern complete determination, so Kant’s discussion in the Pölitz lectures, which refers back to that text,
is not about the idea of complete determination. See Stang (2012) for a discussion of the complete
determination principle in the CPR.
8
I am using ‘content’ here in a very minimal sense (roughly, the meaning or intension of the category),
not in Kant’s technical sense: its relation to a (knowably really possible) object. In Kant’s technical sense,
categories do not have content (Inhalt) for noumena. See Ch. 6 and Tolley (2011) for further discussion.
9
A253, CPrR (Ak. 5: 54, 103).
10
On the ‘one object’ interpretation, empirical and non-empirical standpoints on objects.
11
At B110 Kant distinguishes between the mathematical categories (of quantity and quality) and the
dynamical categories (of relation and modality), a distinction to which he returns at A160/B199. The
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

categories lack objective validity for some domain of objects, then we cannot cognize
those objects, which entails that we cannot know anything about them, aside from
logical truths and perhaps some negative facts about them (e.g., that they are not in
space and time).12 Categories as applied to phenomena have objective validity because
they can be schematized, i.e. provided with a priori rules for application to objects
given in experience. I will refer to them as ‘schematized categories.’ Categories, as
concepts of objects in general (phenomena and noumena), cannot be schematized;
I will sometimes refer to them as unschematized categories. The question from earlier
can be rephrased: what are we doing when we think of an object under the un-
schematized categories <necessary> or <possible>? What is the content of such
thoughts?
Before going any further in trying to answer that question, I want to investigate
further what is lacking or defective about unschematized categories. In the ‘Phenom-
ena and noumena’ section, Kant makes the familiar point that unschematized
categories do not have objective validity for noumena because we cannot prove the
real possibility of noumena falling under them.13 However, he goes on in the next
paragraph to suggest that unschematized categories—the categories as applied to all
objects in general, both phenomena and noumena—are defective for cognitive use by
us for a slightly different reason:
That this is also the case with all categories [that we cannot use them to cognize anything about
objects in general] however, and the principles spun out from them, is also obvious from this:
that we cannot even give a real definition of a single one of them, i.e. make intelligible the
concept of their object, without immediately descending to conditions of sensibility, thus to the
form of the appearances, to which, as their sole objects, they must consequently be limited,
since, if one removes this condition, all significance, i.e. relation to the object, disappears, and
one cannot grasp through an example what sort of thing was really intended by concepts of
that sort. (A241/B300 – underlined material added in B edition)

distinction seems to be that in thinking about objects of mere intuition (mere appearances in the
terminology developed above, e.g., objects constructed in pure intuition) we must use the mathematical
categories but not the dynamical categories, but we must represent objects of full-blooded experience using
both sets of categories. So it is not the case that we must use all moments of the table of categories in
thinking about every object. However, Kant’s reasons for thinking that we must use dynamical categories in
thinking about certain objects is that those categories “concern the existence of the objects of a possible
empirical intuition” (A160/B199). I take this to mean that we must use dynamical categories in thinking of
these objects because they exist independently of any particular intuition of them (unlike objects con-
structed in pure intuition). However, in thinking about noumena we are also thinking about objects that
exist independently of our thinking about them, so we must also use dynamical categories. So the point in
the main text holds.
12
Kant also makes at least one positive claim about noumena: that they are the causal ground of the
matter of experience (see Ch. 7). However, I will not attempt in this study to reconcile that doctrine with
Kant’s doctrine of noumenal ignorance.
13
A230/B298. The full title of this section is “On the ground of the distinction of all objects in general
[überhaupt] into phenomena and noumena.”
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

Kant here identifies the lack of objective validity of the categories in general with
our inability to give real definitions of them for all objects in general (phenomena and
noumena). He identifies a real definition of a category as making ‘intelligible’
[verständlich] to ourselves how it is really possible that objects instantiate it; this is
most naturally read as meaning that a real definition of a concept reveals not merely
that it is really possible for objects to instantiate it, but reveals or makes compre-
hensible to us why this is really possible. In his own copy of the A edition, Kant wrote
in the margin of this passage: “we cannot explain their [the categories’] possibility.”
Giving a real definition of a concept is proving that it is really possible that an object
instantiates that concept by eliciting the grounds of that real possibility and thereby
explaining that real possibility. When we really define a concept we come to know
what it is in virtue of which objects of that concept are really possible.
In Chapter 8 I discussed the distinction between logical and real definition (and
the correlative distinction between logical and real essence) in the case of empirical
concepts, but this distinction also applies to a priori concepts. A nominal definition
of an a priori concept reveals what we ‘think’ in the concept, the marks contained in
it, or what Kant sometimes calls the ‘logical essence of the concept.’ A real definition,
by contrast, reveals the essential inner marks of the object that make it possible and
that explain its other properties. A real definition would reveal not only that the
object of the concept is possible but why. For instance, a real definition of a
mathematical concept displays the grounds of the real possibility of such an object,
and paradigmatically takes the form of a construction of the object in pure intuition.
As with empirical concepts, Kant is quite skeptical about our ability to give real
definitions of ‘given’ a priori concepts, concepts given by the nature of the mind itself,
like the categories, as opposed to made a priori concepts like mathematical concepts,
which we arbitrarily define. Kant allows that we may be able to analyze given a priori
concepts up to a point and determine some of the marks contained in them, but, in
general, we will not be able to analyze them fully and give them complete nominal
definitions.
By denying that we can give real definitions of the categories for all objects
überhaupt, Kant is denying that we can know the grounds of the real possibility of
objects in general falling under the categories. In the A edition, Kant expands upon
the lack of a real definition of the categories as concepts of objects in general:

Above, in the presentation of the table of the categories, we spared ourselves the definitions of
each of them, on the ground that our aim, which pertains solely to their synthetic use, does not
make that necessary, and one must not make oneself responsible for unnecessary undertakings
that one can spare oneself. [ . . . ] But now it turns out that the ground of this precaution lies
even deeper, namely, that we could not define them even if we wanted to [ . . . ]. (A241)

The reason Kant did not define the categories earlier in the CPR is not, as he earlier
claimed, that putative definitions “would distract us from the chief point of the
investigation by arousing doubts and objections that can well be referred to another
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

occasion” (A83/B109), but because real definitions of the categories in general are
impossible for us; we can only give real definitions of categories restricted to
phenomena (schemata).
Kant admits that we can give definitions of a sort for the categories as concepts of
objects überhaupt, both phenomena and noumena, and gives a series of examples of
such quasi-definitions:
If I leave out persistence (which is existence at all times), then nothing is left in my concept of
substance except the logical representation of the subject, which I try to realize by representing
to myself something that can occur solely as subject (without being a predicate of anything).
(A242/B300)
From the concept of a cause as a pure category (if I leave out the time in which something
follows something else in accordance with a rule), I will not find out anything more than that it
is something that allows an inference to the existence of something else. (A243/B301)
No one has ever been able to define possibility, existence, and necessity except through obvious
tautologies if he wanted to draw their definition solely from the pure understanding. For the
deception of substituting the logical possibility of a concept (since it does not contradict itself)
for the transcendental possibility of things (where an object corresponds to the concept) can
deceive and satisfy only the inexperienced. (A244/B302)14

The ‘obvious tautologies’ to which Kant alludes in the third passage are principles
like: the necessary is that whose non-being is not possible.15 The third passage also
tells us what is incomplete in these putative definitions—it is not that they are false,
but they do not show that it is really possible for an object to fall under the concept,
much less why. At most they show that the concept is logically consistent. Kant does
not deny, for instance, that a cause in general is something from which the existence
of something else follows, but this principle does not show us what makes it really
possible for there to be a cause. Similarly, it is surely the case that a necessary being is
one whose non-existence is not possible. But this does not tell us what would make
such a being necessary. Given that Kant has just invoked the notion of a real
definition—and thus implicitly contrasted the real definition of the categories with
mere nominal definitions—it is natural to think of these as partial nominal defin-
itions. They unpack conceptual connections between categories (e.g., the inter-
definability of the necessary, the possible, and the contingent) but they make no
progress in making intelligible to us the real possibility that objects fall under these
concepts.

14
Cf. A459/B487 and esp. A593/B621, where Kant explicitly describes as a nominal definition the
principle that a necessary being is one whose non-existence is not possible but criticizes this for not making
“verständlich” to me “whether through the concept of an unconditionally necessary being I am still
thinking something or perhaps nothing at all.”
15
Cf. MV (Ak. 28: 418), MvS (Ak. 28: 498), ML2 (Ak. 28: 556–7), and MD (Ak. 28: 633).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

9.3. Unschematized Modal Categories

I have argued that by not being able to give real definitions of the categories as
concepts of objects in general we are not able to identify what, in general, grounds the
real possibility that objects fall under these categories, e.g., we are not able to answer
perfectly generally what makes it really possible that something is a cause, or a
substance. But what does ‘real possibility’ mean here? In Chapters 7 and 8
I distinguished various different kinds of real possibility in Kant’s Critical system
(formal possibility, empirical-causal possibility, noumenal-causal possibility, nomic
possibility), but these are defined only for propositions about phenomena. When
Kant claims that we cannot give a real definition of a single one of the categories as
concepts of objects in general, he presumably does not have these notions of real
possibility in mind because they do not apply to all objects überhaupt.
Before I address that question directly, I want to point out that the modal
categories should be treated slightly differently than the other categories. The prob-
lem of the real definition of <cause–effect>, for instance, is the problem of specifying
the ground of the real possibility of objects in general standing in cause–effect
relations, i.e. one object being such as to posit the existence of an object distinct
from itself. But applying the same scheme to the problem of the real definition of
<possibility> we get: the problem is to determine the ground of the real possibility of
objects in general being possible. In other words, we get an iterated modality: the
ground of the real possibility of possibility. This is very indirect and artificial. It is
much more natural to think that what we want from a real definition of the category
<possibility> is to determine the grounds of the real possibility of objects in general;
no modal iteration is necessary. After all, possibility is already included in the idea of
a real definition of a concept, so it makes sense that when we come to the question of
the real definition of <possibility> itself we do not ask about the grounds of the real
possibility of possibility, we ask about the grounds of the real possibility period. To
take a related example, Kant’s complaint about what he calls the ‘nominal definition’
of necessary existence (A593/B621)16—a being exists necessarily just in case its non-
existence is impossible—is not that this fails to tell us what makes a necessary being
possible (iterated modality) but that it fails to tell us, if there is a necessary being,
what grounds the necessary existence of that being, i.e. why it is not possible for that
being not to exist. Likewise, what is missing in our grasp of the category <possible> is
an account of what grounds the possibility of all objects in general—not an account
of what grounds the real possibility of the possibility of all objects in general.
But this means that the question being raised about the non-modal categories—
e.g., what is the ground of the real possibility of objects in general being causes?—is
the same kind of question being raised about the modal category—what is the ground
of the possibility of objects in general?—only if the notion of ‘real possibility’ involved

16
Kant makes the same point in Beweisgrund at Ak. 2: 81; I discuss these passages in Ch. 1.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

in the first question is the same as the notion of possibility involved in the second
question, which is itself the modal category <possibility>. This means that the sense
of ‘real possibility’ involved in a demand for a real definition of the non-modal
categories is the same notion of possibility involved in a demand for a real definition
of the modal category <possibility>. In other words, the unschematized category
<possibility>, applied to all objects in general, is the sense of real possibility at issue
when Kant denies we have insight into the grounds of real possibility of objects in
general falling under the categories.
The categories, including the modal categories, are concepts we apply when we
think about all objects in general (überhaupt). So if the unschematized category
<possibility> (henceforth, <possibility>UC) is a concept of real possibility, then it must
be the highest or most general concept of real possibility. If there were a more general
concept of real possibility than <possibility>UC then in thinking about that kind of
real possibility we would not be using the unschematized category, and this is
impossible; any thought about objects requires thinking about them under the
categories.17
Conversely, if the most general concept of real possibility were less general than
<possibility>UC—if logical possibility were a species of the unschematized category—
then Kant’s claim that we cannot give a real definition of a single one of the categories
as concepts of objects in general would be misplaced with respect to <possibility>UC.
For in this case, we could give a real definition of a species of <possibility>UC for all
objects in general, logical possibility, for we know quite generally what logical
possibility is (not containing mutually contradictory marks).18 The categories are
concepts at exactly the level where Kant thinks we cannot give real definitions for all
objects in general. If there is a higher genus of the concept and we cannot give a real
definition of that genus for all objects in general, then we have too narrowly specified
the category. And if there is a species of the concept for which we can give a real
definition for all objects in general, then we have identified the category too
generically.
What does it mean that the unschematized category is the most general concept of
real possibility? First of all, it means that the category of possibility is a concept of real
possibility, rather than of logical possibility. What does this mean? In Chapter 6
I claimed this principle captures part of what it is to be a kind of real possibility in
Kant’s Critical system:
(Real Possibility) For any kind of possibility ◇xp (and its associated kind of
necessity □xp, where □xp $ ¬◇x¬p), ◇x is a kind of real possibility (and □xp is a
kind of real necessity) only if

17
Or, at least, modal thought requires modal categories, just as thought about the quantities of objects
requires categories of quantity, and, as I argued above, thought about the existence* of objects requires the
relational categories.
18
Disc., Ak. 8: 195.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

(i) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇xp (equivalently, it is

not a conceptual truth that □xp ⊃ □Lp), and
(ii) Groundedness: if ◇xp then the fact that ◇xp has a real ground in some actual
object or principle.
When I introduced this principle I described it as an interpretive hypothesis that
will be confirmed in the course of my interpretation. We can now see that the Pölitz
passage is an important source of evidence for clause (ii), for it is precisely what Kant
means when he says: “yet we have no concept of real possibility except through
existence [Dasein], and in the case of every possibility which we think realiter we
always presuppose some existence” (Ak. 28:1036). This is a claim about the unsche-
matized category of (real) possibility in general. This is why our concept of real
possibility carries with it the rational necessity to inquire into the ground(s) of real
possibility: in thinking about noumena we have to think about them using the
unschematized category of (real) possibility and this carries with it the thought
that (real) possibilities for noumena—as for all objects in general—have grounds in
actuality.
Some readers might note that in the passage quoted from the Pölitz lectures Kant
says that a real possibility must be grounded in existence, while I claimed it is
grounded in actuality. As we have seen, Kant sometimes uses these terms inter-
changeably, and sometimes he distinguishes them. In previous chapters
I distinguished a broad notion of existence (on which it is identical to what is
‘absolutely posited’) and a narrow notion, which I designated existence*, which
applies only to the subset of (absolutely posited) objects that are causally efficacious.
Formulated in terms of this distinction, I think Kant’s view in the Pölitz lectures
cannot be “in the case of every possibility which we think realiter we always
presuppose something that exists*,” for clearly there are kinds of real possibility
that are not grounded in existing* objects, for instance, formal possibility; nor can it
be “we presuppose something there is [absolute positing],” for formal possibility is
grounded in the unity of apperception, and the unity of apperception is not the object
of an absolute positing for it is not an object of intuition. 19 So the notion of actuality
involved in clause (ii) above must be more general than absolute positing or exist-
ence*; I explore it in Chapter 10.7.20
If we cannot give a real definition of real possibility for all objects in general, and
cannot know a metaphysical theory about all objects in general, then one might
wonder about the status of Real Possibility, as formulated above. It seems to be a
global claim about the nature of real possibility for all objects in general. Recall,

19
See B Deduction, }25.
20
I also think there is a distinction between existence in the broadest sense (what there is, the objects of
absolute positing) and actuality, but we will not be in a position to understand this until we examine, in
Chapter 10, Kant’s doctrine that the intuitive intellect represents no difference between actuality and
possibility.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

though, that although he denies that we can give real definitions of categories for all
objects in general, Kant does allow that we can give partial (partial nominal defin-
itions) analyses of them. In section 2 I argued that these partial analyses of unsche-
matized categories include such principles as a substance has predicates but is not
predicated of anything else and a cause is something from which the existence of
something else follows according to a rule.21 But these analytic truths also include
modal principles:
No one has ever been able to define possibility, existence, and necessity except through obvious
tautologies if he wanted to draw their definition solely from the pure understanding. For the
deception of substituting the logical possibility of a concept (since it does not contradict itself)
for the transcendental possibility of things (where an object corresponds to the concept) can
deceive and satisfy only the inexperienced. (A242–4/B300–2)

These obvious tautologies, I take it, are partial nominal definitions of the relevant
unschematized categories (e.g., a cause is something from which the existence of
something else follows according to a rule). I have included the interdefinability of
possibility and necessity in Real Possibility, which I take to be a partial nominal
definition of our most general concept of real possibility, <possibility>UC. The first
clause of Real Possibility just unpacks what we mean by talking about ‘real’ possibility
in the first place. By itself, it does not entail even that real possibility and logical
possibility are really distinct (i.e. differ in extension). Finally, when Kant says in the
Pölitz lectures that, “yet we have no concept of real possibility except through
existence, and in the case of every possibility which we think realiter we always
presuppose some existence,” I take him to be asserting the second clause of Real
Possibility as an analytic mark of our most general concept of real possibility.
We can now return to Kant’s claim that we can only give real definitions of the
categories for phenomena. The problem of the real definition of the modal categories
is the problem of knowing what in general the real grounds of possibility are. So we
can now see that Kant’s claim that we can give real definition of the categories only
for phenomena means, in the case of the modal categories, that we only know the
grounds of real possibility for phenomena. If p is some proposition, and p is really
possible, then we know the grounds of the real possibility that p only if p is a
proposition concerning phenomena only. But phenomena have two kinds of
grounds: immanent grounds and transcendent grounds.22 We only know the ‘imma-
nent’ grounds of real possibility of phenomena, so the only kinds of real possibility
for which we can give real definitions are kinds of real possibility with immanent
grounds. Three of the kinds of real possibility distinguished in Chapters 7 and 8 have
immanent grounds: formal possibility (forms of experience), empirical-causal possi-
bility (the past and laws), and nomic possibility (real essences of natural kinds).
Although we cannot cognize the noumenal grounds of phenomena, we can think

21 22
A240/B300. See Ch. 7 for this distinction.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

about their noumenal grounds, and whether they ground certain real possibilities for
phenomena. For instance, do the causal powers of noumena ground the real possi-
bility that I could be so affected I would have an experience with a completely
different matter than the matter of my actual experience? Likewise, we know that,
since I actually made a lying promise at time t, that it was not empirically-causally
possible for me to omit that promise at t; but could my will, considered as a
noumenon, have determined itself to act according to a different maxim, one that
would not have produced this action? These are questions about what, in Chapter 7,
I called ‘noumenal-causal possibility’:
(Noumenal-causal possibility) It is noumenally-causally possible that p if and only
if (i) it is formally contingent that p, and (ii) there is some noumenon with the
causal power to make it the case that p, where p is a synthetic proposition about
phenomena.
Noumenal-causal possibility is one species of what I have called ‘noumenal1’ real
possibility in Figure 9.1, real possibility for phenomena that is grounded in noumena.
However, since we cannot cognize noumena, we cannot give a real definition of this
kind of real possibility.

Concepts of possibility

Logical possibility Real Possibility = unschematized category of possibility for all

objects in general (section 3)
Non-logicality
Groundedness
Absolute necessity (section 4)

Real Possibility for Phenomena = Real Possibility for Noumena=

category of possibility applied to phenomena
unschematized category of possibility
(hence schematizable):
applied to noumena
(noumenal2 possibility)

Where the actual ground of real possibility is Where the actual ground of real
‘immanent’ : possibility is ‘transcendent’
Formal possibility: grounded in actual (noumenal1 possibility)
forms of experience. Noumenal-causal possibility:
Empirical possibility: grounded in actual grounded in causal powers of
laws and the actual past. noumena
Nomic possibility: grounded in real Noumenal possibility
essences of actual empirical natural kinds. in general: real
possibility grounded
in noumena

Fig. 9.1. Kinds of real possibility in Kant’s Critical system

OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

Finally, for propositions or states of affairs involving noumena themselves we

might ask whether they are really possible, using the unschematized category of
(real) possibility.23 In doing so, we would be talking about a kind of real possibility
whose grounds must be transcendent, that is, noumenal, because noumena ground
phenomena, and not vice versa (so phenomena cannot be among the grounds of the
real possibility of noumena or any state of affairs involving noumena). We cannot in
principle know what the grounds of this kind of real possibility would be. This is the
notion of real possibility that I have called ‘noumenal2’ in Figure 9.1. But notice that
‘noumenal1’ and ‘noumenal2’ are really species of a single conception of real possi-
bility, which I have dubbed simply ‘noumenal possibility in general’: the concept of a
real possibility (whether for phenomena or for noumena themselves) that is
grounded in noumena. We cannot give a real definition of this kind of possibility,
for we do not know the noumenal grounds of real possibility. Figure 9.1 summarizes
the distinctions between, and relations among, these different notions of possibility.
To return to the original passage, we can now answer one of the questions from the
Introduction: what is the notion of ‘real possibility’ for which we are rationally
necessitated to postulate an absolutely necessary ground? We can get a grip on that
question by noting that Kant there regards this postulate as directed at the same
object whose existence he took himself to have demonstrated in Beweisgrund: God,
the absolutely necessary ground of real possibility. We need to ask ourselves, there-
fore, which of these various kinds of real possibility corresponds to the pre-Critical
conception of real possibility? I think the answer is clear: noumenal real possibility.
From the perspective of Kant’s Critical theory, the God of Beweisgrund was a
noumenon, a supersensible being, and he grounded real possibility for all beings,
including other noumena. Consequently, the absolutely necessary being we are
rationally necessitated to postulate is an absolutely noumenally necessary being
that grounds all noumenal real possibility.

9.4. Absolutely Necessary Existence

Kant’s primary concern in the Pölitz passage was the notion of ‘absolutely necessary
existence’ and the claim that we must postulate an absolutely necessarily existing
ground of real possibility, or, as we can now identify it, noumenal real possibility. In
sections 5 and 6 I will explore why these considerations make it ‘rationally necessary’
to postulate such a being, but in this section I will focus on the content of this
postulate.

23
Kant seems to be applying modal concepts directly to noumena—rather than merely considering
noumena as putative grounds of the possibility of empirical objects—at Refl. 5177 (Ak. 18: 109), 5184
(Ak. 18: 111), and 5723 (Ak. 18: 335).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

Kant’s notion of absolutely necessary existence in Pölitz is the same as his

conception of absolutely necessary existence in Beweisgrund. Consider these two
passages, the first of which is from Pölitz, the second of which is from Beweisgrund:
In addition to the logical concept of the necessity of a thing (where something is said to be
absolutely necessary if its non-existence [Nichtseyn] would be a contradiction, and conse-
quently impossible), we have yet another rational concept of real necessity. This is where a
thing is eo ipso necessary if its non-existence would cancel all possibility. (Pöl.RT, Ak. 28: 1036)
Something may be absolutely necessary either when the formal element of all that can be thought
is cancelled by means of its opposite [the ‘logical concept of the necessity of a thing’—NS], that
is to say, when it is self-contradictory; or, alternatively, when its non-existence [Nichtsein]
eliminates the material element and all the data of all that can be thought. (OPG, Ak. 2: 82)

The “data of all that can be thought,” in the context of Beweisgrund, is a reference
to really possible predicates. In Chapter 5 I argued that absolute necessity should be
interpreted as follows:
(1) It is absolutely really necessary that p if and only if ¬p ) (nothing is really
possible).
Kant invokes the same notion of absolute necessity in the Pölitz lecture. Principle
(1) defines absolute necessity in terms of a counterpossible conditional: if, per
impossibile, God were not to exist, nothing would be really possible. As I argued in
Chapter 5, this ‘canceling’ relation is a consequence of the grounding relation between
God and real possibilities: because God is the ground of all real possibility, if, per
impossibile, he were not to exist, none of those real possibilities would be possible.
Kant is claiming in the Pölitz lectures that we are rationally required to postulate
the existence of an absolutely necessary ground of noumenal real possibility (God),
where noumenal real possibility is one of several specific kinds of real possibility. If
my interpretation is correct up to this point, then we should expect that, for each
specific kind of real possibility, there will be (i) a corresponding notion of absolute
necessity, and (ii) an absolutely necessary entity or principle that is the first ground of
real possibility of that specific kind. After claiming at Ak. 28: 1034 that we are
rationally necessitated to assume the kind of “absolute necessary being” whose
existence Kant (he now thinks) mistakenly took himself to have proved in Beweis-
grund, Kant claims at Ak. 28: 1036 that “we have no concept of real possibility except
through existence [Existenz], and in the case of every possibility which we think
realiter we always presuppose some existence [Daseyn]; if not the actuality of the
thing itself, then at least an actuality in general [Wirklichkeit überhaupt] which
contains the data for everything possible.” Earlier I interpreted those as claims
about real possibility as such; they are conceptual truths about our most general
modal concept, the category of (real) possibility. If we are rationally required to form
the idea of an absolutely necessary ground of real noumenal possibility, then it would
stand to reason that we are rationally required to form the idea of an absolutely
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

necessary ground for each different lower-level kind of real possibility. Thus, for each
lower-level kind of possibility, we should expect to find some absolutely necessary
principle p such that:
(1*) It is absolutely necessaryx (absolute-□x) that p if and only if ¬p ) (nothing
is really possiblex).24
The subscript x refers to the different kinds of possibility, from above. Some care is
required in deploying the formula, though, because the catch-all phrase ‘nothing
would be really possible’ covers over differences in what the various kinds of real
possibility apply to (experiences, alterations, etc.). In the case of the ‘immanent’
real modalities—modalities with immanent grounds—we can give more specific
definitions:
(1F) It is absolutely formally necessary (absolute-□F) that p if and only if ¬p )
(there is no formally possible experience).
(1E) It is absolutely empirically-causally necessaryx (absolute-□E) that p if and
only if ¬p ) (there are no empirically-causally possible alterations).
(1N) It is absolutely nomically necessary that p (absolute-□No) if and only if
¬p ) (there are no nomically possible truths).
Precisely because we know the grounds of these kinds of real possibility, we should
expect that we can know what is absolutely necessary in the case of each of them.
This expectation is most clearly satisfied in the case of formal possibility. In several
texts Kant identifies space as the absolutely necessary ground of formal possibility for
outer objects and explicitly compares space in this respect to God as the absolutely
necessary ground of noumenal possibility:
All reality must be completely given, and so some actuality is prior to possibility, just as space is
not merely something possible, but the ground of all possible figures. (Refl. 4119, Ak. 17: 424)
From this it follows only that the ens realissimum must be given prior to real concepts of
possibility, just as space cannot be thought antecedently as possible, but as given. But not as an
actual [wirklich] object in itself, rather as a merely sensible form, in which alone all objects can
be intuited. (Refl. 6290, Ak. 18: 558)25

24
‘Absolute-□x’ indicates that this is a definition of the absolute form of the relevant kind of necessity;
the argument given above shows that this is not simply necessity of the relevant kind (□x).
25
Kant makes this connection between God and space even more explicit in the Transcendental Ideal
section of the CPR, where he writes: “all manifoldness of things is only so many different ways of limiting
the concept of the highest reality, which is their common substratum, just as all figures are possible only as
different ways of limiting infinite space” (A578/B606). In this passage, Kant compares the role that the ens
realissimum, God, plays in grounding possibility to the role that space plays in grounding the possibility of
figures in space. However, the connection he makes here rests on the idea that God contains all realities,
which reflects neither his pre-Critical understanding of how God is the ground of possibility nor his
considered view in the Critical period. See A579/B607, where Kant makes clear that some realities are
grounded in God without being limitations of infinite realities instantiated by him.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

These texts suggest that something about space is absolutely formally necessary, in
the sense specified by (1*) above (according to my regimentation, there is some
proposition p about space that is absolutely formally necessary). It cannot be that
space exists* because (as the second text reminds us) space is not causally efficacious
(it is not wirklich). It is a pillar of Kant’s Critical theory that without being given in
space and time, no outer or inner object is (formally) possible; if objects were not
given to us in space and time, there would be no experience.26 The absolute formal
necessity of space and time consists, then, not in the fact that space exists* absolutely
formally necessarily, but in the fact that it is absolutely formally necessary that outer
objects are given to us in space and inner objects are given to us in time. Per
impossibile, if objects were not given to us in space and inner objects were not
given to us in time, there would be no formally possible experience of outer or
inner objects, respectively. Space is the absolutely necessary ground of the formal
possibility of outer objects;27 time is the absolutely necessary ground of the formal
possibility of inner objects.
It is also a pillar of Kant’s Critical theory of experience that experience has two
kinds of grounds: sensible grounds (the givenness of objects in intuition) and
intellectual grounds (thinking about objects). The unity of apperception, the source
of the intellectual conditions of experience, is just as absolutely formally necessary as
space and time, as Kant makes clear in this passage from the B Deduction:
The supreme principle of the possibility of all intuition in relation to sensibility was, according
to the Transcendental Aesthetic, that all the manifold of sensibility stand under the formal
conditions of space and time. The supreme principle of all intuition in relation to the
understanding is that all the manifold of intuition stand under conditions of the original
synthetic unity of apperception. (B136)28

Kant’s description of space and time as “the supreme principle of possibility [der
oberste Grundsatz der Möglichkeit]” of objects of intuition is a reference, I take it, to
the analogy between the role that space and time play in grounding the formal
possibility of objects of intuition and the role that God plays in grounding (nou-
menal) real possibility. Kant here extends that analogy to the “original synthetic” unity
of apperception: the unity of apperception is the ground of the real formal possibility of
intellectual representation of sensibly given objects, that is, of objects of experience
(phenomena) in the full sense. Formal possibility has three grounds, each of which is
absolutely formally necessary: space and time (sensible grounds), and the unity of

26
A23–25/B38–39 and A30–32/B46–47 are especially illuminating when read in light of the idea of
absolute formal necessity.
27
One would expect Kant to say similar things about time. However, from my overview of the relevant
texts, it seems he more frequently draws the comparison between space and God as first grounds of (formal
and noumenal, respectively) possibility.
28
Cf. A107: “the numerical unity of this apperception grounds all concepts a priori just as the
manifoldness of space and time grounds the intuitions of sensibility.”
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

apperception (the intellectual ground). If any of these were ‘canceled’ there would be no
formally possible experience of objects.29
Applying the definition of absolute necessity to empirical-causal possibility we get:
(1E) It is absolutely empirically-causally necessary (absolute-□E) that p =def ¬p
) (no alterations occur that are compatible with the actual empirical laws and the
actual past).
Since the actual alterations are all and only the empirically-causally possible
alterations, we can replace the final consequent of (1E) with: none of the actual
alterations occur. Consider the following conditionals:
(i) (t)(An arbitrary non-zero interval of the past before t is different than it
actually is ) no actual alterations occur at t).
(ii) (t)(An arbitrary non-zero interval of the future after time t is different than it
actually is ) no actual alterations occur at t).30
(iii) (t)(The laws are different than the actual laws ) no actual alterations
occur at t).
Conditional (i) is true only if the determinism that obtains at the empirical level
excludes ‘convergent’ empirical series: empirical series, governed by the actual laws,
that are distinct from the actual series before a time but then ‘converge’ to the actual
series after that time. If the series before t were different from the actual series (for a
non-zero interval), the laws would remain the same;31 so whether any actual alter-
ations would occur after t depends upon whether different possible world-series with
deterministic laws can converge. However, Kant does not seem to think that the
determinism of natural laws excludes convergence; to put it roughly, he holds a ‘same
cause, same effect’ principle, but not a ‘same effect, same cause’ principle;32 causation
introduces temporal asymmetry, according to Kant, so the determinism that

29
For each kind of possibility distinguished above, we can distinguish a rigid definition of that kind of
possibility and a non-rigid definition. On the non-rigid definition, we eliminate ‘actual’ from the definition.
This will produce an important difference when we embed these definitions of possibility within modal
contexts, as we are doing with the definitions of absolute necessity above. In the main text I will focus on
the rigid definitions, and thus on rigid absolute necessity. For the non-rigid notion, and its difference from
the rigid notion, see the supplementary article “Rigid and non-rigid absolute necessity” on my website (see
Notes on the Text).
30
In Chapter 7.3 I pointed out that, for a given time t, “the past before t” and “the future after t” do not
refer to anything, on Kant’s view, because the complete totality of the past or future is not a possible object
of experience. The same fix I proposed there also applies here: let “the past before t” refer to any non-zero
interval before t, and let “the future after t” refer to any non-zero interval after t.
31
I am assuming that laws are more ‘counterfactually robust’ than alterations; in the contemporary
idiom of worlds and closeness, for any world in which the past is different and the laws are different, there
is a world closer to the actual world where the past is different and the laws are the same. See Lange (2009)
for a different way of formulating this idea.
32
See his discussion, in the metaphysics lectures, of the principle that a cause posits a determinate
(specific) effect, while an effect entails that there is some cause or other, but does not posit a determinate
(specific) cause: MV (Ak. 28: 401–2, 408), MvS (Ak. 28: 487), and MM (Ak. 29: 808).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

causation brings into the empirical series goes one way: the past determines a unique
future, not the other way around.
Ironically, while the past is the ground of the empirical-causal possibility of
alterations in the future, there are reasons to think that the future is absolutely
empirically-causally necessary. Assume that events in the future (past time t) are
different than they actually are during some non-zero interval. Holding fixed the
natural laws, and given determinism, it follows that the empirical series before t is
different than the actual series at every time. Given Kant’s determinism, if the past
were the same as the actual past for any non-zero interval, the future would be the
same as the actual future, contra hypothesis. However, if the past is different than the
actual past for every non-zero interval before t, it follows that if alteration E is actually
empirically possible at time t (i.e. alteration E actually occurs), it does not occur.33
So if we change the future after time t we cancel the empirical possibility of any
actually empirically possible alterations at time t. There are good reasons, then, to
think that (ii) is true and that the future is absolutely empirically-causally necessary.
Determining whether the actual laws are absolutely empirically-causally necessary
(i.e. whether (iii)) is true, is somewhat harder. The laws (I argued in Chapter 8) are
truths grounded in the essences of empirical natural kinds; therefore, for the laws to
be non-actual is for there to be non-actual empirical natural kinds (because nothing
could have a different essence than it actually has, and the relevant grounding
relations are essential to the natural kinds as well). There are two ways for there to
be non-actual empirical natural kinds: (a) there is a non-actual kind of matter, or (b)
there is a non-actual kind that is not a kind of matter. Option (b) divides into two
sub-options: (b1) there is matter in addition to the non-actual kind, or (b2) there is
no matter. If (b2) obtains, then none of the alterations that occur are actual alter-
ations because actual alterations are individuated by the substances of which they are
alterations, and if there is no matter, then no actual substances exist (because they are
all material substances).34 So we are left with (a) and (b1). But intuitively, there could
be a slight difference in laws from the actual world (either (a) because there is a non-
actual kind of matter or (b) because there is a non-material kind) while some, if not
most, actual alterations remain, if the difference were small enough (e.g., if the non-
actual kind were very extremely rare). It is hard to evaluate the counterfactual

33
See the discussion of continuity at A207-8/B253-4. There is a complication here. The argument in the
main text shows that the ‘complete state’ of the world for any non-zero interval of time before t cannot be
identical to the ‘complete state’ of the actual world during that interval (for, by determinism, this would
entail that the series after t is identical to the actual series). But it may be that some actual alteration occurs,
even though it is part of a different ‘complete world-state’ than it is in the actual world. This problem can be
partly alleviated by individuating alterations very finely, to include the ‘complete world-state’ at the time of
alteration. Another complication is that there is no such thing as ‘the complete world-state’ at a given time,
just as there is no ‘complete world-history’ before a given time, because this is not a possible object of
experience. Resolving these problems would require going into more detail on Kant’s views on the infinite,
ideas of reason, and the metaphysics of the natural world than I have space to do here.
34
I am assuming that any material substance is de re necessarily a material substance.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

conditional, if there were a non-actual kind of matter no actual alterations would

occur, without knowing much more about Kant’s views on causal determinism. It is
clear, though, that if there were no matter (e.g., if the laws of matter did not obtain)
then no actual alteration would occur; the existence of matter, the highest empirical
natural kind, is absolutely empirically-causally necessary. Since the laws of matter are
grounded in the real essence of matter, these are absolutely empirically-causally
necessary as well.
When we apply this notion to nomic necessity we obtain:
(1N) It is absolutely nomically necessary that p (absolute-□No) =def ¬p ) (nothing
is compatible with the essences of the actual empirical natural kinds).35
The actual natural kinds form a system of species and genera: the highest natural
kind is matter, and all of the other natural kinds are, ultimately, species of matter.36 If
there were no matter then there would be no actual natural kinds, and nothing would
be compatible with essences of actual empirical natural kinds (because these would
also be canceled). It is absolutely nomically necessary that there is matter. But no
other actual empirical natural kind is absolutely nomically necessary; for any other
such kind, if that kind did not exist, there would still be matter (matter is the highest
kind), so some things would be compatible with actual empirical natural kinds (in
particular, with matter). This is summarized in Table 9.1.
Now we are in a position to see that Kant’s Critical modal metaphysics, far from
being a mere list of different conceptions of modality, is highly systematically unified,
and moreover, this unity is provided by the elements that are retained from his pre-
Critical modal theory. Kant has a very general concept of real possibility, the
unschematized category <possibility>UC. Certain conceptual truths hold of this very

Table 9.1. Absolute necessities for immanent real possibility

Kind of possibility Absolute necessity

Formal Actual forms of intuition (space, time) and actual forms of
understanding (unity of apperception, categories)
Empirical-causal Actual future, actual highest natural kind (matter) and its laws
Nomic Actual highest natural kind (matter) and its laws

35
Some readers will object that in a world w where none of the actual natural kinds are instantiated,
any p is compatible, given how I defined the compatibility relation: p is compatible with the essences of
actual natural kinds if and only if ¬(¬p is ground in the essences of actual natural kinds). If there are no
actual natural kinds in w, goes the objection, then, trivially, everything satisfies the right-hand of this
biconditional. However, this objection does not hold for the more involved technical definition of nomic
possiibility in Chapter 8.7.
36
See Ch. 8.8.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

generic conception of real possibility, truths which are retained from the pre-Critical
theory of real possibility. In particular:
(Real Possibility) For any kind of possibility ◇xp (and its associated kind of
necessity □xp, where □xp $ ¬◇x¬p), ◇x is a kind of real possibility (and □xp is a
kind of real necessity) only if
(i) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇xp (equivalently, it is
not a conceptual truth that □xp ⊃ □Lp), and
(ii) Groundedness: if ◇xp then the fact that ◇xp has a real ground in an actual
object or principle.
(iii) Absolute necessity: there is an associated notion of absolute necessity (abso-
lute-□xp) defined as follows: absolute-□xp =def ¬p ) (for all q, ¬◇xq). There
is some actual p about the grounds of all possibilityx such that absolute-
□xp.37
Kant generalizes from his pre-Critical conception of metaphysical real possibility
to a conception of real possibility in general by dropping the pre-Critical requirement
that real possibility be grounded in some feature of the world independent of our
minds. He retains the requirement that each kind of real possibility be interdefinable
with a related concept of real necessity, be conceptually distinct from logical possi-
bility, be grounded in some features of actuality, and that there be a corresponding
conception of absolute necessity.
This conception of absolute necessity is most problematic, but also most interesting,
in the case of ‘noumenal’ real possibility, my generic term for real possibility grounded
in noumena. Kant claims in the Pölitz lectures that all noumenal real possibilities are
grounded in an absolutely necessarily existing being. This was precisely the conception
of God and his absolutely necessary existence that Kant defended in Beweisgrund,
translated into the Critical terminology. In the Pölitz text he claims that it is “subject-
ively necessary” to posit such a being. In the next section I use Kant’s theory of ‘doctrinal
belief ’ to explain what it could mean that it is “subjectively necessary” for us to posit the

37
In Chapter 5.2 I discussed the ‘reductive’ reading that identifies absolute necessity and necessity
simpliciter. Similarly, one might argue that, in the Critical system, ‘absolute real necessity’ of some kind
(e.g., absolute formal necessity) is just real necessity simpliciter, interdefined with real possibility. However,
this cannot be so, as this brief argument shows:
(1) It is really possible that p if and only if it is not absolutely necessary that ¬p. [Assumption]
(2) (It is absolutely necessary that p) if and only if (¬p □! nothing is really possible). [Definition of
absolute necessity]
(3) ∴ (It is really possible that p) if and only if ¬ (p □! nothing is really possible). [From (1) and (2)]
(4) (There is a ground of the real possibility that p) if and only if (it is really possible that p).
[Assumption]
(5) ∴ (There is no ground of the real possibility that p) if and only if (¬p □! nothing is really possible).
But this conclusion is absurd, for it entails that if there is no ground of the real possibility that I act
otherwise than I actually do (perhaps because my noumenal will lacks the freedom to do so), the hypothesis
that I act otherwise than I do cancels all real possibility. Absolute real necessity cannot be identified with
real necessity simpliciter and thus interdefined with real possibility; premise (1) should be rejected.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

existence of something, and why, in particular, this is the case with the absolutely
necessary ground of all noumenal real possibility.

9.5. Postulates of Pure Theoretical Reason

Kant’s theory of doctrinal belief is found in the Canon of Pure Reason in the CPR, in
the section entitled “On opining [meinen], knowing [wissen], and believing [glau-
ben]” (A820/B848–A831/B859).38 Kant there distinguishes several different “modes”
of “holding-to-be true” [Fürwahrhalten], different epistemic statuses that a judgment
can have for an agent. Since the epistemic status of a judgment can vary from agent to
agent (e.g., Friedrich merely believes that p, while Gottfried knows that p), Kant
needs to distinguish between judgment-contents (or judgment-types individuated by
their content, which I have called propositions) and the relation of that judgment-
content to an agent who judges that it is true. Kant introduces Fürwahrhalten to refer
to the second: an agent’s act of judging that some content is true.39 Consequently, one
and the same judgment-content can be the target of different acts of Fürwahrhalten
by different agents. I have used ‘proposition’ to refer to judgment-contents through-
out this study; I will sometimes refer to Fürwahrhalten simply as judging.40
The different modes of Fürwahrhalten correspond to different epistemic statuses
that an act of judging can have: the agent can have the mere opinion that p, believe
that p, know that p, etc. Kant’s classification of these different epistemic statuses is
quite complex, but for our purposes the most important such status is belief [Glaube],
which Kant defines as follows: “if taking something to be true is only subjectively
sufficient and is at the same time held to be objectively insufficient, then it is called
believing” (A822/B850).41 This is only informative if we know what ‘subjective
sufficiency’ and ‘objective (in)sufficiency’ are. Kant writes in the same passage that
“subjective sufficiency is called conviction [Überzeugung] (for myself ),” a remark he
does not elaborate on until two pages later, where he writes that “the usual touch-
stone of whether what someone asserts is mere persuasion or at least subjective
conviction, i.e. firm belief, is betting.”42 The ‘subjective sufficiency’ that defines belief
that p as a mode of Fürwahrhalten is what would now be called ‘credence’ and is

38
See also the parallel discussions of Fürwahrhalten in JL (Ak. 9: 65–73) and the logic lectures: LB (Ak.
24: 148–53), LP (Ak. 24: 436–41), LPö (Ak. 24: 541–5), LBu (Ak. 24: 637–9), LDW (Ak. 24: 732–5), and WL
(Ak. 24: 850–7). My discussion of these issues is deeply indebted to Chignell (2007a) and (2007b). I will not
bother to indicate minor differences in our interpretations, though, for I am interested primarily in Kant’s
theory of ‘doctrinal belief ’. See also Hanna (2006), 258–61.
39
Cf. JL: “the judgment through which something is represented as true, the relation to an understand-
ing and thus to a particular subject, is subjectively, holding-to-be-true” (Ak. 9: 65–6). See A74–5/B100–1.
40
Kant’s complete theory of judgment includes the modality of judgment, a complication I suppress in
the body of the text. A subject can relate to a judgment-content in one of three ways: problematically
(entertaining the content without asserting it), assertorically (asserting it), or apodictically (asserting it by
deriving it from a rule). See A74–6/B100–1.
41
Cf. JL (Ak. 9: 67), LPö (Ak. 24: 541), LBu (Ak. 24: 638), LDW (Ak. 24: 732), and WL (Ak. 24: 852).
42
Cf. JL (Ak. 9: 73) and WL (Ak 24: 852–3, 855).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

measured by what bets one is willing to take on the proposition that p: the more one
is willing to wager, the greater the degree of ‘subjective sufficiency’ or ‘subjective
conviction’ in one’s judgment that p. Kant does not state a precise lower bound on
‘subjective sufficiency’ that differentiates belief from mere opinion, but the idea,
I take it, is clear enough; opining that p is subjectively insufficient insofar as one
entertains the proposition that p but would not be willing to bet anything on it. Kant
allows that we can believe things we could never (in principle) verify or falsify, but
maintains that the ‘usual touchstone’ still applies: we are to imagine that there is some
way of verifying or falsifying the proposition and ask ‘what would one be willing to
wager that p?’ “If it were possible by any sort of experience to verify whether there are
inhabitants of at least some of the planets that we see, I might well bet everything I have
on it” (A825/B853). Kant is expressing his strong, albeit unverifiable (at least in
1781–2016), belief that some planets in the universe other than Earth are inhabited.
Whether it makes sense to apply this test to propositions that could not in principle be
verified because they concern non-sensible objects (e.g., that God exists, or that the soul
is immortal) is a point Kant does not address. Even if it is incoherent to ask ‘how much
would you bet that there is an absolutely necessary ground of possibility?’, Kant’s idea
still applies: belief requires subjective conviction, this subjective conviction comes in
degrees (beliefs require more than mere opinion), and the ‘betting’ test (“the usual
touchstone”) is one intuitive way to measure this degree of subjective conviction. This
means that belief is defined in terms of one’s subjective degree of credence, not in terms
of any positive epistemic status (like justification); there can be beliefs that lack any
rational standing, and it is psychologically possible to have greater credence in what
one believes than in what one knows to be true.43
Before we understand the positive epistemic status of ‘rational belief,’ though, we
need to understand the ‘objective sufficiency’ that defines knowledge [Wissen] and
distinguishes it from belief (so that we can understand why even rational beliefs are
not knowledge). In the paragraphs leading up to his discussion of belief, Kant
distinguishes mere persuasion [Überredung] from conviction: “the touchstone [Pro-
bierstein] of whether taking something to be true [Fürwahrhalten] is conviction or
mere persuasion is therefore, externally, the possibility of communicating it and
finding it to be valid for the reason of every human being to take it to be true
[Fürwahrhalten]” (A820–1/B849). Kant here is talking about the intersubjective
validity of a judging, that is, whether it is based on grounds (reasons) that are
grounds (reasons) for other epistemic agents to concur in the judging that p.
I interpret ‘touchstone’ [Probierstein] here as meaning a necessary but not sufficient
condition on conviction [Überzeugung],44 so that if some judging is an instance of

43
See JL (Ak. 9: 70), LB (Ak. 24: 148–50), LPö (Ak. 24: 543), LBu (Ak. 24: 638), LDW (Ak. 24: 734), and WL
(Ak. 24: 852, 855). I will not discuss here Kant’s further claim (made in several of these texts) that it is rational
to have a higher degree of credence in moral beliefs than in what one knows to be true.
44
Not the subjective conviction discussed above, but objective conviction. Cf. Chignell (2007a), 40–4.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

conviction, then it is based on intersubjectively valid grounds (but not vice versa).
Since this notion of judging on intersubjectively valid grounds will be crucial for
Kant’s theory of rational belief, it is worth being precise about it:
(IS-valid ) S’s judging that p is intersubjectively (IS) valid if and only if 9g such
that (i) S bases her judging that p on g, and (ii) g is an epistemic ground (reason)
for S and other rational agents to judge that p.
The range of rational agents for which g is a ground to judge that p can vary; this
allows us to distinguish degrees of IS-validity. At one extreme will be judgments based
on grounds that are valid for all finite rational agents as such; I will refer to this as
maximal IS-validity. The quoted sentence continues with this remark: “for in that
case [IS-validity] there is at least a presumption that the ground of the agreement of
all judgments, regardless of the difference among the subjects, rests on the common
ground, namely the object, with which they therefore all agree and through which the
truth of the judgment is proved” (A821/B849). I interpret this to mean that, in the
case of judging on IS-valid grounds, there is a presumption that these IS-valid
grounds are objectively valid (in a sense to be explained below), and hence constitute
a case of conviction (knowledge), but that there can be cases of judging on IS-valid
grounds that are not conviction (knowledge).
This is Kant’s point when he introduces opinion, belief, and knowledge: “taking
something to be true, or the subjective validity of judgment, has the following three
stages in relation to conviction (which at the same time is valid objectively): having
an opinion, believing, and knowing” (A822/B850). When I opine that p, I do not even
have subjective conviction that p because I would not bet that p. When I believe that
p, I am willing to bet that p, and in some cases (which I will discuss further below),
my belief that p is based on IS-valid grounds, but I am aware that my grounds are not
objectively valid. Only in the case of knowledge do I judge that p with subjective
sufficiency (I am willing to bet that p) on the basis of grounds that are IS-valid and
objectively valid.45 This is shown in Figure 9.2.46
Before continuing to the main topic of this section, Kant’s theory of doctrinal
belief, I want to discuss, briefly, the ‘objective validity’ of grounds that distinguishes
knowledge from belief. It is crucial to realize that the grounds in question (the grounds
for S’s judging that p) are not a prioriG grounds of the truth of p, either causal or non-
causal grounds, but epistemic grounds. If they were a prioriG, then all knowledge (all
knowledge from ‘sufficient grounds’) would be a prioriG knowledge, but this is clearly
not Kant’s view; he allows that in both empirical and mathematical cases there is
genuine knowledge from the consequences (a posterioriG knowledge).47 So the

45
I am persuaded that p when I have subjectively sufficient grounds for p (I am willing to bet that p) but
I mistakenly take these grounds to be either IS-valid or objectively valid, or even both. Persuasion
[Überredung] is not simply the complement of conviction. When one has rational belief that p one is
neither persuaded nor convinced that p.
46 47
Cf. Chignell (2007b), 358. JL (Ak. 9: 70–1), LPö (Ak. 24: 544).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

Modes of Fürwahrhalten

Opinion Belief Knowledge

~SS SS, ~OV SS, ISV, OV

Mere Belief Rational Belief

SS, ~OV, ~ISV SS, ~OV, ISV

SS=subjective sufficiency (subjective OV=objectively valid grounds

conviction) ISV+OV=conviction
ISV=intersubjectively valid grounds

Fig. 9.2. Modes of validation

notion of grounding that defines knowledge is not the metaphysical notion of a

ground that makes something possible or makes something exist, but the epistemic
notion of a ground that confers warrant or justification on a judgment (although the
epistemic ground may also be the ground of possibility or of existence, in the case of a
prioriG knowledge).
In the empirical case, Kant seems to regard g as an objectively sufficient ground for
judging that p just in case g makes p sufficiently probable (where p is also true); this
allows for empirical knowledge by testimony, for instance, knowledge of history
from fallible sources.48 In the a priori case, though, Kant requires “consciousness
of necessity” (JL, Ak. 9:71).49 This cannot simply mean that the proposition
(judgment-content) p is itself necessary, for this will not distinguish between cases
where one knows a necessary truth and where one merely thinks one knows a necessary
truth. The ‘necessity’ characterizes not the proposition (judgment-content) itself, but
the relation between one’s ground and the proposition one judges: the ground must
necessarily entail the proposition (in the a prioriJ case).50 In other words:
(Obj. validity) S’s judging that p is objectively valid if and only if 9g such that (i)
S bases her judging that p on g, and (ii) g obtains, and either (iii) g makes p
sufficiently probable (empirical knowledge) or (iv) necessarily, if g obtains then p is
true (a priori knowledge).51

48 49
JL, Ak. 9:70–1. Cf. Chignell (2007a). Cf. LPö, Ak. 24: 544.
50
Hence Kant’s characterization of Wissen as apodictic judgment (JL, Ak. 9: 66); the apodictic mode of
judging is defined at A76/B101 as judging that some content is a necessary consequence of a rule (“the
apodictic judgment [Satz] thinks the assertoric one as determined through these laws of the understanding
itself ”). The example in that passage is an example in which the apodictic judgment is a logical conse-
quence of the epistemic ground on which it is based (the premises in a hypothetical syllogism). But Kant
must generalize the necessity here beyond logical necessity in order to make sense of, for instance, synthetic
a priori mathematical knowledge.
51
Chignell rejects my clause (i), which requires that S be aware of the ground and base her judgment on
that ground. On this issue, he opts for a counterfactual analysis: on reflection, S would cite g as her ground;
see Chignell (2007a), 45. He cites the Blomberg logic (Ak. 24: 87–8), but I don’t see how that passage
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

The obvious question about this principle, especially in the context of this study, is
what notion of necessity is involved? My answer is that the notion of necessity
involved depends upon the content of p: if p is an analytic judgment, knowing that
p requires judging on the basis of a logically necessitating ground; if p is a mathem-
atical claim, knowing that p requires judging on the basis of a mathematically
(formally) necessitating ground; if p is about an alteration in a substance then
knowing that p requires judging on the basis of an empirically-causally necessitating
ground;52 and if p is about laws, then knowing that p requires judging on the basis of
a nomically necessitating ground. My definition also remains neutral on the onto-
logical status of grounds; in some cases they might be other propositions, or objects,
etc. The phrase ‘g obtains’ is meant to be category-neutral: whatever kind of thing a
given ground is, necessarily, if that ground is present (true, in the case of proposi-
tions; absolutely posited in the case of objects, etc.) then the proposition for which it
is the objectively sufficient ground is sufficiently probable or true.
Since Kant defines belief as subjectively sufficient judging consciously based on
objectively invalid grounds, we can also now state his definition of belief:
(Belief ) S believes that p if and only if S subjectively sufficiently judges that p and
S is aware that her judging that p is not objectively valid.
Note that the definition of belief leaves open whether it is based on IS-valid grounds
(rational belief) or whether it lacks rational grounds (mere belief).
The rest of the section “On opining, believing, and knowing” is devoted to rational
belief, of which kant gives various examples, but never precisely defines what makes
rational belief rational. The one feature all of his examples share, though, is that they
are based on intersubjectively valid grounds. The degree of inter-subjective validity
varies from example to example, and the nature of the grounds themselves vary (e.g.,
practical or theoretical grounds), but their in-principle shareability does not vary.
I propose, consequently, that we define rational belief in terms of IS-validity, from
above:
(Rat. Belief ) S rationally believes that p if and only if (i) S believes that p, and
(ii) S’s judging that p is IS-valid.

supports his reading. Clause (iv) is not intended as an analysis of a prioriJ knowledge, but merely to indicate
that in the a prioriJ case one’s ground must do more than merely make p probable. One might also want to
insert a clause (v) to the effect that S must be aware of (ii)–(iv). However, adjudicating whether Kant
maintains that knowledge that p requires being aware that one’s grounds for judging that p are objectively
sufficient (rather than merely requiring that they are objectively sufficient) would require more space than
I have here.
52
There is a complication here. Knowledge of alterations in substances is not a prioriJ so we might think
this should fall under the empirical clause (mere probability is sufficient). However, I think there are two
ways to know alterations in substances: purely empirically (in which case an objective ground that makes it
sufficiently probable is enough) and knowledge a priori secundum quid (see Stang 2011 and Ch. 7.4), in
which one derives the alteration from the antecedent state of its cause and thus cognizes its empirical-
causal necessity.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

Consequently, we can distinguish the range of intersubjective validity that rational

beliefs have: some rational beliefs will be intersubjectively valid only for agents who
share my contingent ends (what Kant calls ‘pragmatic’ beliefs, A824/B852), while
others will be intersubjectively valid for all agents who share the moral end of
bringing about the highest good (an end shared by those with a good moral
disposition [Gesinnung], according to A829/B857).53
While the lion’s share of scholarly attention has focused on the latter class of
practically rational beliefs (what Kant calls ‘moral beliefs’), my focus will be on what
he calls doctrinal belief, a “theoretical analogue” (A825/B853) of practical beliefs.
Kant’s clearest articulation of what doctrinal belief is occurs in this long passage:
Now we must concede that the thesis of the existence of God belongs to doctrinal belief. For
although with regard to theoretical knowledge of the world I have nothing at my command
that necessarily presupposes this thought as the condition of my explanations of the appear-
ances of the world, but am rather obliged to make use of my reason as if everything were mere
nature, purposive unity is still so important a condition of the application of the reason to
nature that I cannot pass it by, especially since experience liberally supplies examples of it. But
I know of no other condition for this unity that could serve me as a clue for the investigation of
nature except insofar as I presuppose that highest intelligence has arranged everything in
accordance with the wisest ends. Consequently, the presupposition of a wise author of the
world is a condition of an aim which is, to be sure, contingent but yet not inconsiderable,
namely that of having a guide for the investigation of nature. The outcome of my experiments
also so often confirms the usefulness of this presupposition, and nothing can be decisively said
against it, so that I would say too little if I called my taking it to be true merely having an
opinion, but rather even in this theoretical relation it can be said that I firmly believe in God;
but in this case this belief must not strictly be called practical, but must be called a doctrinal
belief, which the theology of nature (physico-theology) must everywhere necessarily produce.
(A826–7/B854–5)54

Kant’s point here is that the judgment that there is an intelligent author of nature
(the role in which God is being cast here) is not based on an objectively valid ground.
My reason for judging that there is an intelligent author of nature is not that I directly
perceive one or possess some proof of its existence (Kant undermined such proofs
in the Transcendental Dialectic), but that postulating an intelligent author is a means
to the end of finding natural explanations of natural phenomena.55 For instance,

53
That these ends are morally obligatory does not entail that they are actually shared by other rational
agents; however, the fact that the moral obligation to set those ends extends to all finite rational agents
entails, presumably, that all finite rational agents have sufficient reason to set those ends, and hence to share
those beliefs. Cf. CPrR (Ak. 5: 143), JL (Ak. 9: 86 n.), and WL (Ak. 24: 851).
54
Kant claims that belief is only possible on practical grounds (ends of action) at A823/B851, but he
appears to take this back three pages later when he gives his theory of doctrinal beliefs. He is more skeptical
about the possibility of theoretical rational belief in the logic lectures: JL (Ak. 9: 68 n., 69), LBu (Ak. 24:
638–9), and WL (Ak 24: 852, 855 f.).
55
Some caution is required here. Kant denies that belief in an intelligent cause of nature or (by parity of
reasoning) in an absolutely necessary ground of possibility is a “Hypothese” because it does not contain a
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

if I postulate that nature is the product of an intelligent author, if I were to find that
there are two species of natural object that seem to have nothing to do with one
another, to be entirely incommensurable and governed by entirely different laws,
I will not stop at this stage of inquiry, but will press on until I find some more
fundamental understanding of both species that explains how they are particular
determinations of a more general kind. I will do so because I have postulated a wise
author who would not create a system of nature that is so disunified, one that lacks
‘purposive unity.’56 The postulate will lead to greater and richer empirical knowledge.
Consequently, postulating the existence of an intelligent author of Nature is a means
to one of my theoretical ends—to have richer and deeper explanations of the natural
world. What is more, Kant claims it is a necessary means to that end: “purposive unity
is still so important a condition of the application of the reason to nature that
I cannot pass it by.” This fact is an intersubjectively valid ground for all other agents
who share the end of understanding Nature to believe that there is an intelligent
author, which makes this belief a case of rational belief. Since this ground does not
necessarily entail (or even significantly raise the probability) that there is an intel-
ligent author of Nature, this is a case of mere belief, not knowledge. Notice, though,
that this is a case of rational belief only if it is based on an intersubjectively shareable
ground. Kant is assuming, not implausibly, that our theoretical ends are in principle
shareable by other epistemic agents; since our theoretical ends are given by the
nature of reason itself (more about this in section 6), and reason has the same nature
in all rational agents, theoretical ends of one agent are in principle shareable by
others.
Kant goes on in the rest of the section to develop his theory of moral beliefs, cases
where believing something (e.g., that the highest good is possible) is a rational
requirement on setting an end (e.g., the achievement of the highest good) that is
morally obligatory for all (finite) rational agents. Because these ends are morally
obligatory, the relevant belief is based on grounds that are valid for all rational
agents.57 In his discussion of doctrinal belief, Kant claims that the theoretical end
in question—systematicity in our explanations of nature—is merely a “contingent”

determinate enough concept of its object (A827/B855) to explain why nature exists or why anything is
really possible. As we have already seen, though, there has been a problem in Kant’s view at least since 1763
about representing the determinate content of the idea of God as the ground of real possibility (see Ch. 4);
in the next chapter I will address this issue in the Critical philosophy, and thus address (in the terms of
A827/B855) whether rational belief in the existence of an absolutely necessary being is a genuine
‘hypothesis’ or not. Cf. the discussion of hypotheses in LP (Ak. 24: 439–40). Thanks to Noam Hoffer for
pressing me to clarify my views on this point.
56
I take ‘purposive unity’ to be the unity a thing has if it is created by a wise and intelligent being. The
notion of ‘purposiveness’ is discussed extensively in the CJ. This passage, on the necessary doctrinal belief
in a wise nature, should be compared with the two Introductions to that work, as well as to the discussion of
the regulative use of the idea of God in the CPR in the Appendix to the Transcendental Dialectic
(A669–704/B697–732).
57
A828/B856 and CPrR, Ak. 5: 119–46.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

theoretical end of the agent in question.58 This raises the intriguing possibility,
though, of a theoretical analogue of moral belief: judgments that must be believed
in order to satisfy a necessary end of theoretical reason.59 Kant calls moral beliefs—
beliefs mandated by the nature of practical reason itself—‘postulates of pure practical
reason.’ We might therefore think of necessary theoretical beliefs as postulates of
pure theoretical reason. To make the analogy complete, we will need to impose on
theoretical belief a requirement that Kant imposes on moral belief: its object must be
logically possible, and must not be knowably really impossible.60 No matter how
much it would serve the ends of reason for us to believe that p, if p is logically
inconsistent or provably really impossible, then we cannot have any rational warrant
for accepting it. I will call this the ‘no impossibilities’ rule.
Note, though, that if all rational agents as such have end E, and subjectively
sufficiently judging that p is a necessary means to E, then all rational agents as
such have grounds to judge that p. So, if there is some end E that is set by the nature
of theoretical reason (that is, solely in virtue of having theoretical reason we set
theoretical end E) and there is some p such that subjectively sufficiently judging that
p (holding p to be true with subjective conviction) is a necessary means for satisfying
E, then we can have rational belief that p (belief based on IS-valid reasons). Recall,
however, what Kant says about the postulate of an absolutely necessary being: it is
“subjectively necessary” and “grounded in the nature of human reason.” This sug-
gests that belief in an absolutely necessary being is a necessary theoretical belief,
where the end in question is explaining why anything is (noumenally) really possible,
that this end is given by the nature of theoretical reason, and that this belief is a
necessary means for that end because it is the only means by which we can explain
why anything is (noumenally) really possible.61
In the next section I argue that, on Kant’s view, these conditions are satisfied. We
can have the theoretical rational belief that there is an absolutely necessary being if we
base our belief on the fact that, without this being, we cannot explain noumenal real
possibility. What is more, this theoretical belief is rationally obligatory for us: it is the
only way to satisfy a necessary theoretical end.62 This is belief rather than knowledge

58
Although he denies this in both the Appendix to the Transcendental Dialectic and in the CJ; see
previous discussion.
59
Kant encourages the thought when he describes doctrinal belief as “a merely theoretical [ . . . ]
analogue of practical belief ” (A825/B853).
60
Cf. Chignell (2007b), 347–50.
61
Chignell (2007b), 349 anticipates my point here, even discussing the Pölitz text with which I began
and to which I return later in this chapter. Chignell also points to the following Reflexion: “the principle of
the self-preservation of reason is the basis of rational belief, in which assent [Fürwahrhalten] has the same
degree as knowledge, but is of another kind which comes not from the cognition of grounds in the object
but rather from the true needs of the subject in respect to theoretical as well as practical application” (Refl.
2146, Ak. 16: 371–2).
62
Robert Hanna develops an analogous explanation of Kant’s belief that cognitive subjects are ‘nou-
menally’ affected by things in themselves: reason demands an explanation of our sensory states. See Hanna
(2001), 117.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

because it is based on a ground that is not objectively sufficient for the existence of
God; the fact that this postulate allows us to satisfy a necessary end of theoretical
reason (explain why anything is noumenally really possible) does not make more
probable or necessarily entail that God exists.63

9.6. The Necessary Ends of Theoretical Reason

The conclusion of the previous section is that we can have a species of Kantian
belief—which I characterized as ‘necessary theoretical belief ’—in the existence of an
absolutely (noumenally) necessary being that grounds all (noumenal) real possibility
if two conditions are met:
(a) It is a necessary end of theoretical reason to explain why anything is (nou-
menally) really possible at all.
(b) The necessary means to explain this is to posit the existence of an absolutely
(noumenally) necessary being that grounds all (noumenal) real possibility.
In this section I explore Kant’s theory of (theoretical) reason in the Transcendental
Dialectic of the CPR and his Critical reconstruction of the Beweisgrund argument in
the Transcendental Ideal. I argue that they entail (a) and (b). This is what Kant means
when he says in the Pölitz lectures that “[the Beweisgrund argument] can in no way
be refuted, because it has its ground in the nature of human reason” (Pöl.RT, Ak. 28:
1034).64
Claim (a) concerns the faculty of theoretical reason [Vernunft], which differs in
crucial respects from the faculty of understanding [Verstand]. The understanding is
the faculty by which we think about sensibly given objects using concepts; conse-
quently, the understanding is the faculty of cognition: representations of knowably
really possible objects (those that can be given in intuition).65 The faculty of reason
does not produce cognition but brings unity to the cognitions generated by the faculty
of understanding. Reason has two uses: a logical use, and a ‘pure’ use (which I discuss
below). The logical use of reason is to bring logical unity to our cognition. Our
cognition has logical unity to the extent that individual cognitions (e.g., individual
judgments) are explained by higher-level principles. Reason discharges its logical

63
Kant does remark in “Opining, believing, and knowing” that “there is something unstable about
doctrinal belief; one is often put off from it by difficulties that come up in speculation, although, to be sure,
one inexorably returns to it again” (A828/B856). I take this to be a reference to the antinomial conflicts that
afflict reason when it attempts to think about supersensible objects. Kant’s resolution of those antinomial
conflicts, however, clears the way for the possibility of a non-wavering doctrinal belief in an absolutely
necessary being, once we have proved that it is the unique way to answer reason’s demand for a ground of
possibility and have understood that it is merely a belief, not a claim to know supersensible reality.
64
In the rest of this section ‘reason’ means specifically theoretical reason, or reason in its theoretical use.
I am not attempting to give an interpretation of Kant’s theory of practical reason, much less a unified
account of reason as such.
65
See Ch. 6.6.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

function by subsuming individual items of cognition under higher-level explanatory

principles.66
The logical use of reason is governed by the rule: seek a condition [Bedingung] for
every conditioned [bedingt] object. I will refer to this as the logical principle of reason
(LPR for short). I take ‘condition’ in the Transcendental Dialectic to mean ground in
the very general sense of ‘ground’ used in pre-Critical German rationalism (including
by Kant himself): it includes logical grounds, real grounds, causal grounds, etc.,
though it excludes merely “consequentially determining grounds” (ND, Ak. 1: 392),
grounds of mere a posterioriG knowledge. Reason seeks conditions, not conse-
quences.67 Kant uses this term in the Dialectic to indicate that it is wider than any
of the grounding relations he has considered up to that point in the CPR. Conse-
quently, the relation of ‘a condition to what it conditions’ is a very broad notion that
embraces, among other things, the relation of a spatial region to its parts, the relation
of a material part to a material whole, the relation of a cause to its effect, etc.
Kant writes: “we see very well that the proper principle of reason in general (in its
logical use) is to find the unconditioned for conditioned cognitions of the under-
standing, with which its unity will be completed” (A307/B361). The LPR leads to the
idea of the complete series of conditions for a given object; since the complete series
contains all of the conditions of the object, it is not conditioned by anything outside
it, so we can also think of it as the unconditioned series of conditions for that object.
Since there are different relations of condition to conditioned (of ground to conse-
quence), the concept of the unconditioned series of conditions is always the concept
of the series of objects that stand in a specific grounding relation to the original
conditioned object, which has no grounds of the relevant kind outside that series.68
For instance, given an alteration in time, reason forms the idea of the complete series
of causes of that alteration, which itself has no causal condition external to the series
itself (though it might have spatial conditions, e.g., the larger region of space which
bounds and conditions the region in which this causal series occurs). The concept of
the unconditioned is an idea in Kant’s technical sense: a (logically consistent) concept
of an object that could never be given in experience because it requires a complete-
ness (of conditions) that is never present in any object of experience.69
This “principle of reason in general (in its logical use)” expresses a necessary end
of theoretical reason, an end theoretical reason sets us in virtue of being the faculy
that it is.70 It “is merely a subjective law of economy for the provision of our
understanding” (A306/B362). That we should search for a condition for every
conditioned object and form the idea of an unconditioned being does not entail, or
presuppose, that this goal could ever be reached, or that there is an unconditioned
being. It does not tell us whether there is an unconditioned series of conditions for a

66 67 68 69
A305/B361. A331/B388. A323/B379. A320/B377.
70
See Stang (2014a) for more on the idea that Kantian faculties have certain ends in virtue of being the
faculties they are.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

given object. In particular, Kant warns us not to make a “pure use” of reason and
assume that “when the condition is given then the whole series of conditions,
subordinated to one another, which is itself unconditioned, is also given” (A308/
B364). To avoid confusion, I will refer to this assumption as the pure principle of
reason (PPR). Mistaking the LPR, a mere “subjective law of economy,” for the PPR is
what Kant calls “transcendental illusion,” and the goal of the Transcendental Dia-
lectic is to uncover transcendental illusion as the source of dogmatic metaphysics
and, as much as possible, dispel it.71 What does the ‘givenness’ of an unconditioned
series mean here? Recall that intuitions ‘give’ us objects; in Chapter 5 I interpreted
this to mean that intuitions allow us to have thoughts about the objects there are
(q-objects). Applying this reading of ‘givenness’ to the PPR, I interpret it as the
assumption that, necessarily, if there is a conditioned object, then there is an
unconditioned series of its conditions. The two requirements on theoretical belief,
(a) and (b) from above, can be reformulated as claims about the principle of reason:
(a*) The LPR requires us to think of an unconditioned series of conditions for
the fact that anything is (noumenally) really possible (the conditioned).
(b*) The only conception of the unconditioned series of conditions of the
(noumenal) real possibility that would satisfy the LPR is that it terminates in an
absolutely necessary being, one whose non-existence would cancel all (noumenal)
real possibility.
I will argue that both are the case.
The Transcendental Dialectic contains a discussion of reason’s search for a ground
of (noumenal) real possibility that mirrors Kant’s Beweisgrund argument, “The tran-
scendental ideal (prototypon transcendentale)” (A571–83/B579–611). However, the
close connection between that text and Beweisgrund is partly obscured by Kant’s
framing of the discussion around the issue of complete determination. Readers will
recall from Chapters 4 and 5 that the complete determinacy of existents or of
possibilities played no significant role in the positive argument of Beweisgrund.
The reason he gives them such prominence in the Ideal is that he assigns the Ideal
an additional dialectical function: to diagnose the transcendental illusion involved
in Baumgarten’s complete determination theory of existence. Hence, my explication of
the Transcendental Ideal, and Kant’s Critical attitude towards the Beweisgrund argu-
ment, will begin with a brief discussion of complete determination. I will then attempt
to bracket the issue of complete determination and read the Transcendental Ideal as
Kant’s reconstruction, in Critical terms, of the original Beweisgrund argument.
It is impossible for us to cognize objects as completely determinate. For any object
of a possible experience, there are always (empirical) predicates of the object that we
do not experience. To make things concrete, take a body, which exercises mutual

71
A297/B353. However, Kant describes transcendental illusion as inevitable at A297/B354, A298/B355,
and A339/B397. For a discussion of the ‘inevitability’ thesis, see Grier (2001), 126–30.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

attractive force on all other bodies in space. For any possible experience of that object,
there is some distant region of space R, such that the experience leaves open whether
there are bodies in R on which the given object exerts a (very small) mutual attractive
force, so our experience leaves open whether it has, or whether it lacks, the predicate
exercises mutual attractive force on bodies in region R (call this predicate FR). Now,
for any such region R, we could have a more complete experience that includes
experience of region R, which would represent the body as determinately having or
lacking predicate FR. But for that experience there will be another distant region R*
such that the more complete experience leaves it indeterminate whether the body has
predicate FR*. The point is that for any such predicate, there is a possible experience
that would settle whether the object determinately has or determinately lacks the
predicate. But for any such experience, there is some predicate such that the
experience does not settle whether the object determinately has or lacks it. Conse-
quently, for any possible experience the set of predicates it represents its objects as
having is incompletely determinate; there will always be a predicate F such that
neither F nor its negation ~F is included in the set.
An experience of an object that does not represent it as having or lacking (being
determinate with respect to) a given predicate is a limitation of (is conditioned by)
some possible ‘larger’ experience that would be determinate with respect to that
predicate. So the logical principle of reason (LPR) sets us a task: to have a more
complete experience of the object that would represent a more determinate set of
predicates. Reason thus forms the idea of an unconditioned series of such conditions:
an unconditioned series of ever more inclusive experiences, corresponding to an ever
more determinate set of experienced predicates. Reason also forms the idea of the
complete set of predicates that could be represented in any such series of experiences
and thus appear in any corresponding series of sets of experienced predicates. Call
this the set of all possible predicates.72
The first question one might have about this story is: what does ‘possible’ mean
here? Let us take this, for now, to be a discussion of formal possibility, so the idea of
the complete space of possible predicates is the complete space of formally possible
predicates. But notice that there are two places where we can (illegitimately) apply
the PPR. First, from the fact that we set ourselves the theoretical end of more
completely experiencing objects (experiencing a more complete set of their predi-
cates) it does not follow that there is a completely determinate experience, or even an
unconditioned series of ever more completely determinate experiences. We cannot

72
This is my reconstruction of Kant’s highly condensed discussion in the last paragraph of the Ideal,
A582–3/B610–11. First, we illegitimately apply the PPR and assume that the unconditioned series of
experiences is given, and hence that the complete space of possible predicates of phenomena is given. Then
(as Kant explains in the previous paragraph, A581/B609) we represent every phenomenon as having its
properties in virtue of how it is represented in this one sum total of all experience. Then we apply this
principle (individual things have fully determinate sets of possible properties in virtue of their relation to
some sum-total of all possible properties) to objects in general (phenomena and noumena).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

assume that the unconditioned series of conditions is given. Consequently, it does

not follow that the set of predicates of objects forms an unconditioned series. In other
words we have no guarantee that ‘the set of all possible predicates of objects’ refers
to any set. This constitutes Kant’s explanation of the transcendental illusion (the
illegitimate application of the PPR) that generates the assumption that objects are
completely determinate and that there is a complete space of all possible predicates of
objects. This in turn, Kant thinks, leads to Baumgarten’s view that beings as such (not
just objects of possible experience) are completely determinate.
But the Transcendental Ideal contains a thread that is separable from the idea of
complete determinacy, the thread that represents Kant’s original thinking in Beweis-
grund: “if we consider all possible predicates not merely logically but transcenden-
tally, i.e. as to their content which can be thought in them a priori, then we find that
through some of them a being is represented, and through others a mere non-being
[ . . . ] all concepts of negations are thus derivative, and the realities contain the data,
the material, so to speak, or the transcendental content, for the possibility and the
complete determination of all things” (A574–5/B602–3). Not only does this recap-
itulate exactly the line of thought from Beweisgrund—logically complex predicates
are possible only if their atomic constituents, the data of all possibility and all
thought, are themselves really possible—it remains a compelling line of thought for
reason [Vernunft] even if we eliminate the issue of complete determination. We do
not need to assume that there is a set that contains all possible predicates in order for
reason to ask: what is the condition that explains why there are really possible
predicates at all? Nor do we need to assume that there is a set of all possible predicates
in order to ask whether there is a set of fundamental predicates (a subset of the
realities Kant speaks of above at A575/B603) such that there is no really possible
predicate (at no point in the ascending series of sets of possible predicates) that is not
ultimately derivative of those fundamental predicates. To return to my concrete
example, it is implausible that the predicate exerts mutually attractive force on matter
in region R is a fundamental empirical predicate. So even if ever more complete
experiences will reveal ever more predicates in a series of conditions in indefinitum,
this does not mean that they will reveal further fundamental predicates.
To apply LPR and its quest for conditions to the possibility of properties we need to
specify what concept of real possibility is involved here. We already have a complete
account of the conditions of formal, empirical-causal, and nomic possibility. However,
even if we can know that some putatively possible predicate is compatible with our
forms of experience (it is formally possible that an object have it), with the laws (it is
nomically possible that an object have it), and with the laws given the past (it is
empirically-causally possible that an object have it), we can still ask whether it is
noumenally possible, that is, whether it has a noumenal ground of its possibility.73

73
Cf. the discussion of the possibility of a different “single, all-encompassing” experience (A231–2/
B284) in Ch. 7.5.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

Kant himself explicitly attributes this question to reason (not the understanding) in the
Postulates of Empirical Thinking in General: “[these] are proper questions, and are, be
sure, to be solved synthetically, though they fall under the jurisdiction of reason alone”
(A230/B284). Consequently, Kant’s discussion of possibility in the Transcendental
Ideal (prior to his discussion of the complete determinacy of experience at A581/
B609) is a discussion of noumenal possibility. We are to consider, he says, “the whole of
possibility, as the sum total [Inbegriff] of all predicates of things in general (Dinge
überhaupt)” (A572/B600). I take it that Kant is talking about ‘things’ here to remind us
that we are not talking only about objects of experience, but about ‘beings as such,’ as
Wolff or Baumgarten might put it, which includes phenomena and noumena; we are
trying to reconstruct a certain kind of pre-Critical metaphysics (indeed, Kant’s own) as
the product of transcendental illusion. In other words, we are now to apply LPR to the
noumenal real possibility of predicates of objects in general, phenomena and noumena.
Take some putatively noumenally really possible predicate P of a phenomenon or a
noumenon. By applying the LPR we are set the theoretical task of thinking of the
grounds (conditions) of the possibility of P. By definition, these are noumenal
grounds. By applying the LPR again, we are set the task of thinking of the noumenal
grounds (conditions) of the grounds (conditions) of the possibility of P, i.e. what
predicates in noumena make possible P. We can separate out logically complex
properties; the possibility of the logically complex properties is ultimately grounded
in (conditioned by) the possibility of their constituents. If a predicate is a mere
absence of positive determination, then it is a negation of a reality. The reality is its
condition, so we are interested in the real possibility of that reality (the condition).
Likewise for properties that are limitations of realities; since limitations are condi-
tioned, we can inquire into the ground of the possibility of the unlimited reality of
which they are limitations. As Kant writes in the Ideal, “all negations [ . . . ] are mere
limitations of a greater and finally of the highest reality; hence they presuppose it, and
as regards their content they are merely derived from it” (A578/B606). ‘Content’
here, as in Beweisgrund, means (noumenal) real possibility.
This means that by applying the LPR to the concept of noumenal real possibility
we are led to consider the fundamental predicates of noumena that make all other
preperties of noumena really possible, and which make noumenally possible (specif-
ically, noumenally-causally possible) predicates of phenomena possible. We can
distinguish predicates that are derivative from those that are fundamental, as we
did above. We can ask, further, whether the fundamental predicates are instantiated
by a plurality of different noumena, or whether there is a unique noumenon that
possesses all fundamental predicates, and whose possession of those predicates makes
all other noumenally possible predicates really possible. Note, though, that all of this
is independent of the idea of the complete determinacy of existing or possible things,
and from the assumption that there is a space of all possible predicates. For the
application of the LPR requires us to think for every possible predicate of the ground
of its possibility; this is compatible with it being the case that the space of all possible
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

predicates is indefinitely extendable, as may be the case with the space of all possible
predicates of objects of experience (see previous discussion).
This regressive search for the conditions of (noumenal) real possibility in the
Transcendental Ideal is the project of Beweisgrund, but transposed from a dogmatic-
metaphysical to a critical-subjective key: the project is not to give a metaphysical
theory of reality but to satisfy our “subjective law of economy,” the LPR. What are
presented in 1763 as metaphysical arguments appear, from the perspective of 1781, as
expressions of reason’s search for conditions. For instance, in Beweisgrund, having
established (in Critical terms) that noumenal real possibilities must have noumenal
real grounds, Kant argues that these possibilities cannot be ‘parceled’ out among a
plurality of grounds (the ‘pluralist’ view). In Chapter 5.5 I argued that, even on the
most charitable reconstruction, Kant’s argument relies on assuming (i) that there
cannot be an infinite ascending chain of grounds74 and (ii) that there must be a single
ground that explains why the various grounds of possibility postulated by the
pluralist can interact with objects in the same world.75 From a Critical standpoint
we can see these assumptions as grounded in reason’s regressive search for the
unconditioned and for unity. From this perspective, (i) expresses reason’s idea of a
series of conditions that terminates in an unconditioned condition; (ii) expresses
reason’s demand for a condition of the plurality of grounds. Kant addresses precisely
this point in the Ideal: “because one also cannot say that an original being consists in
so many derivative beings, since each of the latter presupposes the former and so
cannot constitute it, the ideal of the original being must also be thought of as simple”
(A579/B607). If the fundamental predicates are ‘parceled out’ among a plurality of
fundamental grounds (grounds of possibility, which are themselves ungrounded), any
one of which grounds only a subset of them, this will violate reason’s demand for
unity: the series of conditions that explain why various predicates are really possible
will terminate in multiple unconditioned beings. The precise point at which Kant’s
Beweisgrund argument was weak is the point at which Kant can make philosophical
progress by transposing his claims from the register of metaphysical arguments to that
of the subjective demands of reason: a plurality of fundamental grounds is rationally
unsatisfying because it lacks explanatory unity.
This is why it is crucial in Kant’s Beweisgrund conception of the GARP that
its existence and possibility lack grounds; if they had grounds, then reason’s
regressive search for conditions could not stop there. The absolute necessity of
Beweisgrund’s GARP is ultimately ‘outward looking’: the GARP is necessary
because it grounds all real possibility, and, if it were canceled, nothing would
be really possible. This is perfectly suited to Kant’s conception of reason and
the unconditioned in the Dialectic, because the GARP there plays the role of the

74
See Fig. 5.2 and the discussion surrounding it.
75
See Figs. 5.3 and 5.4 and the discussion surrounding them.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE UNITY OF KANT ’ S MODAL METAPHYSICS 

stopping point of reason: once reason reaches it, its regressive search for the
conditions of possibility can cease.
What is more, the picture of the ground of all (noumenally) really possible
predicates at which reason arrives in the Transcendental Ideal is precisely Kant’s
picture of the GARP in Beweisgrund:

The derivation of all other possibility from this original being, strictly speaking, also cannot be
regarded as a limitation of its highest reality and as a division, as it were; for them the original
being would be regarded as a mere aggregate of derivative beings, which, according to the
above, is impossible, even though we represented it in such a way at the beginning in our first
crude outline. Rather, the highest reality would ground the possibility of all thing as a ground
and not as a sum total; and the manifoldness of the former rests not on the limitation of the
original being itself, but on its complete consequences; to which our whole sensibility,
including all reality in appearance, would then belong, which cannot belong to the idea of a
highest being as an ingredient. (A579/B607)

This is just the point, which I discussed extensively in Chapter 4, that (contra
Chignell 2009a) the GARP does not ground really possible predicates by instantiating
all unlimited realities; some really possible predicates are “consequences” of its
fundamental predicates but are not limitations of them. As Kant points out earlier
in the Ideal, not all unlimited realities are equally fundamental, and some are
inconsistent with one another: the idea of a GARP “excludes multiplicity of predi-
cates, which, as derived from one another, are already given [in the fundamental
predicates—NS] or cannot coexist with one another” (A573–4/B601–2). This was
precisely Kant’s anti-logicist point in Beweisgrund: some primitive (hence unlimited
and unnegated) realities are inconsistent with one another, so no being can possess
all of them.
Let us return to the passage with which I began this chapter, where Kant says that
in Beweisgrund “it was shown that of all possible proofs, the one which affords us the
most satisfaction is the argument that if we remove an original being, we at the same
time remove the substratum of the possibility of all things” (Pöl.RT Ak. 28: 1034).
The most rationally satisfactory explanation (the explanation that best satisfies LPR’s
regressive search for conditions) of why anything is (noumenally) really possible is
that there is an absolutely necessary being that grounds these possibilities. But, Kant
now claims, this should not have been mistaken for a proof that there is an absolutely
necessary being: “this proof can in no way be refuted, because it has its ground in the
nature of human reason. For my reason makes it absolutely necessary for me to
assume a being which is the ground of everything possible, because otherwise I would
be unable to know what in general the possibility of something consists in” (Ak. 28:
1034). I take this to mean that:

(a*) the LPR directs us to find an explanation (condition) for the (noumenal)
real possibility of predicates (the conditioned).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE UNITY OF KANT ’ S MODAL METAPHYSICS

We have the concept of noumenal real possibility, and noumenal real possibilities
are subject to reason’s search for conditions. Furthermore:
(b*) the only conception of the ground of (noumenal) real possibility that would
satisfy the LPR is that of an absolutely necessary being, one whose non-existence
would cancel all (noumenal) really possible predicates.
Kant asserts that his own pre-Critical conception of God as the ground of all real
possibility is the only way to satisfy reason’s demand for conditions of possibility
(“otherwise I would be unable to know what in general the possibility of something
consists in”).
Kant’s pre-Critical view of God as the absolutely necessary ground of all real
possibilities satisfies two other important requirements on theoretical belief: the
availability of intersubjectively valid reasons grounded in the nature of theoretical
reason itself, and our inability to know such a being to be impossible. Since the
putative grounds for theoretical belief in God are a necessary means to a necessary
theoretical end of reason ((a*) and (b*)), they are grounds that are intersubjectively
valid for all (finite) discursive rational intellects. Since this is a concept of a nou-
menon and it is logically consistent, we can never know (at least on theoretical
grounds) that it is impossible. This entails that if a human epistemic agent (i) judges
that [hält für wahr] there is such an absolutely necessary ground of (noumenal) real
possibility, (ii) bases this judgment on (a*) and (b*), (iii) judges further that (a*) and
(b*) do not (noumenally) necessarily entail that there is such a being, then that agent
has theoretical belief in the metaphysical theory of modality first offered in Beweis-
grund. Any human epistemic agent can come to have rationally grounded theoretical
belief in an absolutely necessary being.
Earlier in the same passage Kant says of his Beweisgrund proof that “even this
proof is not apodictically certain; for it cannot establish the objective necessity of an
original being, but establishes only the subjective necessity of assuming such a being”
(Pöl.RT, Ak. 28:1034). Read in the light of Kant’s theory of the necessary ends of
theoretical reason in the Transcendental Dialectic, this means that the original 1763
proof of a GARP was the product of transcendental illusion: it was illegitimately
assumed that if the conditioned (a really possible predicate) is given then so is its
unconditioned condition (GARP). Kant’s Critical attitude towards the Beweisgrund
proof is that it mistook the intention for the deed: although the postulate that God is
the ground of all real possibilities is the most rationally satisfactory explanation of
real possibility, and this is sufficient to warrant theoretical belief that there is such a
being, this does not entail that there is such a being. Consequently, the epistemic
ground on which we base our belief in the GARP does not necessitate its existence, so
this is not a case of knowledge. Furthermore, because we cannot so much as cognize
[erkennen] the real (noumenal) possibility of such a being, it cannot in principle be
an object of knowledge [Wissen] for us.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

10
The Antinomy of Kant’s
Modal Metaphysics

10.1. Introduction
The title of this chapter does not refer, as some readers might expect, to the fourth of
the Antinomial Conflicts of Pure Reason, which concerns the conflict between a
putative proof that there is a necessary being that grounds all contingent beings
(Thesis) and a putative proof that there can be no such necessary being either in
space and time or outside them (Antithesis). Like each of the Antinomial conflicts,
the Fourth Antinomy is resolved in the Transcendental Dialectic: there can be no
necessary being in space and time, but there may be (for all we know) a necessary
being outside space and time, among the noumena. The title of this chapter refers
instead to what I take to be an unresolved antinomy in Kant’s modal theory, between,
one the one hand, his theory of freedom, and, on the other hand, his famous claim in
}}76 and 77 of the Critique of Judgment (CJ) that an intuitive intellect does not
cognize modal properties in its objects.
As I discussed in Chapter 7, Kant’s theory of freedom is intended to reconcile the fact
that I am bound by the moral law with the fact that all of my actions are empirically-
causally necessary (they are the consequences of deterministic laws, given the past).
That I am bound by the moral law means that, although I actually commit an
impermissible act, I ought not to have done so, which in turns entails that I could
have omitted that act (ought implies can). But, considered merely as a phenomenon, it is
not (empirically-causally) possible that I do otherwise. However, I am also a noumenon,
which means that I have a nature independently of how I cognize myself (and how other
discursive intellects cognize me) theoretically, a nature that appears to me (and to
others) in space and time. This noumenal nature includes a will, and this will has the
power to determine itself otherwise than it actually does (to adopt a maxim of action
other than its actual one). For any act I commit, I possess the power, as a thing in itself,
to have so determined my noumenal will that I would not have committed the action if
I had so determined my will. As Kant writes in the second Critique:
So considered [as a noumenon] a rational being can now rightly say of every unlawful action he
performed that he could have omitted it even though as appearance it is sufficiently deter-
mined in the past, and so far, is inevitably necessary [ . . . ] (CPrR, Ak. 5: 98)
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

This is not a separable or minor commitment of Kant’s theory of free will; Kant’s
proof of the compatibility of the empirical necessity of my actions with freedom and
thereby with the bindingness of the moral law rests on attributing a modal property to
my noumenal will: as a noumenon, I could have determined my will otherwise, so my
actual acts are contingent.
However, this means that Kant’s theory of freedom stands in tension with his
claim, articulated most forcefully in }76 of the CJ, that an intuitive intellect would not
represent objects as merely possible or as contingent:
For an intellect [Verstand] to which this distinction [between intuiting and thinking] did not
apply, all objects that I cognize would be (exist), and the possibility of some that did not exist,
i.e. their contingency if they did exist, as well as the necessity that is to be distinguished from
that, would not enter into the representation of such a being at all. (CJ, Ak. 5: 403)

An intuitive intellect is one whose faculty of intuition is spontaneous rather than

receptive (like ours); consequently, the distinction between passive intuition and
spontaneous thought, which holds for us, would not hold for such a mind. Kant is
claiming in this passage that a being with an intuitive intellect would not represent
some objects as merely possible (those it had conceived but not intuited), some as
contingent (those it intuited but which it could conceive not being present) and some
as necessary. In what follows, I will refer to this thesis as the thesis that an intuitive
intellect does not cognize modal properties in its objects: it does cognize its objects as
being possible, contingent, etc.
These two claims—that noumena have modal properties, and that an intuitive
intellect does not cognize modal properties—stand in tension because God has an
intuitive intellect and God is omniscient; God cognizes not merely the truth about
noumena but the whole truth about them. As Kant says in the Pölitz lectures on
rational theology, “God cognizes all things as they are in themselves immediately and
a priori through an intuition of the understanding” (Pöl.RT, Ak. 28: 1052). He is
therefore committed to an inconsistent set of claims:
(P0) If there is a God, he has an intuitive intellect.1
(P1) Some noumena have modal properties. [Consequence of the theory of freedom.]
(P2) No intuitive intellect cognizes its object(s) as having modal properties. [CJ }76.]
(P3) Noumena have modal properties if, and only if, if there is a God, God would
cognize those properties.2 [Omniscience of God, see previous discussion.]
(P1) is inconsistent with the ‘noumenal amodalist’ consequence of (P2) and (P3):
noumena lack modal properties. Although I argued at length in Chapter 9 that Kant

1
Pöl.RT (Ak. 28: 996, 1043, 1052), Danz.RT (Ak. 28 :1259, 1267), and Volck.RT (Ak. 28: 1165).
2
My presentation of this problem is indebted to Kohl (2015). Whereas Kohl tries to use the intuitive
intellect to refute what he calls the ‘Leibnizian’ reading of Kant (that noumena have categorial properties),
I try to reconcile the intuitive intellect and the ‘Leibnizian’ reading.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

is committed, not only for practical, but for theoretical reasons, to belief [Glaube] in
the existence of God, I have formulated these premises as conditional on the
existence of God in order to highlight that this problem is independent of Kant’s
own positive commitment to the existence of God; even if there is no God the
problem remains. Nor can it be dismissed by pointing out (correctly) that God is
merely an object of (theoretical and practical) belief rather than knowledge; regardless
of the epistemic status of these inconsistent claims, as long as Kant is committed to
each of them, there is an inconsistency in his view. Kant’s doctrine that the intuitive
intellect does not represent modal properties appears to be incompatible with what,
according to Kant himself, is one of the two foundations of the Critical system:
freedom of the will. The modal properties in question are modal properties of
noumenal wills, e.g., contingently subordinating the moral law to self-interest, pos-
sibly doing otherwise, etc.
Although it does not take the canonical Kantian form of two incompatible claims
(Thesis and Antithesis), this potential inconsistency deserves to be called the (or least
an) antinomy of Kant’s modal metaphysics, for, unlike the Fourth Antinomy, it is
neither recognized nor resolved by Kant himself. What is more, while it may have
been recognized by readers of Kant, to my knowledge no sustained attempt has been
made to resolve it within Kant’s system.3 In the case of post-Kantian figures like
Fichte, Schelling, and Hegel, who thought the intuitive intellect was one of Kant’s
most important ideas, this is because they were less interested in reconciling the
various aspects of his system (which they tended to see as shot through with
inconsistencies, at least if taken at face value) than with developing an adequate
form of idealism on the basis of key Kantian insights. If they recognized this
inconsistency at all, Schelling and (at least the early) Hegel would regard it as a
sign that the intuitive intellect is incompatible with certain limitations in Kant’s own
system (e.g., the sharp distinction between inclination and duty, and hence between
necessity and freedom) and embrace what they saw as the necessitarian (indeed,
Spinozistic) consequences of CJ }76; 4 nor would Fichte, who understood intellectual
intuition very differently than Schelling or Hegel, necessarily see this ‘modal anti-
nomy’ as a problem in need of a solution.5

3
There is an extensive literature on the Kantian notion of intuitive intellect and its fate in post-Kantian
German idealism; see esp. Gram (1981); Tüschling (1990) and (1992); Allison (2000); Longuenesse (2000);
Westphal (2000); Beiser (2002), 294–301, 580–2; Förster (2011); and Leech (2014).
4
Cf. the discussion of intuitive intellect and CJ, }76 in Faith and Knowledge (in H.Werke, 2: 316–7,
325–7). Tüschling (1992) argues that CJ }76 is already Spinozistic and proto-Schellingian; Allison (2000)
rebuts that interpretation. On Schelling's and the young Hegel’s embrace of what they saw as the Spinozistic
consequences of intellectual intuition, see Beiser (2002), 580–2. Longuenesse (2000) points out correctly,
though, that CJ }76, taken at face value, does not entail noumenal necessitarianism, but noumenal
amodalism: noumena have no modal properties whatsoever (Longuenesse (2000), 279 n. 19).
5
For an analysis of Fichte’s various notions of ‘intellectual intuition’ see Breazeale (1998) and Beiser
(2002), 294–9.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

In this chapter I explore the path not taken by these German Idealists, and attempt
to resolve this apparent inconsistency in Kant’s theory of the intuitive intellect
(represented by (P2) and (P3)) and the modal commitments of his theory of freedom
(represented by (P1)) using the resources of Kant’s own system. I will take premise
(P0) as a fixed point in what follows and consider the inconsistency among (P1),
(P2), and (P3) given Kant’s commitment to (P0). In section 3 I explore several ways
one might resolve this inconsistency and argue that none of them succeed. In section
4 I present my own resolution of the problem; in sections 5 and 6 I expand upon that
solution and explain how Kant’s theory of freedom can be made consistent with
‘noumenal amodalism.’ In section 7 I use CJ }76 to clarify the relations among three
notions that have been important throughout this book: existence, existence* (causal
efficacy), and actuality. I conclude in section 8 by showing how my solution to the
‘modal antinomy’ helps clarify, retrospectively, central issues in Kant’s modal meta-
physics in Beweisgrund. First, though, in section 2, I briefly explain what an ‘intuitive
intellect’ would be.

10.2. Intuitive Intellect, Intellectual Intuition

Although the concept of an anschauende Verstand (less frequently, ‘intuitive Verstand’)
is one of the most historically influential concepts in the whole Critical philosophy
(due to its reception in post-Kantian German philosophy), it is only explicitly
discussed in a few passages in the entirety of Kant’s published writings.6 Although
I have elsewhere followed standard practice and translated ‘Verstand ’ as ‘under-
standing’, in this book I am going to translate ‘intuitiver Verstand ’ as ‘intuitive
intellect’ to bring out the fact that it refers to a very different kind of mind than our
discursive understanding.
Kant introduces the idea of intuitive intellect by contrasting it with our discursive
form of intellect. Objects are given to a discursive intellect like ours passively through
causal affection of our sensible faculty (hence, our intuition is said to be ‘sensible’),
and we then must think spontaneously about those objects using general (discursive)
representations, concepts. The idea of an intuitive intellect is the idea of a kind of
mind whose understanding (spontaneous faculty) would give it objects, that is, a
mind that would spontaneously intuit objects.7 A representation is spontaneous if the

6
The main texts for the concept of an intuitive intellect are CJ }}76–7 (Ak. 5:401–10), various remarks
in the B Deduction about “an understanding that would intuit” (B135, B138–9, B145, B149, B159); see also
the first Introduction to CJ (Ak. 20: 227), Prol., }34 (Ak. 4: 316–17), Disc. (Ak. 8: 216), Corr. (Ak. 10: 130),
and ID, }10 (Ak. 2: 396–7); in the unpublished writings, see Refl. 1832 (Ak. 16: 131), 6048 (Ak. 18: 433),
MK1 (Ak. 28: 1256), ML1 (Ak. 28: 241, 330), LP (Ak. 24: 361), MM (Ak. 29: 759), MV (Ak. 28: 372), MK2
(Ak. 28: 782), and MK3 (Ak. 29: 972), in addition to the theology lectures cited above. Förster (2011)
provides a comprehensive survey of German philosophy from Kant to Hegel, with a focus on intuitive
intellect; see subsequent discussion, though, for some reservations about his reading of CJ }76. See also
Westphal (2000) for the reception of this Kantian concept in post-Kantian German philosophy.
7
CJ, Ak. 5: 406; B135.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

ground of that representation is contained within the mind whose representation it

is, so a spontaneously intuiting intellect is one that is the ground of its own intuitions.
We are the ground of the possibility of the objects we intuit because our minds
contain the forms (space and time) in which any object can be given to us. An
intuitive intellect is not merely the ground of the possibility of its objects but of its
objects themselves (their existence, absolute positing); an intuitive intellect ‘posits its
objects absolutely’ in intuiting them. This is why Kant characterizes an intuitive
intellect as “an intellect through whose representation [durch dessen Vorstellung] the
objects of this representation thereby [zugleich] exist [existieren]” (B138–9).8 Since
an intuition is a singular and immediate relation to an object, it may not be fully
coherent to say that an intuitive intellect’s intuition of an object grounds the existence
of that object, since the object must exist (absolute positing) to be the object of an
intuition in the first place.9 Instead, I think it is more coherent to say that: the
intuition of an object by an intuitive intellect just is its being absolutely posited. For
the objects of an intuitive intellect, their existing (being absolutely posited) just is
their being intuited by such an intellect. This is why Kant also describes an intuitive
intellect as possessing ‘intellectual intuition’ [intellektuelle Anschauung]: it intuits its
objects spontaneously (intellectually) rather than passively (sensibly).10,11
In this book I have focused on two technical meanings Kant associates with the
term ‘cognition’ [Erkenntnis]: (1) concepts we can know to be really possibly
instantiated, and (2) thinking a given object under a concept. But neither of these

8
In this passage and elsewhere in the B Deduction Kant is mainly focusing on an intuitive intellect’s
representation of itself, but the same point applies, mutatis mutandis, to an intuitive intellect’s represen-
tations of objects distinct from itself: its intuition of them is their existence. Cf. Corr. (Ak. 10: 30), ID }10
(Ak. 2: 397).
9
The idea is this: x cannot exist in virtue of standing in relation R to some other object, for the fact that
it stands in that relation is partly explained by the fact that x exists (otherwise, it would not be ‘available’ to
stand in that relation). So it is more perspicuous to say: x existing just is x being an object of an intuitive
intellect.
10
In fact, ‘intellektuelle Anschauung’ occurs more frequently in Kant’s writings; the main discussion is
in the ‘Phenomena and noumena’ section (A252–6, B307–313; cf. Kant’s handwritten marginal notes at
A248, E CXXX and CXXXI (Ak. 23: 36)). See also Bxl, B68, A279/B335, Prol. (Ak. 4: 375 n.), CPrR (Ak. 5:
31, 99, 123), CJ (Ak. 5: 409), Disc. (Ak. 8: 219, 220), Rel. (Ak. 6: 67), Prog. (Ak. 20: 267), Tone (Ak. 8: 389),
Corr. (Ak. 10: 123, 11: 51), Refl. 4207 (Ak. 17: 456), Refl. 4677 (Ak. 17: 658), Refl. 5637 (Ak. 18: 274–5), Refl.
5653 (Ak. 18: 306), Refl. 6050 (Ak. 18: 434), MK1 (Ak. 28: 1520), ML1 (Ak. 28: 179, 206, 241), MM (Ak. 29:
800, 880), MV (Ak. 28: 371), MD (Ak. 28: 653), MK3 (Ak. 29: 972, 974), and VE (Ak. 29: 14–15). In two pre-
Critical Reflexionen, Refl. 4228 (Ak. 17: 467) and 4336 (Ak. 17: 509–10), Kant claims that I know my
freedom through intellectual intuition of myself.
11
The thesis that ‘intellectual intuition’ and ‘intuitive intellect’ refer to well-defined concepts, much less
the same thing, is controversial. Gram (1981) argued that Kant has several non-equivalent notions of
intellectual intuition. Förster (2011), 150–60 argues that the concept of an intellectual intuition and the
concept of an intuitive intellect are distinct. I concur with the argument of Leech (2014), though, that while
these are different concepts (one is a concept of the kind of intuition a mind might have, the other is a
concept of a kind of intellect) they are ‘two sides of the coin,’ or, more precisely, a being with intellectual
intuition is an intuitive intellect, and an intuitive intellect has intellectual intuition. They are concepts,
respectively, of a part of a kind of mind (intellectual intuition), and that kind of mind itself (intuitive
intellect), which necessarily come as a package.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

can be what Kant means when he talks of the cognitions of the intuitive intellect,
because (1) the intuitive intellect does not represent its objects modally, and (2) it
does not use general representations (concepts) to think about given objects.12 So
Kant must mean cognition in some even more general sense when discussing the
cognition of the intuitive intellect, which I take to mean, roughly: representing an
object correctly as the object it is.13

10.3. Some (Unsuccessful) Attempts

to Resolve the Antinomy
In this section I argue that several tempting strategies for resolving the inconsistent
triad of (P1)–(P3) do not succeed.
1. The first attempted solution to the problem arises from Kant’s distinction
between ‘negative’ and ‘positive’ noumena. In the A edition Kant distinguishes
between phenomena and noumena and defines the latter as objects of a non-sensible
or ‘intellectual’ intuition (“coram intuitu intellectuali”—A249). In the B edition he
makes a further distinction between negative and positive noumena, defined, respect-
ively, as objects that are not objects of a sensible intuition, and objects of a non-
sensible intuition (noumena simpliciter, in the terminology of the A edition).14
Notice, though, that <negative noumenon> is, in principle, a more general concept
than <positive noumenon>: if something is an object of a non-sensible (intellectual)
intuition then it is not the object of a sensible intuition, but the converse does not
hold as a matter of definition alone. The concepts <negative noumenon> and
<positive noumenon> are not logical contradictories, but stand in a relation of logical
subordination. I will refer to negative noumena that are not positive noumena as
merely negative noumena (see Fig. 10.1).
With these distinctions in place, we can see that ‘noumena’ in (P2) and (P3) means
positive noumena (objects of God’s non-sensible intuition) but the inconsistency
arises only if (P1) is also true of positive noumena as well. Our grounds for holding
(P1) is that our noumenal wills have modal properties. But why should we assume
that our noumenal wills are positive noumena? If our noumenal wills are ‘merely
negative noumena’ then the contradiction is resolved. However, the concept of God
is the concept of an omniscient intuitive intellect, an intellect that cognizes all

12
CJ, Ak. 5: 403, 406. Nor can the intuitive intellect have Wissen since Wissen is a mode of Fürwahr-
halten, which is a subject’s relation to judgment, and judgments are discursive representations of objects.
13
It is, of course, hard to determine Kant’s most general concept of cognition for he almost always discusses
specifically discursive cognition; my hypothesis (cognition in general = objective representation of an object as
the object it is) receives some indirect support from Bix-x, A58/B83, B137, B166, A258/B314, A320/B376, and
JL (Ak. 9: 51). I am drawing here on some ideas of Smit (2000) and Schafer (forthcoming). For more, see the
supplementary article “Intuitive Intellect and Cognition” on my website (see Notes on the Text).
14
B307.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

Negative noumena
Fig. 10.1. Negative and positive ~SI
noumena

Merely negative Positive noumena

noumena NSI, ~SI
~NSI, ~SI

SI = object of a sensible intuition NSI = object of a non-

sensible intuition

noumena.15 It follows that, if there is a God, then all noumena are positive noumena,
and, thus, they cannot have any properties that positive noumena lack, including
modal properties. The proposed resolution of the antinomy does not succeed, for it
makes freedom incompatible with the existence of God, but we know that both are
commitments of Kant’s view.
In a similar vein, though, one might point out that the grounds for holding (P3) is
that it is an instance of a more general principle:
(P3*) For any property F, noumena have property F, if, and only if, if there is a
God, God would cognize them as having property F.
However, (P3*) has a counterexample. If there is a God then noumena have the
property being cognized by God and God cognizes them as having that property (he is
aware that he cognizes them). If we substitute being cognized by God for F in the
right-hand side of (P3*), (P3*) entails all by itself that noumena are cognized by God,
hence that there is a God. Intuitively, though, (P3*) should not entail that God exists!
The point behind this objection is that the presence of God might alter the properties
(at least the relational ones) of noumena and so we should restrict (P3*) to the ‘God-
independent’ properties of noumena. However, having modal properties may be a
God-dependent property: if there is a God then all noumena are positive noumena, as
I argued in the last paragraph, and by (P2), it follows that none of them have modal
properties. If (P3*) is restricted to God-dependent properties (properties noumena
have whether or not they are positive noumena) then (P3) is not an instance of it, so
the original contradiction is resolved. However, this objection to (P3) only succeeds if
there is no intuitive intellect. But Kant thinks that we are rationally required by
practical reason and (I argued in Chapter 9) by theoretical reason to believe (in
Kant’s technical sense of Glaube) that there is a God, and he thinks that if there is a
God, he has an intuitive intellect (premise (P0)), so this way of resolving the
antinomy is not available to him.16

15
Pöl.RT, Ak. 28: 1052.
16
A more ambitious response to this objection is possible. See the unpublished paper “Kant’s Concept
of an Object” on my website (see Notes on the Text).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

2. Another way one might attempt to resolve the triad is pointing out that (P1) is
cognized on practical grounds while (P2) (and perhaps (P3)) represents a mere
postulate (rational belief) of theoretical reason. Consequently, the conflict is resolved:
from a theoretical standpoint noumena lack modal properties, while from a practical
one we have to attribute modal properties to them (e.g., the property could have done
otherwise).
This solution involves relativizing (P1)–(P3) to different ‘standpoints’ or ‘perspec-
tives.’ As such, it depends upon the ‘epistemic’ or ‘methodological’ interpretation of
Kant’s idealism pioneered by Graham Bird, Gerold Prauss, and Henry Allison.17 In
this book I have attempted to refrain from directly engaging with the debate about
the meaning of Kant’s transcendental idealism, but it may be necessary, at this point,
to briefly discuss the relation between the ‘modal antinomy,’ this proposal for
resolving it, and the epistemic interpretation of transcendental idealism. I will
focus on Allison, whose version of the ‘methodological’ reading has been most
extensively discussed by Anglo-American scholars.
The core idea of Allison’s interpretation of transcendental idealism is to relativize
various claims we might make about objects to a standpoint and to the conditions on
objects appropriate to that standpoint. There is a theoretical standpoint, in which we
consider objects (including ourselves) as subject to the conditions under which
knowledge of them is possible, and a practical standpoint, in which we consider
objects (most importantly, ourselves and our actions) as subject to the conditions
under which judgments about morally right action is possible. My problem will not
arise, on Allison’s reading, because it is really a problem about the competition
between standpoints—the moral standpoint (in which (P1) is made) and what we
might call the standpoint of speculative theology (in which (P2) and (P3) are
made)—and Kant, according to Allison, can dismiss such conflicts because there is
no way things are independent of a standpoint, no ‘meta-standpoint’ from which the
different standpoints can be judged. The Allisonian answer to my problem is to say
that in considering ourselves as moral agents we must consider ourselves as possibly
having done otherwise (P1) and when doing speculative theology we have to consider
the objects of divine cognition as lacking modal properties ((P2) and (P3)), and there
is no coherent question about which one is correct.18
I do not have the space here to argue against Allison’s interpretation. I will simply
point out that Allison’s view is best equipped to resolve conflicts between the
practical standpoint and the empirical-theoretical standpoint, the standpoint we
adopt when we cognize empirically given objects using the faculty of understand-
ing.19 But (P1)–(P3) represent a conflict between the practical standpoint and what

17
The loci classici of this view are Bird (1962), Prauss (1974), and Allison (1983) and (2004).
18
See Allison (2004), 20–49.
19
This comes out clearly in the essays on the freedom of the will collected in Allison (2012).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

I referred to above as the ‘speculative-theological’ standpoint, the standpoint we

adopt when we merely think about how God’s intuitive intellect would cognize our
noumenal wills. Allison has not explained how the latter standpoint functions, and
how it is compatible with, but different from, the practical standpoint. It is not clear,
then, that Allison’s reading can be extended to make sense of (P2) and (P3).20 What
is more, even if the Allisonian reading can be so extended, the resulting view will
inherit all of the problems of Allison’s interpretation of transcendental idealism
generally.21 While none of this entails that Allison’s view is false, or the Allisonian
response to the modal antinomy cannot succeed, it provides significant motivation
for a solution to the ‘modal antinomy’ that does not depend upon the epistemic
reading of transcendental idealism.
3. A third alternative would be to draw a distinction, within the set of modal
properties, between theoretical modal properties and practical modal properties. The
idea would be that the modal categories as they are introduced in the CPR have a
certain content or meaning and that content is what Kant denies applies to noumena
in CJ }76, but the modal categories acquire a new practical content from their relation
to the moral law and the will. Kant is not denying that content applies to noumena.
In the Critique of Practical Reason (CPrR) the modal categories do acquire a kind of
practical content in the guise of what Kant calls ‘categories of freedom’: the category
<possibility–impossibility> becomes “the permitted and the forbidden,” the category
<existence–nonexistence> becomes “duty and what is contrary to duty,” and <necessity–
contingency> becomes “perfect and imperfect duty.”22 The modal categories can now be
used, not to express the relation of sensibly given objects to our cognitive faculties, but
the relation of maxims of action to the moral law, which, for sensible beings like us, is
expressed as duty (what we ought to do). The category <possibility>, which can be used
to express the fact that an object is compatible with the sensible and intellectual forms
of our experience, can also be used, in the guise of the concept <permissible> (a ‘category
of freedom’), to express the fact that a maxim is compatible with duty.23

20
This is a point on which Allison’s reformulation of his view in (2004) is especially vulnerable. Allison
there identifies transcendental idealism as the rejection of the “theocentric” view that there is a “way things
are,” the way they would be disclosed to God’s mind (if there is a God), and that human cognition is
defective because it does not fully grasp the “way things are.” But Kant endorses precisely such a
theocentric model in Pöl.RT (see esp. Ak. 28: 1053–4). The conflict I am discussing (P1–P3) is really a
conflict between the practical standpoint and that ‘divine’ standpoint. Cf. the discussion of the intuitive
intellect in Allison (1983), 340 n. 2 and (2000), as well as the discussion of a related problem in Brewer and
Watkins (2012).
21
See Ameriks (1982a); Guyer (1987), 336–8; Robinson (1994); Langton (1998), 8–12; Van Cleve
(1999), 4, 148–9; and Stang (2014b) and (forthcoming). See also Allison’s response to his critics in (2004),
6–11, 42–9.
22
Ak. 5: 65–6. For critical discussion see Beck (1960), 144–53 and Bader (2009).
23
There is an important parallel between Kant’s denial that an intuitive intellect would cognize modal
properties in its objects and his denial that duty (and hence the modal categories of freedom, e.g., the
permissible and the forbidden) applies to a holy will. The context in which Kant introduces (P2) in CJ }76 is
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

But the modal categories of freedom are not the modal properties attributed to
noumena in (P1);24 (P1) concerns alethic rather than deontic modality. To attribute a
modal property to my noumenal will is not thereby to claim that any of my actions
are permissible or impermissible; it is to make a claim about the necessary condition
of that very permissibility or impermissibility: I could have done otherwise. In Kant’s
paradigm case, I do what is impermissible (I subordinate the moral law to my self-
interest) but I could have done otherwise, where ‘could have’ is alethic not deontic. It
is not that my doing otherwise was permissible (it was obligatory, in fact); it is that
because I could have (alethic possibility) done otherwise that the moral law applies to
me and makes it the case that I ought to (deontic necessity) have done otherwise. It is
true that the modal categories, indeed all of the categories, acquire a new kind of
practical content as “modi of the concept of causality of freedom” in the second
Critique, but Kant’s theory of freedom is committed to applying the modal categories
with their original non-practical non-deontic alethic content to our noumenal wills
and that appears inconsistent with CJ }76.25
4. Finally, it might be argued that this represents a development in Kant’s view
from the first edition of the CPR in 1781 (the first published work that contains the
transcendental idealist theory of freedom)26 to the publication of the third Critique in
1790. On this narrative, Kant originally wanted to attribute real modal properties to
noumena and, in particular, to distinguish the (empirical-causal) necessitarianism
obtaining at the phenomenal level from the contingency of the free acts of rational
agents (noumenal wills) that obtains at the noumenal level, but was eventually
pushed towards a more unified holistic and necessitarian conception of noumenal
reality under the pressure of the idea of a divine intuitive intellect. Some might even
see this as a positive development in Kant’s thought, away from his earlier concern
with eighteenth-century German debates about free will and determinism, and
towards a more ‘speculative’ or ‘absolute idealist’ view on which the appearance of
contingency is merely an artifact of our (discursive) cognitive architecture, which
would fall away if we had intellectual intuition of noumenal reality.27

his claim that in three different cases we must recognize that our need to represent objects in a certain way
is a subjective consequence of our kind of mind, which we express by thinking of an alternate kind of mind
that would not so represent its objects: modality (intuitive intellect), ought (holy will), and purposiveness
(intellectual intuition). I do not have space to explore this parallel further here.
24
More precisely, (P1)’s denial that noumena have modal properties (fall under modal concepts) is not
the denial that they fall under modal categories of freedom.
25
CPrR, Ak. 5: 97–8.
26
By which I mean the combination of phenomenal determinism and noumenal freedom. Another
component of Kant’s mature Critical view, that we can only know our freedom through awareness of
ourselves as bound by the moral law, may not have been in place by 1781. Some scholars hold that Kant still
entertained the possibility of a theoretical proof of freedom in 1781 and perhaps even in 1785 (GMM); see
Ameriks (1981) and (1982b), 189–209.
27
This is essentially the view of Tüschling (1990) and (1992). According to Beiser (2002), 580–2, it also
represents the view of Schelling and Hegel at the turn of the nineteenth century.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

As attractive as such a developmental narrative may be to some, it does not fit well

with the texts, because Kant’s doctrine that the intuitive intellect would cognize
objects without modal properties is already present in nuce in the CPR, and based
on reasons similar to those in the CJ. I will hold off on a detailed analysis of Kant’s
reasons for holding (P2) in the CJ until the next section, but very broadly, his reasons
are these. Discursive intellects like ours are given objects and must think them under
concepts. This gives rise to a distinction between what is actual (what is given in
intuition) and what is possible (what agrees with the conceptual and intuitional form
that the actually given objects fill out or instantiate). So, Kant reasons, an intellect
that immediately intuited the whole nature of its object in virtue of being the ground
of the existence of that object would not represent a distinction between what is
actually the case and what might be the case.
The roots of this doctrine are already present in Kant’s claim in the CPR that “the
categories of modality have this special feature: they do not augment the concept to
which they are ascribed in the least, as a determination of the object, but rather
express only the relation to the faculty of cognition” (A219/B266). If the modal
categories of the intuitive intellect (assuming it has them) express the relation of its
objects to its cognitive faculty then there are grounds to suspect that these modal
categories will at least be very different from ours, because the cognitive faculty to
which they are related is quite different. Consider that, in our case, the category of
actuality is closely connected to intuition28 (which gives us the objects there actually
are) while the category of possibility is tied to concepts (of objects that can be given in
intuition).29 From this alone, we should expect that an intellect in which there is no
distinction between concepts and intuitions would represent no difference between
what is actual and what is possible. The discussion of the intuitive intellect and
modality in }}76 and 77 of the CJ draws out a consequence of doctrines already
present in the CPR; it does not represent a radical new development in Kant’s
thinking.

10.4. Resolving the Antinomy

In this section I present my solution to the inconsistent triad of (P1)–(P3). Recall
Kant’s reasons for denying that the intuitive intellect cognizes modally:
if our understanding were intuitive, it would have no objects except what is actual. Concepts
(which pertain merely to the possibility of an object) and sensible intuitions (which merely give
us something, without thereby allowing us to cognize it as an object) would both disappear.
Now, however, all of our distinction between the merely possible and the actual rests on the
fact that the former signifies only the positing of the representation of a thing with respect to

28
A218/B265–6.
29
A599/B627. Another way of seeing this connection: actuality is singular (there is only one actual
world), while possibility is general (there are many ways the world could be).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

our concept and, in general, our faculty for thinking, while the latter signifies the positing of the
thing in itself (apart from this concept). Thus the distinction of possible from actual things is
one that is merely subjectively valid for the human understanding, since we can always have
something in our thoughts although it does not exist, or represent something as given even
though we do not have any concept of it. (CJ, Ak. 5: 402)

Kant claims that we represent a modal distinction in objects because we have two
different sources of cognition: intuitions, by which objects are given to us, and
concepts, by which we think about objects. This allows us to represent a distinction
between possibility and actuality: if a concept agrees with the conditions under which
objects can be given to us, then it is the concept of a possible object, whether or not it
has any instances; if, furthermore, the concept is instantiated by some object given to
us in intuition, then the concept is the concept of an actual object. Since the
representations of the intuitive intellect immediately present objects, and the ‘abso-
lute positing’ of these objects is either identical with or grounded in their presentation
to the intuitive intellect (see section 2), the intuitive intellect will not have unin-
stantiated general representations (concepts) of merely possible objects (objects it
could spontaneously intuit). Since the representations of the intuitive intellect spon-
taneously generate their objects, the intuitive intellect will not represent a difference
between the actual and the merely possible.30
The principle that the concept <actual> relates an object to our receptive faculty of
intuition, while <possible> relates it to our spontaneous faculty of understanding,
applies to these concepts as they are used by discursive intellects like ours. That the
intuitive intellect would not represent a difference between objects consisting in their
relation to receptivity and spontaneity might mean that the intuitive intellect would
not represent the modal difference between objects that we do. It does not by itself
entail that the intuitive intellect would not represent any modal difference at all
between objects. When Kant claims that necessity and contingency “would not enter
at all into the representations of such a being” (CJ, Ak. 5: 403), he might mean
something more specific by possibility and necessity than necessity and contingency
überhaupt; he might mean that the intuitive intellect would not represent the modal
properties of objects in the specific way we do.
This idea of a less specific way of representing possibility and actuality should put
us in mind of the threefold distinction among the unschematized categories (the

30
While I agree with the critique of Gram and Förster in Leech (2014), I think that her account of why a
discursive intellect does, and an intuitive intellect does not, represent its objects modally, conflates
epistemic and ‘metaphysical’ (for lack of a better word) modality. According to Leech, we use modal
categories because we face an epistemic problem: some of our concepts are not instantiated by objects. The
intuitive intellect does not face this problem, because its objects are ‘guaranteed’ to exist (Leech 2014, 347).
But this is incorrect about the role of modal concepts in Kant’s theory; even if we had a ‘guarantee’ that our
concepts are instantiated, we would still need the modal concepts to think about the form of our experience,
hence to think about which objects and which of their properties are necessary in virtue of the form of our
intellect.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

categories as pure concepts of the understanding of all objects überhaupt), the

categories as concepts of objects of a discursive intellect in general, and the (tem-
porally) schematized categories (the categories insofar as they can be cognitively
applied to objects of our specifically spatiotemporal intuition).31 The modal categor-
ies do not, as we have seen, add to the content of the concept of an object, since they
are not, like the categories of the other three moments, rules for the synthesis of an
intuition of an object in a manifold of representations.32 Instead, schematized modal
categories represent objects’ relations to our specifically discursive and spatiotem-
poral form of cognition:
The schema of possibility is the agreement of the synthesis of various representations with the
conditions of time in general (e.g., since opposites cannot exist in one thing at the same time,
they can only exist one after another), thus the determination of the representation of a thing
to some time.
The schema of actuality is existence at a determinate time (A144/B184).

These are the schemata of the modal categories of possibility and actuality for our
discursive spatiotemporal intellect; in my terminology, they are the real definitions of
formal possibility and of actuality for specifically spatiotemporal objects. At one level
of generality higher, there are principles that apply to modal categories for any
discursive intellect whatsoever:
Whatever agrees with the formal conditions of experience (in accordance with intuition and
concepts) is possible.
That which is connected with the material conditions on experience (of sensation) is actual.
(A218/B266)

For any discursive intellect the actuality of an object is its relation to intuition and
the possibility of an object is its relation to its intellectual and sensible forms; these are
essentially the same as the principles for possibility and actuality Kant refers to in CJ }
76 above.
At the highest level of abstraction there are the unschematized modal categories,
but of these we cannot say much more than a few ‘obvious tautologies’:
No one has ever been able to define possibility, existence, and necessity except through
obvious tautologies if he wanted to draw their definition solely from the pure understanding.
(A244/B302)

These ‘obvious tautologies,’ I argued in Chapter 9, are partial nominal definitions

of the unschematized category of (real) possibility. I argued further that the nominal
definition of the unschematized modal categories includes various principles: Non-
Logicality, Groundedness, Absolute Necessity, etc. But nothing in the nominal

31
See B148–9. Kohl (2015) effectively collapses the first two ways of thinking of the categories.
32
A219/B266.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

definition of the unschematized category of (real) possibility says anything about a

cognitive faculty for any mind and definitely nothing about our specifically discursive
faculties. Consequently, to think of an object under the unschematized category of
(real) possibility is not to think of its relation to our discursive intellect.
But surely Kant’s point in the CJ is not merely that the intuitive intellect does not
use the modal concepts a discursive intellect does, much less the specifically tempor-
ally schematized modal categories we use. That is a trivial consequence of the
definition of the intuitive intellect and the idea of a schema of a pure concept of
the understanding: a rule for combining sensory manifolds in an intuition of an
object.33 The substantive question is whether Kant means to be claiming that the
intuitive intellect does not represent its objects even using the unschematized modal
categories. Of course, the unschematized modal categories are concepts, and the
intuitive intellect does not use (general, mediate) concepts, so if there is modal
content in the intuitive intellect’s cognition then it is not conceptual modal content.
It might be that the intuitive intellect represents modal content but does so through
non-conceptual means, that is, that there is some non-conceptual way of represent-
ing modal properties of objects, or something that corresponds to them.
In fact, this is the view Kant articulates in the Pölitz lectures, but with an
unexpected twist. In those lectures Kant claims that the divine intellect (which he
explicitly characterizes as an intuitive intellect)34 does cognize the modal difference
between the possible and the actual:
We represent to ourselves that in cognizing his own essence (simplex intelligentia) God must also
cognize everything possible, since he is the ground of all possibilities. Thus we derive the
cognition of all possibilities from his nature and call it cognitio simplicis intelligentiae.—We
think of scientia libera as God’s cognition of the actual, insofar as he is simultaneously conscious
of his free choice of things; for either all things are actual by the necessity of God’s nature, which
would be the principle of emanation; or else they exist through his will, which would be the
system of creation. We think of scientia libera in God to the extent that in his cognition of
everything possible, God is at the same time conscious in his free will of those possible things
which he has made actual; hence his representation is grounded on the system of creation,
according to which God is the author of all things through his will. (Pöl.RT, Ak. 28: 1054)

God is the ground of all possibilities, so in cognizing his own essence, God cognizes
everything that is possible. God is the causal ground of everything that exists (other
than himself) through his will, so in cognizing himself as the author of those objects,
God cognizes what is actual. In both cases, God’s cognition is a prioriG: he cognizes
all possibilities and actualities through their ground (his essence and his will,
respectively), and this cognition depends upon nothing external to him. In both
cases God’s cognition is a kind of self-cognition; awareness of his essence and

33
A79/B105, B128, and A141–2/B180–1.
34
Pöl.RT (Ak. 28: 996, 1043, 1052), Danz.RT (Ak. 28: 1267), and Volck.RT (Ak. 28: 1165).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

everything that is made possible through it, and awareness of what he is actually,
timelessly, creating. This means that God does cognize a difference among the objects
of his intellect, their different manners of depending upon him: God cognizes some
things as depending solely upon his essence (possible ways objects could be), and a
subset of those ways (ways objects actually are) as depending upon his will (all of the
things whose actuality depends upon God’s will depend for their possibility upon his
essence).35, 36
However, earlier in the same paragraph we find Kant saying, “in regard to God
there is no distinction between the possible and the actual” (Pöl.RT, Ak. 28: 1053),
which I take to mean: God does not cognize anything as actual or as possible. In fact,
the sentence that immediately precedes the quoted passage might be thought to
undermine my interpretation completely:
The distinction between scientia simplicis intelligentiae (knowledge of simple intelligence) and
scientia libera (free knowledge) is to be found only in our human representation of God’s
cognition, and not in this cognition itself. (Pöl.RT, Ak. 28: 1054)

This is why Kant emphasizes the relation to our thinking in the long passage
quoted earlier: we think of God as differentially cognizing what depends upon his
essence and what depends upon his will, but this does not constitute a modal
distinction. However, the text on which Kant is lecturing at this point is Metaphysica
}}874–5, where Baumgarten defines these kinds of cognitions:
}874. God knows all determinations of all things, insofar as these are considered merely
possible. This is scientia simplicis intelligentiae.
}875. God knows all the determinations of the actual beings of this world, which is scientia
libera (of vision).

Notice, though, that Baumgarten defines these kinds of cognitions in explicitly modal
terms. In claiming that the distinction between “scientia simplicis intelligentiae” and
“scientia libera” lies only in our human representation of God’s cognition (Ak. 28: 1054),
Kant is claiming that the representation of this distinction as a modal distinction is due
to us and that modal representation is not part of God’s cognition. He is not denying
that there is a difference in how God cognizes his own essence and his actions.37

35
As Kant remarks: “the actual is already included within the possible, since what is actual must also be
possible, for otherwise it could not be actual. Thus if God is thinking of everything possible, he is already
thinking of everything actual” (Pöl.RT, Ak. 28: 1053–4). But in cognizing actual objects as depending on his
essence, God is not thereby cognizing them as actual, a point Kant recognizes, for he gives different
accounts of God’s scientia libera and cognitio simplicis intelligentiae.
36
In }5 I will be more precise about what ‘object’ means here; for now I am painting with a broad brush.
37
If the distinction between what depends upon God’s essence and his will were only an artifact of our
representations, then the distinction between the “system of emanation” (the existence of the world is a
necessary consequence of God’s essence) and the “system of creation” (the world is freely created by God)
would be merely subjective. But this is not Kant’s view. See Pöl.RT (Ak. 28: 1092–3), Volck.RT (Ak. 28:
1193–5), Danz.RT (Ak. 28: 1298–9), ML1 (Ak. 28: 330), MD (Ak. 28: 701), and MK2 (Ak. 28: 808).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

On this reading, there is a real noumenal distinction in how objects relate to God
and indirectly, to his intuitive intellect: some things depend upon his essence
(nature), while others also depend upon his will. God is omniscient, so these objects
and their different manner of dependence on him are within the scope of his intuitive
intellect; God cognizes this ‘dependence difference’ among objects. In the Pölitz
passage quoted above he claims that it is correct to describe this distinction in
terms of the modal concepts of possibility and actuality only from our standpoint.
Since we are using modal concepts to think about noumena, we are using the
unschematized concepts of possibility and actuality (see Chapters 8 and 9). We use
these concepts to think about a real noumenal distinction in objects that is not in
itself (in its own nature, considered independently of its relation to our cognitive
faculties) a modal distinction.
I am going to refer to these noumenal properties as noumenal ‘correlates’ of our
modal concepts. The noumenal correlates of the unschematized categories of possi-
bility and actuality are the properties, respectively, of depending upon God’s nature or
upon his will. Kant’s view is that God differentially cognizes objects as depending
either upon his will or merely upon his essence, and we represent that difference using
the unschematized modal categories of possibility and actuality. This means that these
noumenal correlates are in the content of God’s intellectual intuition of objects, but
the modal categories are not, not even the unschematized modal categories.
This allows us to see how Kant’s denial of divine cognition of modality in the CJ is
consistent with his view in the Pölitz lectures. In the CJ Kant emphasizes the point
that the intuitive intellect would not represent modal content. In the Pölitz lectures
he repeats that point, but expands upon it by claiming that the difference in our
thoughts about his objects (between those we think as actual and those we think as
merely possible) corresponds to a real distinction in the way objects depend upon
God and in his awareness of them as so depending.
This means that (P1)–(P3) can be reformulated as follows:
(P1*) Noumena have noumenal correlates of unschematized modal categories
(noumenal properties that we think under the unschematized modal categories).
(P2*) The intuitive intellect does not cognize its objects as having modal prop-
erties (though it does cognize the non-modal noumenal correlates of unschema-
tized modal categories).
(P3*) Noumena have noumenal correlates of modal categories if, and only if, the
intuitive intellect cognizes them as having these noumenal correlates (whether or
not it cognizes them as correlates of modal categories).
In other words, for each unschematized modal category there is a corresponding
non-modal property N such that God cognizes N in objects but does not cognize it
modally (as being a modal property). These are consistent, so the original modal
antinomy is resolved. In the next section I clarify this interpretation and extend it to
the modal properties of noumenal wills.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

10.5. Noumenal Freedom without Noumenal Modality

Up to this point I have referred somewhat loosely to the ‘objects’ of divine intuition
and their ‘modal properties,’ but from now on I will be more precise. Since existence
is not a real predicate, there are no merely possible objects (q-objects, objects of
absolute positing); translated into the noumenal correlate of possibility, this means
there are no noumenal objects that only depend upon God’s essence without being
created by his will. All the noumenal objects exist and are (represented by us as)
actual. Our concept of noumenal (real) possibility does not have a set of noumenal
objects as its correlate, but a set of properties of noumena, the properties we represent
as being (really) possibly instantiated. These properties stand in relation R (to be
specified further below) to God’s essence. This relation is the noumenal correlate of
our concept of the relation of a really possible property to a ground of its real
possibility; the properties that stand in R to God’s essence are the noumenal
correlates of our concept <possibly instantiated>. The noumenal correlate of our
concept of the actual properties of noumena is the set of properties possessed by
noumenal objects; some of those properties are created by God in noumenal objects,
while some are causally determined by those noumena themselves (see later discus-
sion). All of the properties instantiated by noumena (which we represent as their
actual properties) stand in R to God; in other words, all of the properties we represent
as actual we also represent as possible. God cognizes the objects there are (and some
of their properties) as depending for their existence (instantiation) on his will, and
cognizes that all of these properties, as well as some uninstantiated ones, stand in R to
his essence. In this fashion God cognizes the non-modal noumenal correlates of our
unschematized modal concepts.38
In the Pölitz lectures Kant does not explain what relation to his essence God is
cognizing in scientia simplicis intelligentiae, that is, what relation to God’s essence is
the noumenal correlate of possibility. Nonetheless, it is not hard to fill in this gap.
Elsewhere he says this about God’s cognition of actuality:
We think of a scientia libera in God to the extent that in his cognition of everything possible,
God is at the same time conscious in his free will of those possible things which he has made
actual; hence this representation is grounded on the system of creation, according to which
God is the author of all things through his will. (Pöl.RT, Ak. 28: 1054)

The possible is the possibly actual, so in representing a property as possible we are

representing it as something possibly actualized by God. Since actuality is our modal
representation of what depends upon God’s will, it stands to reason that possibility is
dependence upon God’s potential will, his power. The properties God cognizes

38
A problem arises about how God can be said to cognize properties since he does not cognize
discursively (hence does not represent properties conceptually as predicates of possible judgments, as we
do). I do not have space to address this problem here; see, however, the supplementary note “Intuitive
Intellect and Cognition” on my website (see Notes on the Text).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

through scientia simplicis intelligentiae are the properties he has the power to cause to
be instantiated. We represent those properties modally as possibly instantiated
properties.
My aim all along has been to reconcile the contingency of the actions of noumenal
wills with (P2) and (P3), but the model sketched so far only applies to the existence of
objects created by God and the properties he has the power to cause to be instantiated
in them. I will now attempt to extend this model to the noumenal correlates of
modality for the choices of noumenal wills, which I will treat as properties noumenal
wills cause to be instantiated in themselves (e.g., I cause myself to have the property
of being radically evil). The noumenal will of a finite rational agent makes a single
timeless choice either to subordinate morality to self-interest (radical evil) or to
subordinate self-interest to morality (moral goodness).39 The noumenal correlate
of the possibility that I choose otherwise than I do cannot be God’s power to cause me
to choose otherwise (or to create me with a will that chooses otherwise)40 for that
would destroy my freedom. Likewise, if the noumenal correlate of the actuality of my
choice (radical evil) were God’s causing me to choose, or creating me as so choosing,
then freedom would be an illusion.
We have two options. The first would be to claim that the noumenal correlate of
the possibility of my choosing otherwise is also God’s power; likewise, the noumenal
correlate of the actuality of my choice is at least partial dependence on God’s will
(God is either the partial or complete cause of my choice). This could take the form of
Malebranchean occasionalism, Spinozistic monism, or divine concurrence (I do not
have the power to do anything, including to determine my own will, without God’s
participation or ‘concurrence’). Since Kant rejects the first two and expresses an
at-best ambivalent attitude towards divine concurrence,41 I think we should look
elsewhere. The second option is that the noumenal correlates of the actuality and
possibility of determination in my will are, respectively, what my will chooses and my
power to choose otherwise. I choose to subordinate morality to self-interest—my will
is the cause of my radical evil42—but I retain the power to choose otherwise. Restrict-
ing ourselves to properties that are within the power of my noumenal will,43 the

39
Rel., Ak. 6: 25, 36, and the rest of Part 1 of Rel. Obviously, there are serious problems with interpreting
Kant’s ‘rigorism’ (e.g. how is moral conversion possible? Cf. Ak. 6:48), but I cannot address them here.
40
In the former case, I do not essentially choose radical evil, but God creates me and causes me to so
choose; in the latter case, I essentially choose radical evil, so in creating me God is creating me choosing
evil.
41
In Pöl.RT. he describes divine concurrence in free acts as an incomprehensible miracle (Ak. 28:
1105–7), but “quite possible” (Ak. 28: 1110). He is more skeptical of divine concurrence in Danz.RT (Ak.
28: 1039) and Volck.RT (Ak. 28: 1209). Cf. Hogan (2014).
42
Rel., Ak. 6: 31, 36.
43
Some of my properties (e.g., my transcendental freedom, my finiteness) are direct consequences of
God’s creative act; since I actually possess these properties, we need some non-modal noumenal correlate
of their actuality. See section 7.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

noumenal correlate of actuality is the set of properties my will causes to be instantiated

(in the cases of all human beings, radical evil),44 and the noumenal correlate of
possibility is the set of properties my will has the power to cause to be instantiated
(on Kant’s view, this set contains only two properties: radical evil and moral
goodness).
My solution involves replacing the original (P1) with (P1*), or, more specifically,
replacing the claim that noumenal wills are only contingently evil with the more
complicated claim that noumenal wills have the non-modal noumenal correlate of
actually willing evil (they cause themselves to be evil, i.e. they choose to subordinate
morality to self-interest) and the non-modal noumenal correlate of possibly not being
evil (they have the power to choose otherwise, i.e. to subordinate self-interest to
morality). This requires developing a non-modal characterization of freedom, a
‘noumenal correlate’ of freedom that we then represent modally as possibly willing
otherwise than it actually wills.
In fact, Kant has the materials to do precisely that. At least since the mid-1760s
Kant has held the Crusian view that free actions of finite rational agents (noumenal
wills) lack determining grounds.45 In doing so, he is drawing on Crusius’s distinction
between a determining ground and a merely sufficient ground.46 A merely sufficient
ground brings about its effect but retains the power to omit that effect in the same
circumstances (Umstände); a determining ground always brings about the very same
effect in the same circumstances. While Crusius and Kant often characterize this
distinction in modal terms (a merely sufficient ground possibly has a different
consequence in the same circumstances, a determining ground necessarily has the
same consequence in the same circumstances), it is possible to define them non-
modally and use that distinction (between merely sufficient and determining
grounds) to understand modality, as follows.47 Crusius refers to a power that is
indifferent in its nature between two outcomes as a “power of freedom”: when a
substance with a power of freedom acts in a given circumstance and posits a
consequence it retains the power to posit the opposite consequence in the same

44
Rel., Ak. 6: 32–3.
45
This shift from the compatibilist theory of freedom in ND (Ak. 1: 396–410) to a libertarian theory first
appears in a series of Reflexionen from the 1760s: Refl. 3717 (Ak. 17: 260), 4034 (Ak. 17: 392), 4338 (Ak. 17:
510), 4544 (Ak. 17: 588), 4693 (Ak. 17: 677), and 5251 (Ak. 18: 132). See Hogan (2005) and (2009a) for
critical discussion.
46
Ent. }}82, 84 (148); De Usu }}II, V, XXII.
47
Crusius: “to determine means nothing more than to leave a single possibility for how a thing in these
circumstances is constituted or could be constituted” (De Usu }II; cf. }XVI). Kant: “[that] everything which
happens has a determining ground, that is, something else which necessitates it, is the principle of the
changes of all passive substances” (Refl. 4338, Ak. 17: 510). Kant cites Crusius’s distinction in ND, but his
explication of it, while non–modal, is not entirely perspicuous: “since, however, to determine is to posit in
such a way that every opposite is excluded, the term ‘determine’ designates that which is certainly sufficient
to conceive the thing in such and such a way, and in no other” (Ak. 1: 393). Nonetheless, in that early work
he maintains that even the free actions of rational agents have determining grounds.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

circumstances.48 A merely sufficient ground of a consequence is the activation of

a power of freedom. A determining ground of a consequence is the activation of a
power that is not a power of freedom. Consequently, a consequence that has a
determining ground has a completely intelligible explanation: the ground had no
power to posit the opposite. By contrast, a consequence with only a sufficient ground
cannot be made fully intelligible: the ground had the power to posit its opposite in the
very same circumstances, and did not.
Starting in the mid-1760s Kant converts to the Crusian view that the free acts of
finite rational agents (noumenal wills) lack determining grounds. In other words, our
acts are the products of Crusian ‘powers of freedom’; we retain the power to do the
opposite under the same circumstances. Since the notion of a power of freedom can
be defined non-modally (as I did above) and the notions of determining and
sufficient grounds can be defined in terms of a power of freedom, they can also be
defined non-modally. This means that freedom (or at least the ‘leeway’ requirement
on freedom)49 can be characterized non-modally: I am free because my acts are the
consequence of a power of freedom, meaning, I retain the power to have acted
differently under the very same circumstances.50
This will strike some readers as arguing in a very tight circle. Freedom is charac-
terized in terms of the power to have done otherwise under the very same circum-
stances. But isn’t the power to act otherwise to be understood modally as possibly
acting otherwise? It is important to realize that this is not Kant’s view. In CJ }76 and
the Pölitz lectures he claims that an intuitive intellect would not cognize modally, but
he does not claim it would not cognize powers, grounds, etc. In fact, he explicitly
claims that God is aware of himself as the cause of the world.51 Since God is
omniscient and noumenal wills possess transcendental freedom, then either tran-
scendental freedom does not involve power, or noumenal wills do have power and in
cognizing this power God is not cognizing anything intrinsically modal. The repre-
sentation of powers in modal terms (e.g., the possibility of acting otherwise than they
do) is due to us, as I will explain in detail in the next section.

48
Ent. }82.
49
I am not claiming that the lack of a determining ground is sufficient for freedom. Kant still maintains
that spontaneity (being the ground of one’s action) and intelligence (being the ground of one’s action
through representing it) are necessary, but not sufficient, conditions on freedom (CPrR, Ak. 5: 96–7, 101).
There is no reason to think these need to be understood modally. I am arguing that a non-modal
characterization can be given of the ‘leeway’ requirement, which Kant sometimes characterizes in terms
of absolute contingency: the free acts of a rational agent must lack determining grounds.
50
In the Kantian context, we are talking about the free choices of noumenal wills, so circumstances will
be non-spatial and non-temporal; they may include, for instance, facts about which noumenal wills God
has created, which properties he has created in them (e.g., finiteness), and whether they causally interact.
See A541/B569, A551f/B579f, A553/B581, and CPrR (Ak. 5: 100–3). Thanks to Hogan (2014) for drawing
my attention to this point.
51
Pöl.RT, Ak. 28: 1053–5.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

Several difficulties still remain. First of all, there is the difficulty of reconciling
God’s creation of my noumenal will with my free choice.52 Secondly, there is the
difficulty of how God can be said to have a prioriJ cognition53 of my choice if he is not
himself the ground of my choice and there is no completely determining ground of
that choice. Since God is not passive he cannot ‘wait and see’ how I will choose to act.
However, my action has no determining ground, so there is nothing ‘prior’ (in the
order of explanation, not temporally, since my noumenal will is atemporal)54 to my
choice from which God could ‘predict’ it.55 I do not want to minimize these
difficulties, but merely to point out that they are difficulties with Kant’s theory of
freedom, not with reconciling that theory with the amodalist consequences of (P2)
and (P3). Regarding the first problem, Kant claims that it is a mystery how freedom is
compatible with divine creation, but this should not weaken our credence in either
(practical cognition and moral faith, respectively).56 The second problem is really the
problem of how to reconcile Kant’s libertarian theory of freedom (free acts lack
determining grounds) with the doctrine of divine non-passivity; this is a problem
that Kant faces as a libertarian theist, independently of his doctrine of noumenal
amodalism.57

10.6. Representing Noumenal Modality

In this section I am going to systematize the remarks I have made over the course of
sections 4 and 5 about noumenal correlates and how we represent them using
unschematized modal concepts.
In the previous chapter I gave an account of the content of the unschematized
concept of real possibility überhaupt, of which various different kinds of real possi-
bility are specifications. I have also discussed a specific such kind of real possibility,

52
Kant admits as much in Rel.: “it is however, totally incomprehensible to our reason how beings can be
created to use their powers freely, for according to the principle of causality we cannot attribute any other
inner ground of action to a being, which we assume to have been produced, except that which the
producing cause has placed in it. And, since through this ground (hence through an external cause) the
being’s every action is determined as well, the being itself cannot be free. So through our rational insight
we cannot reconcile the divine and holy legislation, which only applies to free beings, with the concept of
the creation of these beings, but must simply presuppose the latter as already existing free beings” (Rel., Ak.
6: 142; cf. 44, 191). Cf. CPrR, Ak. 5: 100–3; and Pöl.RT, Ak. 28: 1105–7, 1110). See Brewer and Watkins
(2012) for further discussion.
53
For reasons I noted in }2 it is not strictly correct to talk of God’s ‘knowledge’ [Wissen] because
knowledge is a mode of holding a judgment to be true [Fürwahrhalten] and judgment is a discursive form
of representation.
54
CPrR (Ak. 5: 101–2), Pöl.RT (Ak. 28: 1095).
55
Kant is aware of this problem, but it is unclear he has a consistent solution; cf. MK2 (Ak. 28: 803) and
Pöl.RT (Ak. 28: 1055). See Hogan (2014) for discussion.
56
Rel., Ak. 6: 142. See previous discussion.
57
Cf. Leibniz’s critique of libertarianism in Theodicy (Th. }}41, 42). This was a point Kant was well
aware of, for he himself wielded it against libertarians in ND: “divine foreknowledge is only possible in
respect of free actions if it is conceded that their futurition is determined by their own grounds” (Ak. 1:
405).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

noumenal real possibility. Because noumenal real possibility is a kind of real possi-
bility it instantiates this scheme:
(Noumenal PossibilityN) Noumenal possibility ◇Np is a kind of real possibility
only if
(iii) Non-logicality: it is not a conceptual truth that ◇Lp ⊃ ◇Np (equivalently, it is
not a conceptual truth that □Np ⊃ □Lp), and
(iv) Groundedness: if ◇Np then the fact that ◇Np has a real ground in some
actual object or principle.
(v) Absolute necessity: there is an associated notion of absolute necessity (abso-
lute-□Np) defined as follows: absolute-□Np=def ¬p ) (for all q, ¬◇Np).
There is some actual p about the grounds of all noumenal possibility such
that absolute-□Np.
This is the concept whose non-modal noumenal correlates I will attempt to identify.
Up to this point, I have not discussed the concept <actuality> in much detail. I will
offer a more complete interpretation of the unschematized concept of <actuality>58 in
the next section, but for now I just want to point out that we need some principles to
guide our use of this concept in thinking about the difference between what is actually
the case with noumenal wills (e.g., that I am radically evil) and what is merely possibly
the case (e.g., that I am morally good). Just as with possibility, we need to distinguish
different kinds of actuality: the different concepts of actuality will differ according to
the different objects to which they are applied (e.g., actuality for phenomena will be
different than actuality for noumena). We can think of the principles of actuality
überhaupt as principles any lower-level concept must satisfy to be a concept of
actuality. The principles of the unschematized concept of actuality (the minimal
conditions any concept must meet to be a concept of actuality) are as follows:
(Actuality) A sentence operator actually(p) and associated object-level predicate
actual(x) are concepts of actuality only if:
(i) Is-Act. What is, is actual: (a) for all propositions p if p then actually(p), and
(b) for all objects x, actual(x).
(ii) Alethic. Actuality entails possibility: for all propositions p if actually(p) then
possibly(p).
Strictly speaking, we should subscript the relevant notions of actuality and possi-
bility and restrict the scope of objects and propositions accordingly: empirical-causal
actuality (being an actual alteration) entails empirical-causal possibility, etc. but for
simplicity I will suppress that complication. Each principle has two parts because we
have two concepts of actuality: as an operator on propositions (what is actually the
case) and a predicate of objects (which objects are actual). Note that clause (ii) holds

58
As well as some reasons to think that <actuality> is indeed a category distinct from <existence>; cf.
MvS, Ak. 28: 493.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

only for the relation between actuality and alethic kinds of possibility; in particular, it
does not apply to the deontic notions of the permissible and the forbidden, the
“categories of freedom” discussed in section 3. Consequently, it is not part of
the nominal definitions of actuality and possibility überhaupt, but it is part of the
nominal definition of all of the kinds of real possibility I have been discussing in this
book so far (with the exception of permissibility in section 3).59 In order to deploy
the unschematized concept of actuality appropriately we also need to respect this
principle.
There is a very close structural isomorphism between, on the one hand, the
concepts of noumenal real possibility and actuality (what is actual for noumena)
and (what I have claimed are) their noumenal correlates. Let p range over logically
contingent monadic propositions about noumena that are not themselves modal and
existential generalizations thereof (i.e. of the form Fa or 9xFx); let x range over
noumena.60 Consider these principles about the noumenal correlates and the cor-
responding principle about possibility and actuality (listed in parentheses):
(i) It is not a conceptual truth that if ◇Lp then some noumenon has the power
to make it the case that p. (Non-logicality.)
(ii) If some noumenon has the power to make it the case that p this is grounded
in a fact about God or some object God creates (the noumenon with that
power). (Groundedness.)
(iii) For all p that some noumenon has the power to make the case, if God did not
exist, nothing would have the power to make it the case that p. (Absolute
necessity.)
(iv) If p is true, then either p is about God himself, or p is made true by the will of
God or some other noumenon. For any x, x is either God, or x exists because
God creates it. (Is-Act.)
(v) If p is true, then either p is about God or God or some other noumenon has
the power to make it the case that p. (Alethic).61
Clearly, the non-modal principles (i)–(iv) correspond to the indicated modal
principles. The reason for this is that when we map the noumenal correlates onto
our modal concepts in the right way, these non-modal principles are transformed into

59
I do not think that Alethic is part of the most general concept of real possibility. In the CPrR (Ak.5:
65–6, see section 3) Kant says that the modal category <possibility> can be used to think about permis-
sibility. Permissibility is not a kind of alethic possibility. I argued in Chapter 9 that the unschematized
modal category <possibility> is the most general concept of real possibility. It follows that real possibility
überhaupt is not specifically a concept of alethic possibility: it is used to think about alethic possibilities
(e.g., formal possibility) and non-alethic ones (e.g., permissibility).
60
Intuitively, the idea behind this restriction is that we want to consider the grounds of real non-logical
possibilities for noumena (hence the restriction to logically contingent propositions) and we want to ignore
iterated modalities and logically complex properties.
61
It is important to note that God himself is a noumenon (he is not a possible object of sensible
intuition).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

Table 10.1. The non-modal noumenal correlates of possibility and actuality

Non-modal noumenal correlate Represented modally as

(a) God’s powers and the powers of other (a*) God’s essence and the essences of other
noumenal wills noumenal wills
(b) Properties instantiated by God and (b*) A subset of noumenally really possible
properties he has the power to cause other properties
noumena to instantiate
(c) Properties finite noumenal wills have (c*) A subset of noumenally really possible
the power to cause themselves to instantiate properties (for finite noumenal wills)
(d) the relation between (b) and God’s (d*) The grounding relation making really
power and between (c) and the powers of noumenally possible
finite noumenal wills
(e) God himself and all the other noumena (e*) Actuality as a predicate of (noumenal)
he has created objects
(f) True propositions about God and true (f*) Actuality as an operator for
propositions about other noumena he has propositions about God and his creative
caused to be true acts
(g) Propositions that attribute to finite (g*) Actuality as an operator for
noumenal wills properties they have caused propositions about the choices of noumenal
themselves to instantiate wills
(h) The relation between (f ) and God’s will (h*) The relation of making actual
and (g) and finite noumenal wills

the corresponding modal principles. Table 10.1 gives the non-modal noumenal correlates
in the left-hand column, and how we represent them modally in the right-hand column.
It can easily be checked that if we map non-modal noumenal correlates onto
modal concepts according to Table 10.1 then the non-modal principles (i)-(iv) are
transformed into the modal principles that correspond to them (in parentheses). For
instance, if we represent properties God has the power to cause to be instantiated (b)
and properties finite noumenal wills have the power to cause themselves to instan-
taite (c) as possible ((b*) and (c*)), the fact that (i) it is not a conceptual truth that if p
is logically possible then some noumenon has the power to cause it to be the case that
p is represented modally as Non-logicality: it is not a conceptual truth that if p is
logically possible then it is noumenally really possible. Likewise, if we represent true
propositions about God and propositions he or some other noumenon causes to be
true ((f) and (g)) as actual ((f*) and (g*)), the fact that everything God or other
noumena cause to be the case is within their power is represented modally by Alethic.
What this means is that we use our unschematized concepts of possibility and
actuality to represent the structurally isomorphic relations among the powers of
God and finite noumenal wills and their actions. At the noumenal level considered ‘in
itself ’ there is no modality; these objects and properties (and propositions about
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

them) have a modal status (possible, actual, contingent) only relative to our mod-
alized thought about them.
Before continuing, I would like to respond to one lingering issue. From our modal
perspective we can say that God is the ground of the possibility of the existence of
noumenal objects in virtue of having the power to create them, but noumenal wills
are the ground of the possibility of their free choices (in virtue of having the power to
so choose). Some readers will object that this inappropriately limits the dependence
of all noumenal possibility on God, but in the Pölitz lectures Kant himself reiterates
that God is (represented by us as) the ground of all possibility whatsoever.62 But,
within my interpretation, we can nonetheless recover this absolute dependence of all
of (what we represent as) possibility upon God. Take my noumenal will and some
property (moral goodness) it has the power to produce in itself. My noumenal will
would not have this power unless God created it with that power, and this would not
be the case without God having the power to create it with that power. So the moral
goodness that is within my power is mediately related to God’s (actual) will that
creates this power, and yet further mediately related to the divine power of which that
creative act is an activation. Representing these noumenal correlates in modal terms,
the possibility of my moral goodness is grounded in my power to so choose. I actually
possess this power (the immediate ground of possibility) because of God’s creation of
me as a transcendentally free finite rational will. But this creative act was possible in
virtue of God’s power. This power has no further ground of possibility. So the
mediate but final ground of the possibility of my moral goodness is God’s power.
By parity of reasoning, this will apply to any property possibly possessed by any finite
being. Consequently, we can retain the dependence of all really possible properties of
noumena upon God (specifically, on his powers).

10.7. Absolute Positing, Existence, and Actuality

Up to this point I have mainly been concerned with possibility and whether it figures
in the intuitive intellect’s cognitions. But Kant does not say merely that possibility
does not figure in its cognitions; he says that every object of an intuitive intellect
would exist and would be actual:
That is, if our understanding were intuitive, it would have no objects except what is actual.
(CJ, Ak. 5: 402)
For an understanding to which this distinction [between thinking and intuiting] did not apply,
all objects that I cognize would be (exist), and the possibility of some that did not exist, i.e. their
contingency if they did exist, as well as the necessity that is to be distinguished from that, would
not enter into the representation of such a being at all. (CJ, Ak. 5: 403)63

62
Ak. 28: 1034, 1054.
63
In Refl. 6020 Kant claims that for an intuitive intellect “only what is actual is possible” (Ak. 18: 426).
But this means either that the intuitive intellect cognizes no difference between the actual and the possible,
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

In this section I want to clarify how the intuitive intellect represents what exists and
what is actual, and whether and how these are different. As I have noted before, Kant
sometimes identifies the second category of modality as ‘existence-nonexistence’
(Dasein-nichtsein), sometimes simply as ‘existence’ (Dasein), and sometimes as
‘actuality’ (Wirklichkeit).64 It has been a running theme of this book that Kant uses
‘existence’ and ‘actuality’ in a number of different ways, and with the idea of an
intuitive intellect in place we finally have the resources to systematize and explain
those different uses.
1. Existence as absolute positing (what there is). The first notion we need to
distinguish is existence as ‘absolute positing’: there being an object. This is an
unschematized use of the concept because we can use it to think about
noumena; the question of whether God exists is the question of whether there
is an object that instantiates the concept <God>. This notion of existence was
the theme of Chapters 1 and 2. This is the notion of existence that is not a ‘real
predicate’ or a determination. Intuitions give us objects that exist in this sense:
objects that are absolutely posited (the objects there are, q-objects). Since the
intuitive intellect only intuits, and what is more, must intuit itself, it does
represent objects as existing in this sense: it absolutely posits its objects. This
is what Kant means when he writes: “for an intellect to which this distinction
[between thinking and intuiting] did not apply, all objects that I cognize would
be (exist)” (CJ, Ak. 5: 403).
2. Existence as causal efficacy. This is the concept I have previously dubbed
existence*: the concept of an object possessing causal powers. This concept is
a determination, for some objects do not exist*, e.g., mathematical objects.65
Admitting that existence* is a real predicate does not, however, open any room
for ontological arguments, for, from the analytic judgment God exists*
(<exists*> is one of the marks of the concept <God>) we cannot prove that
there is a God (that God exists in the ‘absolute positing’ sense above). This is
not properly even a modal category; the concept of existence* is really under-
stood through the categories of relation (esp. <cause–effect>). Since we think
about noumena as having causal powers (e.g., transcendental freedom), there is

in which case it is exactly what he says in }76, or it means that the intuitive intellect cognizes only actual
things and cognizes them as actual, in which case it is inconsistent with his considered view in }76. Thanks
to Ralf Bader for calling my attention to this Reflexion.
64
In the initial table of the categories in the first Critique he gives ‘Dasein–Nichtsein’ as the second
category of modality (A80/B106), but in the Postulates of Empirical Thought, the later section devoted to
the modal categories, the second modal category is now the category of ‘Wirklichkeit’ (A218/B266). See also
MvS, where Kant writes: “real actuality is here the category of existence [Existenz], in contrast to the
possibility of a thing” (Ak. 28: 493). Cf. ND (Ak. 1: 396), OPG (2: 72, 75), Volck.RT (28: 1151), and Danz.RT
(Ak. 28: 1256, 1291). See, however, MK3 (Ak. 29: 986) as well as Refl. 6324 (Ak. 18: 644), where Kant
explicitly distinguishes actuality from existence.
65
See MFNS, Ak. 4: 468.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

an unschematized concept of existence* (thought using the unschematized

categories of relation). Since (if there is an intuitive intellect) the intuitive
intellect exists* and it cognizes itself, it must cognize at least one existing
object.* Presumably it must also cognize itself as existing*, but I do not want
to get involved here in the puzzle of how the intuitive intellect represents
properties of its objects (see section 2), including its own properties. It is
clear, though, that God cognizes himself as existing* since he cognizes himself
as the author of the world.66
3. Actuality as a modal concept. Finally, when we use the unschematized modal
category of <actuality> we represent all the objects there are (existence as
absolute positing) and everything that is the case about them as actual to
distinguish it from what is merely possible (e.g., I actually subordinate the
moral law to my self-interest, but possibly do otherwise). Consequently, the
unschematized modal categories <possibility> and <actuality> are interdepend-
ent; we cannot use one without using the other. Their interconnection is given
by two principles about actuality (see section 6):
(i) Is-Act. If p then actually p. For all objects x, x is actual.
(ii) Alethic. If actually p then possibly p.67
The first principle means that we represent everything that is the case as being
actually the case and we represent all objects as actual. This may seem trivial, but it is
a crucial constituent of applying our concept of actuality and thus modalizing what
were otherwise non-modal concepts: what is the case, and what objects there are. The
second principle means that (working within an alethic modality) we represent all
actual truths as possibly true, and all actual objects as possible.
The possibility–actuality connection is also given by two principles about possibility:
(iii) Groundedness: if ◇p then the fact that ◇p has a real ground in some actual
object object or principle.
(iv) Absolute necessity: there is an associated notion of absolute necessity (abso-
lute-□p) defined as follows: absolute-□p=def ¬p ) (for all q, ¬◇p). There is some
actual p such that absolute-□p.
The idea is this. If we are thinking about some objects, some concepts of them, and
some propositions about them, then in order to think them under the (unschematized)
modal categories we have to represent all of the objects there are as actual and all of the
propositions that are true of them as actually true. By using the concepts <possible> and
<actual> according to these principles, and the other principles of real possibility
überhaupt (see section 6), we come to represent those objects and propositions using

66
Pöl.RT, Ak. 28: 1054.
67
The second principle only applies where the relevant notion of possibility is alethic (see sections 3
and 5).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

the unschematized modal categories <possibility> and <actuality>. To think about them
using various more specific modal notions (e.g., formal possibility) we need to employ
the more specific principles described in Chapters 7 and 8. This notion of actuality—
defined in terms of its relation to the unschematized category of possibility—is the most
general modal notion of actuality/existence. It represents, so to speak, the ‘modalization’
of existence in the ‘absolute positing’ sense, for it requires us to represent all the objects
there are as actual and not merely possible. For the sake of clarity I will refer to it as
actuality to distinguish it from existence (absolute positing) and existence*.
Kant claims in CJ }76 that “if our understanding were intuitive, it would have no
objects except what is actual” (CJ, Ak. 5:402). Some readers might conclude from this
that the intuitive intellect does represent actuality (as I am using this term). However,
I do not think this can be correct. One cannot have the concept of actuality (in this
sense) unless one represents it in contrast to mere possibility; this notion of actuality
is an inherently modal concept. Kant denies that the intuitive intellect represents
anything as possible, so I do not think that he is claiming that the intuitive intellect
cognizes its objects as actual. Instead, I think Kant is speaking from our modal
perspective here. All of the objects of the intuitive intellect’s cognition fall under our
concept of actuality because they include: the intuitive intellect itself (God), its
various powers, and the noumenal objects it creates. The intuitive intellect represents
itself as having the power to instantiate various properties but does not, in doing so,
intuit anything that is non-actual (from our perspective); the object of its intuition is
its power to cause that property to be instantiated. The more precise Kantian claim is
expressed later in the same paragraph:

For an understanding to which this distinction [between thinking and intuiting] did not apply,
all objects that I cognize would be (exist), and the possibility of some that did not exist, i.e. their
contingency if they did exist, as well as the necessity that is to be distinguished from that, would
not enter into the representation of such a being at all. (CJ, Ak. 5: 403)

I take this to mean that (prescinding from our modalized concepts) all of the
objects of an intuitive intellect exist in the ‘absolute positing’ sense. This is not the
case for discursive intellects like ours because we form thoughts involving concepts
that correspond to no existing objects (no objects given in intuition).
In the previous section I identified the noumenal correlate of actuality (as a
predicate of objects) as follows: God and any noumenon that depends upon God’s
will. This may appear ad hoc to some readers; I have picked the noumenal correlate
of the unschematized category <actuality>, they might object, precisely so that the
structural isomorphism between the categories <possibility> and <actuality> and
their noumenal correlates would obtain. The discussion of this section, though,
alleviates this appearance of being ad hoc. Any noumenon is either God himself or
exists because God has willed it to exist (according to our practical and theoretical
belief in the existence of God). God’s will is the ground of the existence of noumena
in the ‘absolute positing’ sense of existence; for instance, there is my noumenal will
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

because God wills it to be. We then represent what there is (according to our
practically and theoretically grounded representation of God) modally by subsuming
it under the principle Is-Act and thereby representing those objects as actual. Since
there are no noumenal objects other than God and what God has willed to exist,
nothing else falls under the noumenal correlate of actuality.

10.8. Back to Beweisgrund

Some readers will wonder how my interpretation in this chapter, on which noumena
possess only non-modal correlates of our unschematized modal concepts, is com-
patible with my argument in Chapter 9 that we can possess theoretical rational belief
[Glaube] in an absolutely necessary ground of all noumenal real possibility. If
noumenal modality is merely a ‘projection’ of our discursive intellect and its modal
concepts then why is theoretical reason engaged in a regressive search for noumenal
grounds of noumenal real possibility?
It is clear, though, that Kant takes his noumenal amodalism ((P2) and (P3)) to be
compatible with the subjective rational necessity of positing an absolutely necessary
ground of noumenal real possibility, for he reiterates the latter claim in CJ }76, the
locus classicus of the former doctrine:
For that [actuality] cannot be inferred from [possibility], hence that those propositions [e.g.,
that not all possibilities are actual] are certainly valid of objects insofar as our cognitive faculty,
as sensibly conditioned, is concerned with objects of these senses, but are not valid of objects in
general, is evident from the unremitting demand of reason to assume some sort of thing (the
original ground) as existing absolutely necessarily, in which possibility and actuality can no
longer be distinguished at all, and for which idea our understanding has absolutely no concept, i.e.
can find no way in which to represent such a thing and its way of existing [ . . . ] (CJ, Ak. 5: 402)

It is precisely because we have a discursive intellect, which gives rise to our

representation of the distinction between between possibility and actuality, that we
are rationally required to postulate an absolutely necessary ground of possibility:
Hence the concept of an absolutely necessary being is an indispensable idea of reason but an
unattainable problematic concept for the human understanding. It is still valid, however, for
the use of our cognitive faculties in accordance with their special constitution, thus not for
objects and thereby for every cognitive being [ . . . ] Now here this maxim is always valid, that
even where the cognition of them outstrips the understanding, we should conceive all objects
in accordance with the subjective conditions for the exercise of our faculties necessarily
pertaining to our (i.e. human) nature [ . . . ] Just as in the theoretical consideration of nature
reason must assume the idea of an unconditioned necessity of its primordial ground [ . . . ]
(CJ, Ak. 5: 402–3)

Here Kant reiterates the very claim he makes in the Pölitz theology lectures. In the
very same section of the third Critique where Kant denies that an intuitive intellect
cognizes its objects modally he reiterates his claim that we are rationally required to
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

postulate that there is a being that grounds all real possibility. What is more, we know
from the Pölitz lectures that this being (God) has an omniscient intuitive intellect.
We represent it as grounding all real possibility, and it cognizes itself as grounding all
objects but not modally.
In Chapter 9 I reconstructed the rational necessity of this postulate as a conse-
quence of reason’s regressive search for the conditions (grounds) of noumenal real
possibility. The idea of God as the ground of all real possibility is the idea of an
unconditioned condition of all real possibility. However, the very same line of
reasoning applies if we abstract from our modal concepts and think about non-
modal noumenal correlates. The non-modal noumenal correlates we represent as
grounding the possibility of our noumenal choices are our powers. Through practical
reason we represent ourselves as having the power to make either of two noumenal
choices (radical evil or moral goodness) but reason requires us to think about a
condition of these powers, since these powers are not themselves thought as uncon-
ditioned. This in turns requires us to find a condition of that condition, and so on,
until we form the idea of an unconditioned condition of all powers of noumena. This
unified ground has powers that are activated when it creates us with a power of
choice (represented as a ground of possibility of choice). This unconditioned condi-
tion of our powers is the non-modal noumenal correlate we represent as the unified
ground of all noumenal real possibility.
In Chapter 4 we examined in detail Kant’s theory of God as the ground of all real
possibility in Beweisgrund. Various models have been proposed in the secondary
literature for how God grounds real possibilities: through his intellect (Yong 2014),
through his power (Stang 2010), or by instantiating them (Chignell 2009a). However,
as I pointed out in Chapter 4, Kant not only explicitly states that none of these views
is correct, but notably fails to identify what it is in God by which he grounds all real
possibilities.68 God’s essence grounds all real possibilities, including his own power,
wisdom (intellect), and goodness; consequently, neither his power, nor his intellect,
nor his goodness can itself be the ground of real possibilities. Not only do we not
understand what it is in God through which he grounds all real possibilities, but Kant
denies that we could ever understand or even identify it:
Though one can form, through an analogy with human action, some concept of how a being
can be the cause of everything actual, one can form no concept of how such a being could
contain the ground of the inner possibility of other things. It appears that this thought rises far
higher than a created being can reach. (OPG, Ak. 2: 152–3)

In Beweisgrund Kant lacks not only an adequate epistemology of how we can know
this metaphysical theory of real possibility to be true (something he will come to
realize in the Critical period); as he himself admits, he lacks an adequate explanation

68
OPG, Ak. 2: 125.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS 

of how its content could even be among the contents represented by our discursive
intellects.
In a reversal of the interpretive strategy that has been popular in recent Kant
scholarship—in which the pre-Critical works are used to shed light on the Critical
philosophy—Kant actually develops an account of the modal content of our repre-
sentation of God in the Critical philosophy that fills the gap he himself noticed in
Beweisgrund, thus retrospectively clarifying that philosophical work, but from a
specifically Critical point of view. We have already seen, in previous chapters,
Kant’s Critical reconstruction of the ‘only possible ground of proof ’ itself as an
expression of the ‘only possible way’ to satisfy reason’s regressive demand for an
explanation of why anything is really (noumenally) possible at all. But now we can
see that Kant’s Critical theory also retrospectively answers the question Kant himself
raises, but does not answer, in Beweisgrund: how can we represent whatever it is in
God by which he grounds all real possibilities?
We represent God as the ground of all real noumenal possibilities by the means
sketched in the previous three sections: we use the unschematized category of
actuality to think about God and everything posited by God’s will and we use the
unschematized category of possibility to think about everything God has the power to
posit. In doing so we come to represent intrinsically non-modal noumenal correlates
modally and we represent God as grounding all real noumenal possibilities (every-
thing we represent as a noumenal real possibility) in virtue of his powers; if God were
canceled, his powers would be canceled, so nothing would be really possible.
In the second passage quoted above Kant denies that we can use an “analogy” with
human power and will to represent God as the ground of all real possibility (Ak. 2:
152–3). Kant does not explicitly say why we cannot use this analogy in this fashion,
but a natural answer suggests itself: God’s essence is the ground of all real possibility,
including the possibility of his own powers (cf. Ak. 2: 125), so those powers
themselves cannot be the grounds of all real possibility.
But this worry about representing God as grounding all real possibility through his
powers disappears when we no longer take God to have modal properties ‘in himself.’
Once Kant has in place the doctrine that the divine intuitive intellect does not cognize
modal properties, it follows (by (P2) and (P3) from section 1) that neither God nor
any other noumena have modal properties ‘in themselves,’ so restricting attention to
God and noumena ‘in themselves,’ no question arises about the grounds of the
possibility of God’s powers. The source of noumenal modality is the content of our
thoughts about noumena, using the unschematized categories of possibility and
actuality. Now that God and noumena have no modal properties ‘in themselves’
(but only relative to our modalized thought about them) there is no barrier to thinking
that God grounds various noumena and their properties through his will, and has the
power to directly produce certain properties, and indirectly to produce powers to
produce other properties (e.g., the choices of noumenal wills), and this comes to be
represented by us as the relation of real grounds to real possibilities “through an
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 THE ANTINOMY OF KANT ’ S MODAL METAPHYSICS

analogy with action.” In Chapters 4 and 5 I argued that, although he claims we cannot
represent God’s grounding of all real possibility through his powers, Kant implicitly
relies on such a model in his argument for a unique GARP. I referred to this as the
‘tension’ within Kant’s pre-Critical modal metaphysics. My solution to the ‘antinomy’
of Kant’s Critical modal metaphysics also shows us, retrospectively, how to resolve the
tension: God’s powers are what we represent modally as the grounds of all real
possibility. The problem that arose from thinking that God grounds all real possibil-
ities through his powers—powers presuppose possibility—is dissolved, since possi-
bility (for noumena) is no longer something ‘worldly’ or ‘noumenal’ but the content of
our modalized thoughts about noumena. God’s powers do not ground real possibil-
ities for noumena ‘in themselves’ so to speak; we think of those powers modally and
thereby represent them as grounds of possibility of the very consequences they are the
powers to bring about.
We have already seen that the modal theory of Beweisgrund raises (implicitly and
explicitly) various questions in the metaphysics of modality, none of which receive
satisfactory answers in that work:
(i) How are really possible predicates given to a discursive intellect?
(ii) How can a discursive intellect know what is really possible?
(iii) What is the content of modal thought? I.e. what, in general, are we as discursive
intellects doing when we represent some predicate as really possible?
(iv) What are the real grounds of real possibility?
(v) Can we prove that there is a unique absolutely necessary real ground of real
possibility?
(vi) What is the real grounding relation between God and really possible predi-
cates? In the terms of Chapters 4 and 5, what are God’s fundamental
predicates and what is the derivation relation between them and derivative
really possible predicates?
We can now see that Kant’s Critical modal metaphysics constitutes a systematic set
of answers to all of these questions. His answers to (i) and (ii) constitute his positive
theory of a priori cognition (Chapter 6). His answer to (iii) constitutes his highly
general theory of (the unschematized category of) real possibility überhaupt (Chapters
7 and 9). His answer to (iv) constitutes his theory of various notions of real possibility
and their grounds, systematically unified by his answer to (iii) (Chapters 7, 8, and 9).
His answer to (v) constitutes his Critical reinterpretation of the epistemic status of the
argument in Beweisgrund: it is a subjectively necessary postulate of theoretical reason
(Chapter 9). And, finally, his answer to (vi) is that the divine intuitive intellect cognizes
no modal properties, so the relation between God and real possibility is a consequence
of our modalized representation of the relation between God and what he has the
power to do.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

Note on Sources and Key to

Abbreviations and Translations

Citations to the Critique of Pure Reason use the customary format of giving the page in the first
edition of 1781 (A), followed by the page in the second edition of 1787 (B) (e.g., A327/B384).
Citations to works of Kant other than the Critique of Pure Reason give the volume and page
number in the Academy edition, Kants gesammelte Schriften [Ak.], edited by the Berlin-
Brandenburg Academy of Sciences (Berlin: Walter de Gruyter, 1902– ). Unless context
makes it clear, I also refer to individual works by abbreviations of their English titles. Below
I list those abbreviations, their complete (German) title, which Ak. volume they are found in, as
well as dates of publication (for published works), of composition (for unpublished works),
and, in the case of Kant’s lectures, of the lectures themselves; where dates are uncertain
I indicate this with a question mark (and, in some cases, a range of possible dates). I also list
the translations from which I quote, often with slight modifications of my own (see Preface).
Where no published translation is listed, either translations are my own or the work is cited but
not quoted.
A Leibniz, G. W. (1923– ). Sämtliche Schriften und Briefe. Darmstadt: Berlin
Academy of the Sciences. Cited by series, volume, and page.
AG Leibniz, G. W. (1989). Philosophical Essays. R. Ariew and D. Garber (trans.).
Indianapolis: Hackett.
Ak. (1902– ). Kants gesammelte Schriften (vols. 1–29). Berlin-Brandenburg
(formerly: Royal Prussian) Academy of Sciences (eds.). Berlin: Walter de
Gruyter. Cited by volume and page number.
Anthr. Anthropologie in pragmatischer Hinsicht (1798). Ak. 7: 117–334.
AT Descartes, R. (1964–76). Oeuvres (vols. 1–12). C. Adam and P. Tannery (eds.).
Paris: Vrin/CNRS. Cited by volume and page number.
AzDM Anmerkungen zur Deutschen Metaphysik. In Werke (Wolf (1965– )), Abt. I, Bd. 3.
CJ Kritik der Urtheilskraft (1790). Ak. 5: 165–485.
Critique of the Power of Judgment, P. Guyer and E. Matthews (trans.). Kant
(2000).
Corr. Kants Briefwechsel. Ak. 10–13. Selected translations in Kant (1999).
CPR Kritik der reinen Vernunft (A:1781, B:1787). Ak. 3 (B) and 4: 1–252 (A). I have
also consulted Kant (1998b).
Critique of Pure Reason, P. Guyer and A. Wood (trans.). Kant (1998a). I have
also occasionally consulted the Kemp Smith translation, Kant (1929).
CPrR Kritik der praktischen Vernunft (1788). Ak. 5: 1–164.
Critique of Practical Reason, M. J. Gregor (trans.). In Kant (1996a).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOTE ON SOURCES

CSM Descartes, R. (1984–1991). The Philosophical Writings of Descartes (vols. 1–3),

J. Cottingham, R. Stoothoff, and D. Murdoch (trans. and ed.). Cambridge:
Cambridge University Press.
Danz.RT Danziger Rationaltheologie (1783/4), Ak. 28: 1226–1319.
De Usu Crusius, C. A. (1743). Dissertatio de usu et limitibus principii rationis
determinantis, vulgo sufficientis. In (1964). Die Philosophischen Hauptwerke,
G. Tonelli (ed.). Hildesheim: Georg Olms, Vol. 4.
Ausfürhliche Abhandlung von dem rechten Gebrauche und der Einschränkung
des sogenannten Satzes vom zureichenden oder besser determinierenden Gründe
(1744), Christian Friedrich Krausen (trans.). Leipzig. Cited by section number.
Disc. Über eine Entdeckung, nach der alle neue Kritik der reinen Vernunft durch eine
ältere entebehrlich gemacht werden soll (1790). Ak. 8: 185–252.
On a Discovery Whereby any New Critique of Pure Reason is to be Made
Superfluous by an Older One, H. Allison (trans.). In Kant (2002).
DM Leibniz, G.W. Discourse on Metaphysics. In AG 35–68.
DSR Leibniz, G. W. (1992). De Summa Rerum, G. H. R. Parkinson (trans. and ed.).
New Haven: Yale University Press.
Dt.Log. Wolff, C., Vernünftige Gedanken von den Kräften des menschlichen Verstandes
und ihrem richtigen Gebrauche im Erkenntnis der Warhheit. In Werke
(Wolff (1965– )), Abt. I, Bd. 1. Cited by chapter and paragraph number
(e.g., }3.3 refers to }3 of Chapter 3; Chapter 0 corresponds to ‘Vorbericht von
der Welt-Weisheit’.)
Dt.Met. Wolff, C., Verunünftige Gedancken von Gott, der Welt und der Seele des
Menschen, auch allen Dingen überhaupt. In Werke (Wolff (1965– )), Abt. I, Bd. 2.
E Erdmann, B. (1881). Nachträge zu Kants Kritik der reinen Vernunft. Kiel:
Lipsius & Tischer. Reproduced in Ak. 14–23. Cited by Roman numeral in
Erdmann’s edition and volume and page number in Ak.
Ent. Crusius, C. A. (1745). Entwurf der nothwendigen Verunft-Wahrheiten
(Leipzig). In (1964). Die Philosophischen Hauptwerke, G. Tonelli (ed.).
Hildesheim: Georg Olms, Vol. 2. Cited by section number.
GMM Grundlegung zur Metaphysik der Sitten (1785). Ak. 5: 385–464.
Groundwork of the Metaphysics of Morals, M. J. Gregor (trans.). In Kant
(1996a).
H.Werke Hegel, G.W.F. (1969–1971). Werke, E. Moldenhauer and K.M. Michel (eds.).
Frankfurt a.M.: Suhrkamp. 20 volumes. Cited by volume and page number.
ID De mundi sensibilis atque intelligibilis forma et principiis (1770). Ak. 2: 385–420.
[Inaugural Dissertation.]
On the Form and Principles of the Sensible and Intelligible World, D. Walford
and R. Meerbote (trans.). In Kant (1992a).
JL Logik, herausgegeben von Gottlob Benjamin Jäsche (1800). Ak. 9: 1–150.
J. M. Young (trans.). In Kant (1992b).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOTE ON SOURCES 

L Leibniz, G. W. (1969). Philosophical Papers and Letters, L. Loemker (trans. and

ed.). Dordrecht: D. Reidel.
LB Logik Blomberg (1771?). Ak. 24: 7–301. J. M. Young (trans.). In Kant (1992b).
LBu Logik Busolt (1789/90?). Ak. 24: 604–86.
LDW Logik Dohna-Wundlacken (1792), Ak. 24: 697–784. J. M. Young (trans.). In
Kant (1992b).
LF Gedanken von der wahren Schätzung der lebendigen Kräfte (1747). Ak. 1: 1–182.
Log. Wolff, C., Philosophia rationalis sive Logica. In Werke (Wolf (1965– )), Abt. II,
Bd. 1.2–3.
LP Logik Philippi (1772?). Ak. 24: 303–496.
LPö Logik Pölitz (1780?). Ak. 24: 499–602.
MD Metaphysik Dohna (1792–3). Ak. 28: 610–704.
Meta. Baumgarten, A. G. (1757). Metaphysica (4th edn.). Halle.
Baumgarten, A. G. (2013). Metaphysics: A Critical Translation with Kant’s
Elucidations, Selected Notes and Related Materials, C. Fugate and J. Hymers
(trans. and ed.). London: Bloomsbury.
MFNS Metaphysische Anfangsgründe der Naturwissenschaft (1786). Ak. 4: 465–566.
Metaphysical Foundations of Natural Science, M. Friedman (trans.). In Kant
(2002).
MH Metaphysik Herder (1762–4?). Ak. 28: 1–181. Selections trans. K. Ameriks and
S. Naragon in Kant (1997).
MK2 Metaphysik K2 (early 1790s). Ak. 28: 705–816. Selections trans. K. Ameriks and
S. Naragon in Kant (1997).
MK3 Metaphysik Vigilantius (1794–5). Ak. 28: 817–838, 941–1047. K. Ameriks and
S. Naragon (trans.). In Kant (1997).
ML1 Metaphysik L1 (1770–5?). Ak. 28: 167–250. Selections trans. K. Ameriks and
S. Naragon in Kant (1997).
ML2 Metaphysik L2 (1780?), Ak. 28: 531–604. K. Ameriks and S. Naragon (trans.). In
Kant (1997).
MM Metaphysik Mrongovius (1782/3), Ak. 29: 747–940. K. Ameriks and S. Naragon
(trans.). In Kant (1997).
Mon. Leibniz, G. W. (1714). Principes de philosophie ou Monadologie.
The Principles of Philosophy, or Monadology. AG 213–24.
MV Metaphysik Volckmann (1784/5). Ak. 28: 351–459.
MvS Metaphysik von Schön (1785–90?). Ak. 28: 460–524.
ND Principiorum primorum cognitionis metaphysicae nova delucidatio (1755). Ak.
1: 385–416.
New Elucidation of the First Principles of Metaphysical Cognition, D. Walford
and R. Meerbote (trans.). In Kant (1992a).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 NOTE ON SOURCES

NE Leibniz, G. W. (1704). Nouveaux Essais sur l’entendement humain.

Leibniz, G.W. (1966). New Essays on Human Understanding, P. Remnant and
J. Bennett (trans. and ed.). Cambridge: Cambridge University Press.
NG Wolff, C. Natürliche Gottesgelahrheit nach beweisender Lehrart abgefasst. In
Werke (Wolf (1965– )), Abt. 1, Bd. 23.1–4.
NM Versuch den Begriff von Negativen Grössen in der Philosophie einzuführen
(1763). Ak. 2: 165–204.
Attempt to Introduce the Concept of Negative Magnitudes into Philosophy,
D. Walford and R. Meerbote (trans.). In Kant (1992).
Ont. Wolff, C., Philosophia prima sive Ontologia . In Werke (Wolff (1965– )), Abt. 2,
Bd. 3.
OPG Der einzig Mögliche Beweisgrund zu einer Demonstration des Dasein Gottes
(1763). Ak. 2: 63–164. I have also consulted Kant (2011). [Beweisgrund]
The Only Possible Argument in Support of a Demonstration of the Existence of
God, D. Walford and R. Meerbote (trans. and ed.). In Kant (1992a).
P Leibniz, G. W. (1966). Logical Papers, G. H. R. Parkinson (trans. and ed.).
Oxford: Clarendon Press.
PE Vorlesungen über Philosophische Enzyklopädie. Ak. 29: 3–45.
Pöl.RT Philosophische Religionslehre nach Pölitz (1783–6?). Ak. 28: 989–1126.
Lectures on the Philosophical Doctrine of Religion, A. Wood (trans.). In Kant
(1996b).
Prize Untersuchung über die Deutlichkeit der Grundsätze der natürlichen Theologie
und Moral (1764). Ak. 2: 273–302.
Inquiry Concerning the Distinctness of the Principles of Natural Theology and
Morality, D. Walford and R. Meerbote (trans.). In Kant (1992a). [Prize Essay]
Prog. Welches sind die wirklichen Fortschritte, die die Metaphysik seit Leibnitzens und
Wolffs Zeiten in Deutschland gemacht hat? (1791/1804), Ak. 20: 253–332.
What Real Progress has Metaphysics Made in Germany since the Time of Leibniz
and Wolff?, P. Heath (trans.). In Kant (2002).
Prol. Prolegomena zu einer jeden künftigen Metaphysik die als Wissenschaft wird
auftreten können (1783). Ak. 4: 253–384.
Prolegomena to any Future Metaphysics that Will be Able to Present itself as a
Science, G. Hatfield (trans.). In Kant (2002).
Refl. Kants handschriftlicher Nachlaß. Ak. 14–23. Cited by four-digit number and
volume and page number. Also referred to as Reflexionen.
Rel. Die Religion innerhalb der Grenzen der bloßen Vernunft (1793). Ak. 6: 1–202.
Religion within the Boundaries of Mere Reason, G. di Giovanni (trans.). In Kant
(1996b).
Search Malebranche, N. (1980). The Search after Truth, M. Lennon and P. J. Olscamp
(trans. and ed.). Columbus, Ohio: Ohio University Press. Repr. (1997),
Cambridge: Cambridge University Press. Cited by Book, Part, and Chapter.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

NOTE ON SOURCES 

Th. Leibniz, G. W. (1710). Théodicée.

Leibniz, G. W. (1985). Theodicy, E. M. Huggard and A. Ferrer (trans. and ed.).
Chicago and La Salle, Ill.: Open Court.
TN Wolff, C., Theologia naturalis. In Werke (Wolff (1965– )). Abt. II, Bd. 7–8.
Tone Von einem neuerdings erhobenen vornehmen Ton in der Philosophie (1796). Ak.
8: 387–406.
On a Recently Prominent Tone of Superiority in Philosophy, P. Heath (trans.). In
Kant (2002).
Volck.RT Natürliche Theologie Volckmann (1783/4). Ak. 28: 1127–1225.
Weg Crusius, C. (1747). Weg zur Gewißheit und Zuverläßigkeit der menschlichen
Erkenntniß. In (1964). Die Philosophischen Hauptwerke, G. Tonelli (ed.).
Hildesheim: Georg Olms, Vol. 3. Cited by section number.
Werke Wolff, C. (1965– ). Gesammelte Werke, J. École, H. W. Arndt, Ch. A. Corr,
J. E. Hoffmann, and M. Thomann (eds.). Hildesheim, Zürich, and New York:
Georg Olms. Cited by division (Abt.) and volume (Bd.).
WL Wiener Logik (1780). Ak. 24: 785–940. J. M. Young (trans.). In Kant (1992b).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

Bibliography

Works of Kant
(1902– ). Kants gesammelte Schriften (vols. 1–29). Berlin-Brandenburg (formerly: Royal
Prussian) Academy of Sciences (eds.). Berlin: Walter de Gruyter.
(1929). Immanuel Kant’s Critique of Pure Reason, N. Kemp Smith (trans.). London: Macmillan.
(1976). Prolegomena; zu einer jeden künftigen Metaphysik, die als Wissenschaft wird auftreten
können, K. Vorländer (ed.). Hamburg: Felix Meiner.
(1977). Prolegomena to any Future Metaphysics, P. Carus and J. Ellington (trans.). New York:
Hackett.
(1992a). Theoretical Philosophy, 1775–1770, D. Walford and R. Meerbote (trans. and ed.).
Cambridge: Cambridge University Press.
(1992b). Lectures on Logic, J. M. Young (trans. and ed.). Cambridge: Cambridge University Press.
(1996a). Practical Philosophy, M. J. Gregor (trans. and ed.). Cambridge: Cambridge University
Press.
(1996b). Religion and Rational Theology, A. Wood and G. di Giovanni (trans. and ed.).
Cambridge: Cambridge University Press.
(1997). Lectures on Metaphysics, K. Ameriks and S. Naragon (trans. and ed.). Cambridge:
Cambridge University Press.
(1998a). Critique of Pure Reason, P. Guyer and A. Wood (trans. and ed.). Cambridge:
Cambridge University Press.
(1998b). Kritik der reinen Vernunft. Nach der 1. And 2. Original Ausgabe, J. Timmerman (ed.).
Hamburg: Felix Meiner.
(1999). Correspondence, A. Zweig (trans. and ed.). Cambridge: Cambridge University Press.
(2000). Critique of the Power of Judgment, P. Guyer and E. Matthews (trans. and ed.).
Cambridge: Cambridge University Press.
(2002). Theoretical Philosophy after 1781, H. Allison and P. Heath (trans. and ed.). Cambridge:
Cambridge University Press.
(2011). Der einzig mögliche Beweisgrund zu einer Demonstration des Dasein Gottes. Historisch-
kritische Edition, L. Kreimendahl and M. Oberhausen (eds.). Hamburg: Felix Meiner.

Other Primary Sources

Baumgarten, A. G. (1757). Metaphysica (4th edn.). Halle. Repr. in Ak. 17.

Baumgarten, A. G. (1783). Metaphysik (2nd German edn.), G. F. Meier and J. E. Eberhard (trans.
and ed.). Halle. Repr. (2004). Metaphysik, D. Mirbach (ed.). Jena: Dietrich Schlegelmann.
Baumgarten, A. G. (2011). Metaphysica = Metaphysik: historisch-kritische Ausgabe, G. Gawlick
and L. Kreimendahl (trans.). Stuttgart: Fromann-Holzboog.
Baumgarten, A. G. (2013). Metaphysics: A Critical Translation with Kant’s Elucidations,
Selected Notes and Related Materials, C. Fugate and J. Hymers (trans. and ed.). London:
Bloomsbury.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 BIBLIOGRAPHY

Crusius, C. A. (1964). Die philosophischen Hauptwerke (vols. 1–4), G. Tonelli (ed.). Hildes-
heim: Georg Olms.
Descartes, R. (1964–76). Oeuvres (vols. 1–12), C. Adam and P. Tannery (eds.). Paris: Vrin/
CNRS.
Descartes, R. (1984–1991). The Philosophical Writings of Descartes (vols. 1–3), J. Cottingham,
R. Stoothoff, and D. Murdoch (trans. and ed.). Cambridge: Cambridge University Press.
Eberhard, J. E. (1789). Über die Unterscheidung der Urteile in analytische und synthetische.
Philosophisches Magazin, 1: 307–32.
Hegel, G. W. F. (1969–71). Werke (vol. 1–20), E. Moldenhauer and K.M. Michel (eds.).
Frankfurt a.M.: Suhrkamp.
Leibniz, G. W. (1969). Philosophical Papers and Letters, L. Loemker (trans. and ed.). 2nd
edition. Dordrecht: D. Reidel.
Leibniz, G. W. (1985). Theodicy, E. M. Huggard and A. Farrer (trans. and ed.). Chicago and La
Salle, Ill.: Open Court.
Leibniz, G. W. (1989). Philosophical Essays, R. Ariew and D. Garber (trans. and ed.). Indian-
apolis: Hackett.
Leibniz, G. W. (1996). New Essays on Human Understanding, P. Remnant and J. Bennett
(trans. and ed.). Cambridge: Cambridge University Press.
Leibniz, G. W. (2008). Protogaea, C. Cohen and A. Wakefield (trans. and ed.). Chicago:
University of Chicago Press.
Maaß, J. G. (1789). Über den höchsten Grundsatz der synthetischen Urtheile, in Beziehung auf
die Theorie von der mathematischen Gewißheit. Philosophisches Magazin, 2(2): 186–231.
Malebranche, N. (1980). The Search after Truth, M. Lennon and P. J. Olscamp (trans. and ed.).
Columbus, Ohio: Ohio University Press. Repr. (1997). Cambridge: Cambridge University Press.
Malebranche, N. (1997). Dialogues on Metaphysics and on Religion, N. Jolley and D. Scott
(trans. and ed.). Cambridge: Cambridge University Press.
Mendelssohn, M. (1997). Philosophical Writings, D. Dahlstrom (trans. and ed.). Cambridge:
Cambridge University Press.
Suárez, F. (1983). On the Essence of Finite Being as Such, on the Existence of that Essence and
their Distinction, N. Wells (trans. and ed.). Milwaukee, Wisc.: Marquette University Press.
Valentini, M. B. (1709). Armamentarium artis & naturae. Gissae-Hassorum.
Warda, A. (1922). Immanuel Kants Bücher. Berlin: Martin von Breslauer.
Watkins, E. (2009). Kant’s Critique of Pure Reason: Background Source Materials. Cambridge:
Cambridge University Press.
Wolff, C. (1965– ). Gesammelte Werke, J. École, H.W. Arndt, Ch. A. Corr, J. E. Hoffmann, and
M. Thomann (eds.). Hildesheim, Zürich, and New York: Georg Olms.
Wolff, C. (2005). Erste Philosophie oder Ontologie }}1–78, D. Effertz (trans. and ed.). Hamburg:
Felix Meiner.

Secondary Literature
Abaci, U. (2008). Kant’s Theses on Existence. British Journal for the History of Philosophy, 16
(3): 559–93.
Abaci, U. (2013). The Coextensiveness Thesis and Kant’s Modal Agnosticism in the ‘Postu-
lates.’ European Journal of Philosophy 23(2).
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

BIBLIOGRAPHY 

Abaci, U. (2014). Kant’s Only Possible Argument and Chignell’s Real Harmony. Kantian
Review, 19(1): 1–25.
Adams, R. M. (1974). Theories of Actuality. Noûs, 8(3): 211–31.
Adams, R. M. (1994). Leibniz: Determinist, Idealist, Theist. Oxford: Oxford University Press.
Adams, R. M. (1997). Things in Themselves. Philosophy and Phenomenological Research, 57
(4): 801–25.
Adams, R. M. (2000). God, Possibility and Kant. Faith and Philosophy, 17(4): 425–40.
Adickes, E. (1924). Kant und das Ding an sich. Berlin: Pan Verlag Rolf Heisse. Repr. (1977).
Hildesheim: Georg Olms.
Allais, L. (2004). Kant’s One World: Interpreting ‘Transcendental Idealism.’ The British
Journal for the History of Philosophy, 12(4): 655–84.
Allais, L. (2006). Intrinsic Natures: A Critique of Langton on Kant. Philosophy and Phenom-
enological Research, 73(1): 144–69.
Allais, L. (2007). Kant’s Idealism and the Secondary-Quality Analogy. Journal of the History of
Philosophy, 45(3): 459–84.
Allais, L. (2009). Kant, Non-conceptual Content, and the Representation of Space. Journal of
the History of Philosophy, 47(3): 383–413.
Allais, L. (2015). Manifest Reality: Kant’s Idealism and his Realism. Oxford: Oxford University
Press.
Allison, H. (1973). The Kant–Eberhard Controversy. Baltimore: Johns Hopkins University
Press.
Allison, H. (1983). Kant’s Transcendental Idealism: An Interpretation and Defense. New
Haven: Yale University Press.
Allison, H. (1994). Causality and Causal Laws in Kant: A Critique of Michael Friedman. In
P. Parrini (ed). Kant and Contemporary Epistemology. Dordrecht: Kluwer, 291–307.
Allison, H. (2000). Is the Critique of Judgment ‘Post-Critical?’. In S. Sedgwick (ed.). The
Reception of Kant’s Critical Philosophy: Fichte, Schelling, and Hegel. Cambridge: Cambridge
University Press, 78–92.
Allison, H. (2004). Kant’s Transcendental Idealism: An Interpretation and Defense. 2nd edn.
New Haven: Yale University Press.
Allison, H. (2012). Essays on Kant. Oxford: Oxford University Press.
Ameriks, K. (1981). Kant’s Deduction of Freedom and Morality. Journal of the History of
Philosophy, 19(1): 53–79.
Ameriks, K. (1982a). Recent Work on Kant’s Theoretical Philosophy. American Philosophical
Quarterly, 19(1): 1–24.
Ameriks, K. (1982b). Kant’s Theory of Mind: An Analysis of the Paralogisms of Pure Reason.
Oxford: Clarendon Press.
Ameriks, K. (1992). The Critique of Metaphysics: Kant and Traditional Ontology. In P. Guyer
(ed.). The Cambridge Companion to Kant. Cambridge: Cambridge University Press, 249–79.
Ameriks, K. (2003). Kant and Short Arguments to Humility. In Interpreting Kant’s Critiques.
Oxford: Oxford University Press, 135–57.
Anderson, E. (1994). Kant, Natural Kind Terms, and Scientific Essentialism. History of
Philosophy Quarterly, 11(4): 355–73.
Anderson, R. L. (2004). It Adds Up After All: Kant’s Philosophy of Arithmetic in Light of the
Traditional Logic. Philosophy and Phenomenological Research, 69(3): 501–40.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 BIBLIOGRAPHY

Aquila, R. (1979). Things in Themselves and Appearances: Intentionality and Reality in Kant.
Archiv für Geschichte der Philosophie, 61(3): 293–308.
Aquila, R. (1983). Representational Mind: A Study of Kant’s Theory of Knowledge. Blooming-
ton, Ind.: Indiana University Press.
Armstrong, D. (1989). Universals: An Opinionated Introduction. Boulder, Colo.: Westview
Press.
Ariew, R. (1998). Leibniz on the Unicorn and Various Other Curiosities. Early Modern Science
and Medicine, 3(4): 267–88.
Audi, P. (2012a). Grounding: Toward a Theory of the In-Virtue-Of Relation. Journal of
Philosophy, 109(12): 685–711.
Audi, P. (2012b). A Clarification and Defense of the Notion of Grounding. In F. Correia
and B. Schnieder (eds.). Metaphysical Grounding: Understanding the Structure of Reality.
Cambridge: Cambridge University Press, 101–21.
Ayer, A. J. (1936). Language, Truth, and Logic. London: Gollancz.
Bader, R. M. (2009). Kant and the Categories of Freedom. British Journal for the History of
Philosophy, 17(4): 799–820.
Baumgardt, D. (1915). Das Möglichkeitsproblem der Kritik der reinen Vernunft, der modernen
Phänomenologie und der Gegenstandstheorie. Kant-Studien Ergänzungsheft 51. Berlin:
Walter de Gruyter.
Barnes, J. (1972). The Ontological Argument. New York: St Martin’s Press.
Beck, L. W. (1960). A Commentary on Kant’s Critique of Practical Reason. Chicago: University
of Chicago Press.
Beck, L. W. (1967). Can Kant’s Synthetic Judgments Be Made Analytic? In M. Gram (ed.).
Kant: Disputed Questions. Chicago: Quadrangle Books, 228–46.
Beck, L. W. (1969). German Philosophy before Kant. Cambridge, Mass.: Harvard University Press.
Beck, L.W. (1978). A Prussian Hume and a Scottish Kant. In Essays on Kant and Hume. New
Haven: Yale University Press, 111–29.
Beck, L. W. (1989). Two Ways of Reading Kant’s Letter to Herz: Comments on Carl. In E. Förster
(ed.). Kant’s Transcendental Deductions. Stanford: Stanford University Press, 21–6.
Beiser, F. (2002). German Idealism: The Struggle against Subjectivism, 1781–1801. Cambridge,
Mass.: Harvard University Press.
Bird, A. (2005). Laws and Essences. Ratio, 18(4): 437–61.
Bird, G. (1962). Kant’s Theory of Knowledge. London: Routledge & Kegan Paul.
Bissinger, A. (1970). Die Struktur der Gotteserkenntnis: Studien zur Philosophie Christian
Wolffs. Bonn: H. Bouvier.
Bliss, R. and Trogdon, K. (2014). Metaphysical Grounding. The Stanford Encyclopedia of Philoso-
phy. E.N. Zalta (ed.). <http://plato.stanford.edu/archives/win2014/entries/grounding/>.
Blumenfeld, D. (1973). Leibniz’s Theory of the Striving Possibles. Studia Leibnitiana, 5(2):
163–77.
Boghossian, P. (1996). Analyticity Reconsidered. Noûs, 30(3): 360–91.
Brandt, R. (1991). Die Urteilstafel: Kritik der reinen Vernunft A67–76/B92–201. Hamburg:
Felix Meiner.
Breazeale, D. (1998). Fichte’s nova methodo phenomenologica: On the Methodological Role of
‘Intellectual Intuition’ in the Later Jena Wissenschaftslehre. Revue internationale de philoso-
phie, 52(206): 587–616.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

BIBLIOGRAPHY 

Brewer, K. and Watkins, E. (2012). A Difficulty Still Awaits: Kant, Spinoza, and the Threat of
Theological Determinism. Kant-Studien, 103(2): 163–87.
Brittan, G. (1978). Kant’s Philosophy of Science. Princeton: Princeton University Press.
Brogaard, B. and Salerno, J. (2013). Remarks on Counterpossibles. Synthèse, 19(4): 639–60.
Buchdahl, G. (1965). Causality, Causal Laws and Scientific Theory in the Philosophy of Kant.
British Journal for the Philosophy of Science, 16(63): 187–208.
Buchdahl, G. (1969). The Kantian ‘Dynamic of Reason’, with Special Reference to the Place of
Causality in Kant’s System. In L. W. Beck (ed.). Kant Studies Today. La Salle, Ill.: Open
Court, 341–74.
Buchdahl, G. (1986). Kant’s ‘Special Metaphysics’ and The Metaphysical Foundations of
Natural Science. In R. Butts (ed.). Kant’s Philosophy of Physical Science: Metaphysische
Anfangsgründe der Naturwissenschaft 1786–1986. Dordrecht: D. Reidel, 127–62.
Buchdahl, G. (1992). Kant and the Dynamics of Reason. Cambridge, Mass.: Blackwell.
Burge, T. (1988). Individualism and Self-Knowledge. Journal of Philosophy, 85(11): 649–63.
Burgess, J. (2013). Saul Kripke. Cambridge: Polity.
Butts, R. (1986). The Methodological Structure of Kant’s Metaphysics of Science. In R. Butts
(ed.). Kant’s Philosophy of Physical Science: Metaphysische Anfangsgründe der Naturwis-
senschaft 1786–1986. Dordrecht: D. Reidel, 163–200.
Byrne, A. and Hilbert, D. R. (1997). Readings on Color, Vol. 1: The Philosophy of Color.
Cambridge, Mass.: MIT Press.
Carl, W. (1989a). Der schweigende Kant: die Entwürfe zu einer Deduktion der Kategorien vor
1781. Göttingen: Vandenhoek & Ruprecht.
Carl, W. (1989b). Kant’s First Draft of the Deduction of the Categories. In E. Förster (ed.).
Kant’s Transcendental Deductions. Stanford: Stanford University Press, 3–20.
Cassirer, E. (1981). Kant’s Life and Thought. (James Haden, trans.). New Haven: Yale University
Press.
Casullo, A. (2002). A Priori Knowledge. In P. Moser (ed.). The Oxford Handbook of Epistem-
ology. Oxford: Oxford University Press, 95–143.
Chalmers, D., Manley, R., and Wasserman, R. (eds.). (2009). Metametaphysics. Oxford: Oxford
University Press.
Chignell, A. (2003). Kant’s Ethics of Assent: Knowledge and Belief in the Critical Philosophy.
Unpublished Ph.D dissertation. Cornell University.
Chignell, A. (2007). Belief in Kant. Philosophical Review, 116(3): 323–60.
Chignell, A. (2009a). Kant, Modality, and the Most Real Being. Archiv für Geschichte der
Philosophie, 91(2): 157–92.
Chignell, A. (2009b). Are Supersensibles Really Possible? The Evidential Role of Symbols. In
V. Rhoden, T. Terra, and G. Almeida (eds.). Recht und Frieden in der Philosophie Kants.
Berlin: Walter de Gruyter, vol. 4, 99–110.
Chignell, A. (2010). Real Repugnance and Belief about Things-in-Themselves: A Problem and
Kant’s Three Solutions. In J. Krueger and B. B. Lipscomb (eds.). Kant’s Moral Metaphysics.
Berlin: Walter de Gruyter, 177–210.
Chignell, A. (2011). Real Repugnance and Our Ignorance of Things-in-Themselves: A
Lockean Problem in Kant and Hegel. Internationales Jahrbuch des Deutschen Idealismus,
7: 135–59.
Chignell, A. (2012). Kant, Possibility, and the Threat of Spinoza. Mind, 121(483): 635–75.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 BIBLIOGRAPHY

Chignell, A. (2014a). Kant and the ‘Monstrous’ Ground of Possibility: A Reply to Abaci and
Yong. Kantian Review, 19(1): 53–69.
Chignell, A. (2014b). Modal Motivations for Noumenal Ignorance: Knowledge, Cognition, and
Coherence. Kant-Studien 105(4): 573–597.
Cicovacki, P. (1991). An Aporia of A priori Knowledge. Kant-Studien, 82(3): 349–60.
Cohen, C. (2002). The Fate of the Mammoth. Chicago: University of Chicago Press.
Corr, E. H. (1973). The Existence of God, Natural Theology and Christian Wolff. International
Journal for Philosophy of Religion, 4(2): 105–18.
Correia, F. and Schnieder, B. (eds.) (2012). Metaphysical Grounding: Understanding the
Structure of Reality. Cambridge: Cambridge University Press.
Cover, J. and O’Leary-Hawthorne, J. (1999). Substance and Individuation in Leibniz. Cam-
bridge: Cambridge University Press.
Cunning, David (2008). Descartes’ Modal Metaphysics. The Stanford Encyclopedia of Philosophy
(Winter 2008 edn.). E.N. Zalta (ed.). <http://plato.stanford.edu/archives/spr2014/en
tries/descartes-modal/>.
Curley, E. (1984). Descartes on the Creation of the Eternal Truths. Philosophical Review, 93(4):
569–97.
Dell’Oro, R. (1994). From Existence to the Ideal: Continuity and Development in Kant’s
Theology. New York: Peter Lang.
De Vleeschauwer, H. J. (1962). The Development of Kantian Thought, trans. A. R. C. Duncan.
Edinburgh: Thos. Nelson.
Di Bella, S. (2005). The Science of the Individual. Dordrecht: Kluwer.
Di Sala, G. (1990). Kant und die Frage nach Gott. Kant-Studien Ergänzungsheft 122. Berlin:
Walter de Gruyter.
Divers, J. (2002). Possible Worlds. New York: Routledge.
Dummett, M. (1983). Could There Be Unicorns? In The Seas of Language. Oxford: Oxford
University Press, 328–48.
Düsing, K. (1986). Die Teleologie in Kants Weltbegriff. Kant-Studien Ergänzungsheft 80. Bonn:
H. Bouvier.
Easton, P. (2009). What is at Stake in the Cartesian Debates on the Eternal Truths? Philosophy
Compass, 4(2): 348–62.
École, J. (1990). La Métaphysique de Christian Wolff. In Wolff, C. Werke, Abt. III, Bd. 12.1–2.
Ellis, B. and Lierse, C. (1994). Dispositional Essentialism. Australasian Journal of Philosophy,
72(1): 27–45.
Falkenstein, L. (1995). Kant’s Intuitionism: A Commentary on the Transcendental Aesthetic.
Toronto: University of Toronto Press.
Ferrari, J. (1998). Das Ideal der reinen Vernunft. In G. Mohr and M. Willaschek (eds.).
Immanuel Kant, Kritik der reinen Vernunft. Berlin: Academie Verlag, 491–523.
Fine, K. (1994a). Essence and Modality. Philosophical Perspectives, 8: 1–16.
Fine, K. (1994b). Compounds and Aggregates. Noûs, 28(2): 137–58.
Fine, K. (1995). Senses of Essence. In W. Sinnott-Armstrong, D. Raffman, and N. Asher (eds.).
Modality, Morality and Belief: Essays in Honor of Ruth Barcan Marcus. Cambridge: Cambridge
University Press, 53–73.
Fine, K. (1999). Things and Their Parts. Midwest Studies in Philosophy, 23(1): 61–74.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

BIBLIOGRAPHY 

Forgie, J. W. (2007). Kant and Gassendi on Existence. Journal of the History of Philosophy, 45
(4): 511–23.
Förster, E. (2011). Die 25 Jahre der Philosophie. Frankfurt a.M.: Vittorio Klostermann.
Friedman, M. (1986). The Metaphysical Foundations of Newtonian Science. In R. Butts (ed.).
Kant’s Philosophy of Physical Science: Metaphysische Anfangsgründe der Naturwissenschaft
1786–1986. Dordrecht: D. Reidel, 25–60.
Friedman, M. (1992a). Kant and the Exact Sciences. Cambridge, Mass.: Harvard University
Press.
Friedman, M. (1992b). Causal Laws and the Foundation of Natural Science. In P. Guyer (ed.).
The Cambridge Companion to Kant. Cambridge: Cambridge University Press, 161–99.
Friedman, M. (2013). Kant’s Construction of Nature: A Reading of the Metaphysical Founda-
tions of Natural Science. Cambridge: Cambridge University Press.
Garber, D. (1985). Leibniz and the Foundations of Physics: The Middle Years. In K. Okruhlik
and J. R. Brown (eds.). The Natural Philosophy of Leibniz. Dordrecht: D. Reidel, 27–130.
Garber, D. (2005). Leibniz: Physics and Philosophy. In Donald Rutherford (ed.). The Cambridge
Companion to Leibniz. Cambridge: Cambridge University Press, 270–352.
Garber, D. (2009). Leibniz: Body, Substance, Monad. Oxford: Oxford University Press.
Gardner, S. (1999). Kant and the Critique of Pure Reason. New York: Routledge, 1999.
Gibson, M. (2011). A Revolution in Method, Kant’s ‘Copernican Hypothesis,’ and the Neces-
sity of Natural Laws. Kant-Studien, 102(1): 1–21.
Gloy, K. (1976). Die Kantische Theorie der Naturwissenschaft: eine Strukturanalyse ihrer
Möglichkeit, ihres Umfangs und ihrer Grenzen. Berlin: Walter de Gruyter.
Goodman, N. (1954). Fact, Fiction, and Forecast. London: Athlone Press.
Gram, M. S. (1981). Intellectual Intuition: The Continuity Thesis. Journal of the History of
Ideas, 42(2), 287–304.
Grier, M. (2001). Kant’s Doctrine of Transcendental Illusion. Cambridge: Cambridge Univer-
sity Press.
Griffin, M. (2013). Leibniz, God, and Necessity. Cambridge: Cambridge University Press.
Grüne, S. (2009). Blinde Anschauung: die Rolle von Begriffen in Kants Theorie sinnlicher
Synthesis. Frankfurt a.M.: Vittorio Klostermann.
Grüne, S. (2011). Is There a Gap in Kant’s B Deduction? International Journal of Philosophical
Studies, 19(3): 465–90.
Guyer, P. (1987). Kant and the Claims of Knowledge. Cambridge: Cambridge University Press.
Guyer, P. (1990a). Reason and Reflective Judgment: Kant on the Significance of Systematicity.
Noûs, 24(1): 17–43.
Guyer, P. (1990b). Kant’s Conception of Empirical Law. Proceedings of the Aristotelian Society,
64: 221–42.
Guyer, P. (ed.). (1992a). The Cambridge Companion to Kant. Cambridge: Cambridge Univer-
sity Press.
Guyer, P. (1992b). The Transcendental Deduction of the Categories. In P. Guyer (ed.). The
Cambridge Companion to Kant. Cambridge: Cambridge University Press, 123–60.
Guyer, P. (1998). The Postulates of Empirical Thinking in General and the Refutation of
Idealism. In G. Mohr and M. Willaschek (eds.). Immanuel Kant: Kritik der reinen Vernunft.
Berlin: Akademie Verlag, 297–324.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 BIBLIOGRAPHY

Guyer, P. (2003). Kant on the Systematicity of Nature: Two Puzzles. History of Philosophy
Quarterly, 20(3): 277–95.
Guyer, P. (2006). Kant. New York: Routledge.
Guyer, P. (ed.). (2010). The Cambridge Companion to Kant’s Critique of Pure Reason. Cambridge:
Cambridge University Press.
Hanna, R. (1998). A Kantian Critique of Scientific Essentialism. Philosophy and Phenomeno-
logical Research, 58(3): 497–528.
Hanna, R. (2001). Kant and the Foundations of Analytic Philosophy. Oxford: Oxford University
Press.
Hanna, R. (2006). Kant, Science, and Human Nature. Oxford: Oxford University Press.
Hanna, R. (2008). Kantian Non-conceptualism. Philosophical Studies, 137(1): 41–64.
Hanna, R. (2011). Kant’s Non-Conceptualism, Rogue Objects, and the Gap in the B Deduction.
International Journal of Philosophical Studies, 19(3): 399–415.
Hardin, C. L. (1988). Color for Philosophers. Indianapolis: Hackett.
Hartmann, N. (1938). Möglichkeit und Wirklichkeit. Berlin: Walter de Gruyter.
Hawthorne, J. (2001). Causal Structuralism. Philosophical Perspectives, 15: 361–78.
Heimsoeth, H. (1956). Metaphysik und Kritik bei. Chr. A. Crusius. Kant-Studien Ergänzung-
sheft 71. Cologne: Kölner Üniversitats-Verlag.
Heidegger, M. (1975). Die Grundprobleme der Phänomenologie. Frankfurt a.M.: Vittorio
Klostermann.
Henrich, D. (1967). Kants Denken 1762/3: Über den Ursprung der Unterscheidung analy-
tischer und synthetischer Urteile. In H. Heimsoeth, D. Henrich, and G. Tonelli (eds.).
Studien zu Kants philosophischer Entwicklung. Hildesheim: Olms, 9–38.
Henrich, D. (1968–9). The Proof-structure of Kant’s Transcendental Deduction. Review of
Metaphysics, 22(4): 640–59.
Henrich, D. (1976). Identität und Objektivität: eine Untersuchung über Kants transzendentale
Deduktion. Heidelberg: C. Winter.
Hintikka, J. (1969). On Kant’s Notion of Intuition. In T. Penelhum and J. J. MacIntosh (eds.).
The First Critique: Reflections on Kant’s Critique of Pure Reason. California: Wadsworth
Publishing Company, 38–53.
Hoffer, Noam. (forthcoming). Was Kant a Spinozist? Kantian Review.
Hogan, D. (2005). Rationalism and Causal Realism in Kant’s Metaphysics. Unpublished Ph.D
Dissertation. Yale University.
Hogan, D. (2009a). Three Kinds of Rationalism and the Non-Spatiality of Things in Them-
selves. Journal of the History of Philosophy, 47(3): 355–82.
Hogan, D. (2009b). Noumenal Affection. Philosophical Review, 118(4): 501–32.
Hogan, D. (2009c). How to Know Unknowable Things in Themselves. Noûs, 43(1): 49–63.
Hogan, D. (2013). Metaphysical Motives of Kant’s Analytic-Synthetic Distinction. Journal of
the History of Philosophy, 51(2): 267–307.
Hogan, D. (2014). Kant on Foreknowledge of Contingent Truths. Res Philosophica, 19(1):
47–70.
Honnefelder, L. (1990). Scientia transcendens: die formale Bestimmtheit der Seiendheit und
Realität in der Metaphysik des Mittelalters und der Neuzeit. Hamburg: Felix Meiner.
Howell, R. (1973). Intuition, Synthesis, and Individuation in the Critique of Pure Reason. Noûs,
7(3): 207–32.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

BIBLIOGRAPHY 

Insole, C. (2011). Intellectualism, Relational Properties and the Divine Mind in Kant’s Pre-
critical Philosophy. Kantian Review, 16 (3): 399–427.
Jago, M. (2014). The Impossible: An Essay on Hyperintensionality. Oxford: Oxford University
Press.
Jolley, N. (1990). The Light of the Soul: Theories of Ideas in Leibniz, Malebranche, and
Descartes. Oxford: Clarendon Press.
Kain, P. (forthcoming). The Development of Kant’s Conception of Divine Freedom. In B. Look
(ed.). Leibniz and Kant. Oxford: Oxford University Press.
Kaufman, D. (2002). Descartes’s Creation Doctrine and Modality. Australasian Journal of
Philosophy, 80(1): 24–41.
Kauppi, R. (1960). Über die Leibnizsche Logik. Acta Philosophica Fennica, XII, 243–67.
Kemp Smith, N. (1918). A Commentary to Kant’s Critique of Pure Reason. London: Macmillan.
Repr. (1984). Atlantic Highlands, NJ: Humanities Press.
Kenny, A. (1968). Descartes: A Study of His Philosophy. New York: Random House.
Kitcher, P. (1975). Kant and the Foundations of Mathematics. Philosophical Review, 84: 23–50.
Kitcher, P. (1986). Projecting the Order of Nature. In R. Butts (ed.). Kant’s Philosophy of
Physical Science: Metaphysische Anfangsgründe der Naturwissenschaft 1786–1986. Dor-
drecht: D. Reidel, 201–37.
Kitcher, P. (1993). Kant’s Transcendental Psychology. New York: Oxford University Press.
Kment, B. (2006). Counterfactuals and the Analysis of Necessity. Philosophical Perspectives,
20(1): 237–302.
Kohl, M. (2015). Kant on the Inapplicability of the Categories to Things in Themselves. British
Journal for the History of Philosophy, 23(1): 1–25.
Korcz, K.A. (2010). The Epistemic Basing Relation. The Stanford Encyclopedia of Philosophy
(Spring 2010 edn.). E. N. Zalta (ed.). <http://plato.stanford.edu/archives/spr2010/en
tries/basing-epistemic/>.
Koslicki, K. (forthcoming). The Coarse-Grainedness of Grounding. In K. Bennett and
D. Zimmerman (eds.). Oxford Studies in Metaphysics. Oxford: Oxford University Press.
Kreimendahl, L. (1990). Kant: der Durchbruch von 1769. Cologne: J. Dinter.
Kreines, J. (2009). Kant on the Laws of Nature: Laws, Necessitation, and the Limits of our
Knowledge. European Journal of Philosophy, 17(4): 527–58.
Kripke, S. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press.
Kroon, F. and Nola, E. (1987). Kant, Kripke, and Gold. Kant-Studien, 78(4): 442–58.
Kuehn, M. (2001). Kant: A Biography. Cambridge: Cambridge University Press.
Laerke, M. (2008). Leibniz lecteur de Spinoza: la genèse opposition complexe. Paris: Champion.
Land, Thomas (2013). Intuition and Judgment: How Not to Think about the Singularity of
Intuition. In S. Bacin, A. Ferrarin, C. LaRocca, M. Ruffing (eds.). Akten des XI. Internatio-
nalen Kant-Kongresses. Berlin: Walter de Gruyter, vol. 2, 221–32.
Lange, M. (2009). Laws and Lawmakers: Science, Metaphysics, and the Laws of Nature. Oxford:
Oxford University Press.
Langton, R. (1998). Kantian Humility. Oxford: Oxford University Press.
Lapointe, S. (2012). Kant, Bolzano, and the Kantian Logicians. Grazer Philosophische Studien,
85(1): 11–32.
Laywine, A. (1993). Kant’s Early Metaphysics and the Origins of the Critical Philosophy. North
American Kant Society Studies in Philosophy, vol. 3. Atascadero, Calif.: Ridgeview.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 BIBLIOGRAPHY

Leech, J. (2012). Kant’s Modalities of Judgment. European Journal of Philosophy, 20(2): 260–84.
Leech, J. (2014). Making Modal Distinctions: Kant on the Possible, the Actual, and the Intuitive
Understanding. Kantian Review, 19(3): 339–65.
Lewis, D. (1973). Counterfactuals. Oxford: Basil Blackwell.
Lewis, D. (1983). New Work for a Theory of Universals. Australian Journal of Philosophy 61,
343–77. Repr. in D. Lewis (1999). Papers in Metaphysics and Epistemology. New York:
Cambridge University Press, 8–55.
Lewis, D. (1986). On the Plurality of Worlds. Oxford: Blackwell Publishers. Cited by page
numbers in 2nd edition (2001).
Lin, M. (2012). Rationalism and Necessitarianism. Noûs, 46(3): 418–48.
Logan, I. (2007). Whatever Happened to Kant’s Ontological Argument? Philosophy and
Phenomenological Research, 74(2): 346–63.
Longuenesse, B. (1998). Kant and the Capacity to Judge. Princeton: Princeton University Press.
Longuenesse, B. (2000). Point of View of Man or Knowledge of God: Kant and Hegel on
Concept, Judgment, and Reason. In S. Sedgwick (ed.). The Reception of Kant’s Critical
Philosophy: Fichte, Schelling, and Hegel. Cambridge: Cambridge University Press, 253–82.
Longuenesse, B. (2005). The Transcendental Ideal and the Unity of the Critical System. In
B. Longuenesse. Kant and the Human Standpoint. New York: Cambridge University Press, 210–35.
Look, B. (2005). Leibniz and the Shelf of Essences. Leibniz Review, 15: 27–47.
Malcolm, N. (1960). Anselm’s Ontological Arguments. Philosophical Review, 69(1): 41–62.
Marshall, C. (2013). Kant’s Appearances and Things in Themselves as Qua-Objects.
The Philosophical Quarterly, 63(252): 520–45.
Mates, B. (1986). The Philosophy of Leibniz: Metaphysics and Language. New York: Oxford
University Press.
McDaniel, K. (2013). A Philosophical Model of the Relation between Things in Themselves
and Appearances. Noûs, 48(4): 1–22.
McFarlane, J. (2002). Kant, Frege, and the Logic in Logicism. Philosophical Review, 111(1):
25–65.
McLear, C. (2011). Kant on Animal Consciousness. Philosophers’ Imprint, 11(15): 1–16.
McLear, C. (2014). The Kantian (Non)-conceptualism Debate. Philosophy Compass, 9(11):
769–90.
McLear, C. (2015). Two Kinds of Unity in the Critique of Pure Reason. Journal of the History of
Philosophy, 53(1): 79–110.
McLear, C. (forthcoming). Kant on Perceptual Content. Mind.
McMichael, A. (1983). A Problem for Actualism about Possible Worlds. The Philosophical
Review, 92(1983): 49–66.
Menne, A. (1982). Das unendliche Urteil Kants. Philosophia Naturalis, 19(1–2): 151–62.
Mondadori, F. (1973). Reference, Essentialism, and Modality in Leibniz’s Metaphysics. Studia
Leibnitiana, 5(1): 74–101.
Mondadori, F. (1975). Leibniz and the Doctrine of Inter-world Identity. Studia Leibnitiana,
7(1): 21–57.
Mondadori, F. (1985). Understanding Superessentialism. Studia Leibnitiana, 17(2): 162–90.
Morujão, C. (2008). Wolff, Kant und der Begriff der Existenz. In J. Stolzenburg, and
O. P. Rudolph (eds.). Christian Wolff und die europäischen Aufklärung. Akten des 1.
Internationalen Christian-Wolff Kongresses. Hildesheim: Georg Olms.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

BIBLIOGRAPHY 

Nachtomy, O. (1997). Possibility, Agency, and Individuality in Leibniz’s Metaphysics. Dordrecht:

Springer.
Nachtomy, O. (2012). Leibniz and Kant on Possibility and Existence. British Journal for the
History of Philosophy, 20(5): 953–72.
Nadler, S. (1992). Malebranche and Ideas. Oxford: Clarendon Press.
Newlands, S. (2010). The Harmony of Spinoza and Leibniz. Philosophy and Phenomenological
Research, 81(1): 64–104.
Newlands, S. (2013). Leibniz on the Ground of Possibility. Philosophical Review, 122(2):
155–87.
Newlands, S. and Jorgensen, L. (eds.). (2009). Metaphysics and the Good: Themes from the
Philosophy of Robert Merrihew Adams. Oxford: Oxford University Press.
Nolan, D. (1997). Impossible Worlds: A Modest Approach. Notre Dame Journal of Formal
Logic, 38(4): 535–72.
Nolan, D. (2014). Hyperintensional Metaphysics. Erkenntnis, 171(1): 149–60.
Nolan, L. (1997). The Ontological Status of Cartesian Natures. Pacific Philosophical Quarterly,
78(2): 169–94.
Nolan, L. (2005). The Ontological Argument as an Exercise in Cartesian Therapy. Canadian
Journal of Philosophy, 35(4): 521–62.
Nolt, J. (2014). Free Logic. The Stanford Encyclopedia of Philosophy. E.N. Zalta (ed.). <http://
plato.stanford.edu/archives/win2014/entries/logic-free/>.
O’Neill, O. (1992). Vindicating Reason. In P. Guyer (ed.). The Cambridge Companion to Kant.
Cambridge: Cambridge University Press, 280–308.
Oliver, A. (1996). The Metaphysics of Properties. Mind, 105(417): 1–80.
Oppy, G. (1995). Ontological Arguments and Belief in God. New York: Cambridge University Press.
Parrini, P. (1994). (ed). Kant and Contemporary Epistemology. Dordrecht: Kluwer.
Parsons, C. (1969). Kant’s Philosophy of Arithmetic. In S. Morgenbesser, P. Suppes, and
M. White (eds.). Philosophy, Science, and Method. New York: St. Martin’s Press, 568–94. Repr.
(with Postscript) in (1983). Mathematics in Philosophy. Ithaca, NY: Cornell University Press, 110–41.
Parsons, C. (1992). Transcendental Aesthetic. In P. Guyer (ed.). The Cambridge Companion to
Kant. Cambridge: Cambridge University Press, 62–100.
Pasqualino, M. (1994). Möglichkeit, Grund und Prädikation: Zur Rezeption einer Leibni-
zischen Argumentation bei Wolff und Baumgarten. In Leibniz und Europa. Akten des
VI. Internationaler Leibniz-Kongreß. Leibniz Gesellschaft: Hannover, 465–71.
Paton, H. J. (1936). Kant’s Metaphysic of Experience. London: Allen & Unwin. Repr. (1951)
New York: MacMillan.
Pereboom, D. (1991). Is Kant’s Transcendental Philosophy Inconsistent? History of Philosophy
Quarterly, 8(4): 357–72.
Pereboom, D. (2007). Kant on Transcendental Freedom. Philosophy and Phenomenological
Research, 73(3): 537–67.
Pichler, H. (1910). Über Christian Wolffs Ontologie. Leipzig: Dürr’schen Buchhandlung.
Pichler, H. (1912). Möglichkeit und Wiederspruchslosigkeit. Leipzig: Johann Ambrosius Barth.
Plaass, P. (1965). Kants Theorie der Naturwissenschaft. Eine Untersuchung zur Vorrede von
Kants ‘Metaphysischen Anfangsgründe der Naturwissenschaft.’ Göttingen: Vandenhoek &
Ruprecht.
Plantinga, A. (ed.). (1965). The Ontological Argument. Garden City, NY: Doubleday.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 BIBLIOGRAPHY

Plantinga, A. (1967). God and Other Minds. Ithaca, NY: Cornell University Press.
Plantinga, A. (1966). Kant’s Objection to the Ontological Argument. Journal of Philosophy, 63
(19): 537–46.
Plantinga, A. (1974). God, Freedom, and Evil. New York: Harper & Row.
Pollok, K. (2001). Kant’s ‘Metaphysische Anfangsgründe der Naturwissenschaft’: ein Kritischer
Kommentar. Hamburg: Felix Meiner.
Pollok, K. (2002). ‘Fabricating a World in Accordance with Mere Fantasy’? The Origins of
Kant’s Critical Theory of Matter. Review of Metaphysics, 56(1): 61–97.
Poppe, B. (1907). Alexander Gottlieb Baumgarten: seine Bedeutung und Stellung in der Leibniz-
Wolffischen Philosophie und seine Bedeutung zu Kant. Borna-Leipzig: R. Noske.
Prauss, G. (1974). Kant und das Problem der Dinge an sich. Bonn: Grundman.
Proops, I. (2015). Kant on the Ontological Argument. Noûs 49(1), 1–27.
Putnam, Hilary (1975). The Meaning of ‘Meaning’. Minnesota Studies in the Philosophy of
Science, 7: 131–93.
Quine, W.V. (1948). On What There Is. Review of Metaphysics 2(1), 21–38.
Raven, M. (forthcoming). Ground. In Philosophy Compass.
Reich, K. (1932). Die Vollständigkeit des Kantischen Urteilstafel. Berlin: Richard Schoetz.
Robinson, H. (1994). Two Perspectives on Kant’s Appearances and Things-in-Themselves.
Journal of the History of Philosophy, 32(3): 411–41.
Rohs, P. (1978). Kants Prinzip der durchgängigen Bestimmung alles Seienden. Kant-Studien,
69(1–4): 170–80.
Rosefeldt, T. (2007). Dinge an sich und sekundäre Qualitäten. In J. Stolzenberg (ed.). Kant in
der Gegenwart. Berlin: Walter de Gruyter, 167–209.
Rosefeldt, T. (2008). Kants Begriff der Existenz. In V. Rohden, V. R. Terra, and G. Almeida
(eds.). Recht und Frieden in der Philosophie Kants Berlin: Walter de Gruyter, vol. 2, 657–68.
Rosefeldt, T. (2011). Frege, Pünjer, and Kant on Existence. Grazer Philosophische Studien, 82:
329–51.
Rosefeldt, T. (2013). Subject-dependence and Trendelenburg’s Gap. In S. Bacin, A. Ferrarin,
C. LaRocca, M. Ruffing (eds.). Akten des XI. Internationalen Kant-Kongresses. Berlin: de
Gruyter, vol. 2, 755–64.
Rosen, G. (2010). Metaphysical Dependence: Grounding and Reduction. In R. Hale and
A. Hoffman (eds.). Modality: Metaphysics, Logic, and Epistemology. Oxford: Oxford University
Press, 109–36.
Rosenkoetter, T. (2010). Absolute Positing, the Frege Anticipation Thesis, and Kant’s Defin-
itions of Judgment. European Journal of Philosophy, 18(4): 539–66.
Rosenkoetter, T. (2013). Towards a Non-Embarrassing Account of the Modal Functions of
Judging. In S. Bacin, A. Ferrarin, C. LaRocca, M. Ruffing (eds.). Akten des XI. Internatio-
nalen Kant-Kongresses. Berlin: Walter de Gruyter, vol. 2, 383–94.
Russell, B. (1910–11). Knowledge by Acquaintance and Knowledge by Description. Proceed-
ings of the Aristotelian Society, 11: 108–28.
Sassen, B. (trans. and. ed.). (2000). Kant’s Early Critics: The Empiricist Critique of the
Theoretical Philosophy. Cambridge: Cambridge University Press.
Schafer, K. (forthcoming). Kant’s Conception of Cognition and Our Knowledge of Things in
Themselves. In N. Stang and K. Schafer (eds.). The Sensible and Intelligible Worlds: New
Essays on Kant’s Metaphysics and Epistemology. Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

BIBLIOGRAPHY 

Schafer, K. (unpublished). Practical Cognition and Knowledge of Things in Themselves.

Schaffer, J. (2009). On What Grounds What. In D. Chalmers, R. Manley, and R. Wasserman
(eds.). Metametaphysics. Oxford: Oxford University Press, 247–83.
Schmaltz, T. (1996). Malebranche’s Theory of the Soul: A Cartesian Interpretation. Oxford:
Oxford University Press.
Schmucker, J. (1983). Kants vorkritische Kritik der Gottesbeweise. Mainz: Akademie der
Wissenschaften und der Literatur.
Schneeberger, G. (1952). Kants Konzeption der Modalbegriffe. Basel: Verlag für Recht und
Gesellschaft.
Schönfeld, M. (2000). The Philosophy of the Young Kant. Oxford: Oxford University Press.
Schwaiger, C. (2011). Alexander Gottlieb Baumgarten. Stuttgart-Bad Canstatt: Frommann
Holzboog.
Sellars, W. (1968). Science and Metaphysics: Variations on Kantian Themes. New York:
Humanities Press.
Sellars, W. (1976). Kant’s Transcendental Idealism. In P. Laberge, F. Duchesneau, and
B. E. Morrisey (eds.). Proceedings of the Ottawa Congress on Kant. Ottawa: University of
Ottawa Press, 165–81.
Shabel, L. (2007). Kant’s Philosophy of Mathematics. In P. Guyer (ed.). The Cambridge Compan-
ion to Kant and Modern Philosophy. Cambridge: Cambridge University Press, 94–128.
Shaffer, J. (1969). Existence, Predication and the Ontological Argument. In T. Penelhum (ed.).
The First Critique: Reflections on Kant’s Critique of Pure Reason. Belmont, Calif.: Wadsworth,
123–42.
Shoemaker, S. (1980). Causality and Properties. In P. van Inwagen (ed.). Time and Cause.
Dordrecht: D. Reidel, 109–35.
Shoemaker, S. (1998). Causal and Metaphysical Necessity. Pacific Philosophical Quarterly,
79(1): 59–77.
Sider, T. (2012). Writing the Book of the World. Oxford: Oxford University Press.
Sleigh, R. (1990). Leibniz and Arnauld: A Commentary on their Correspondence. New Haven:
Yale University Press.
Smit, H. (2000). Kant on Marks and the Immediacy of Intuition. Philosophical Review, 109(2):
235–66.
Smit, H. (2009). Kant on A Priority and the Spontaneity of Cognition. In S. Newlands and
L. Jorgensen (eds.). Metaphysics and the Good: Themes from the Philosophy of Robert
Merrihew Adams. Oxford: Oxford University Press, 188–251.
Smit, H. (2010). A Priority, Reason and Induction in Hume. Journal of the History of
Philosophy, 48(3): 313–43.
Stadler, A. (1874). Die Grundsätze der reinen Erkenntnistheorie in der Kantischen Philosophie.
Zürich: S. Hirzel.
Stalnaker, R. (1984). Inquiry. Cambridge, Mass.: MIT Press.
Stang, N. (2008). Kant’s Modal Metaphysics. Unpublished Ph.D dissertation. Princeton
University.
Stang, N. (2010). Kant’s Possibility Proof. History of Philosophy Quarterly, 27(3): 275–99.
Stang, N. (2011). Did Kant Conflate the Necessary and the A Priori? Noûs, 45(3): 443–71.
Stang, N. (2012). Kant on Complete Determination and Infinite Judgment. British Journal for
the History of Philosophy, 20(6): 1117–39.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

 BIBLIOGRAPHY

Stang, N. (2014a). Kant, Bolzano, and the Formality of Logic. In C. Tolley and S. Lapointe
(trans. and eds.). New Anti-Kant. New York: Palgrave-McMillan, 192–234.
Stang, N. (2014b). Review of Henry Allison, Essays on Kant (Oxford, 2012). Notre Dame
Philosophical Reviews. https://ndpr.nd.edu/news/47450-essays-on-kant/.
Stang, N. (forthcoming). Kant’s Transcendental Idealism. The Stanford Encyclopedia of Philosophy.
Stephenson, A. (forthcoming). Kant on the Object-Dependence of Intuition and Hallucination.
The Philosophical Quarterly.
Stolzenburg, J. and Rudolph, O. P. (eds.). (2008). Christian Wolff und die europäischen
Aufklärung. Akten des 1. Internationalen Christian-Wolff Kongresses. Hildesheim, Zürich,
New York: Georg Olms.
Strawson, P. F. (1966). The Bounds of Sense: An Essay on Kant’s Critique of Pure Reason.
London: Methuen.
Theis, Robert (1994). Gott: Untersuchung zur Entwicklung des theologischen Diskurses in Kants
Schriften zur theoretischen Philosophie bis hin zum Erscheinen der Kritik der reinen Ver-
nunft. Stuttgart-Bad Canstatt: Frommann-Holzboog.
Thompson, M. (1973). Singular Terms and Intuition in Kant’s Epistemology. Review of
Metaphysics, 26(2): 314–43.
Tolley, C. (2011). Kant on the Content of Cognition. European Journal of Philosophy, 20(4):
200–28.
Tolley, C. (2013). The Non-Conceptuality of the Content of Intuitions: A New Approach.
Kantian Review, 18(1): 107–36.
Tolley, C. (forthcoming). Kant on the Distinction between Perception and Experience.
Tüschling, B. (1990). Intuitiver Verstand, absolute Identität, Idee. Thesen zu Hegels frühe
Rezeption der Kritik der Urteilskraft. In H. F. Fulda and R. P. Horstmann (eds.). Hegel und
die Kritik der Urtheilskraft. Stuttgart: Klett-Cotta, 174–88.
Tüschling, B. (1992). The System of Transcendental Idealism: Questions Raised and Left Open
in the Kritik der Urtheilskraft. In H. Robinson (ed.). System and Teleology in Kant’s Critique
of Judgment (Spindel Conference 1991). Southern Journal of Philosophy, 30, Supplement,
109–27.
Vaihinger, H. (1881). Commentar zu Kants Kritik der reinen Vernunft, 2 vols. Stuttgart:
Spemann.
Van Cleve, J. (1979). Foundationalism, Epistemic Principles, and the Cartesian Circle. Philo-
sophical Review, 88(1): 55–91.
Van Cleve, J. (1994). Descartes and the Destruction of the Eternal Truths. Ratio, 7(1): 58–62.
Van Cleve, J. (1999). Problems from Kant. New York: Oxford University Press.
Vander Laan, D. (2004). Counterpossibles and Similarities. In G. Priest and F. Jackson (eds.).
Lewisian Themes. New York: Oxford University Press, 258–76.
Walker, R. (1990). Kant’s Conception of Empirical Law. Proceedings of the Aristotelian Society,
64: 243–58.
Watkins, E. (1997). The Laws of Motion from Newton to Kant. Perspectives on Science 5(3):
311–48.
Watkins, E. (1998a). The Argumentative Structure of Kant’s Metaphysical Foundations.
Journal of the History of Philosophy, 36(4): 567–93.
Watkins, E. (2005). Kant and the Metaphysics of Causality. Cambridge: Cambridge University
Press.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi

BIBLIOGRAPHY 

Watkins, E. (2006). On the Necessity and Nature of Simples: Leibniz, Wolff, Baumgarten and
the Pre-Critical Kant. Oxford Studies in Early Modern Philosophy, 3: 261–314.
Watkins, E. and Fisher, M. (1998). Kant on the Material Grounds of Possibility. Review of
Metaphysics, 52(2): 369–95.
Werkmeister, W. H. (1981). Kant’s Silent Decade. Tallahassee, Fla.: University Press of Florida.
Westphal, K. (1995). Kant’s Dynamic Constructions. Journal of Philosophical Research 20:
381–429.
Westphal, K. (2000). Kant, Hegel, and the Fate of ‘the’ Intuitive Intellect. In S. Sedgwick (ed.).
The Reception of Kant’s Critical Philosophy: Fichte, Schelling, and Hegel. Cambridge: Cam-
bridge University Press, 283–305.
Westphal, K. (2004). Kant’s Transcendental Proof of Realism. Cambridge: Cambridge University
Press.
Willaschek, M. (1998). Phaenomena/noumena und die Amphibolie der Reflexionsbegriffe
(A235/B294-A292/B349). In G. Mohr and M. Willaschek (eds.). Immanuel Kant: Kritik
der reinen Vernunft. Berlin: Akademie Verlag, 324–52.
Williamson, T. (2013). Modal Logic as Metaphysics. Oxford: Oxford University Press.
Wilson, J. (2014). No Work for a Theory of Grounding. Inquiry, 57(5–6): 1–45.
Wilson, K. (1975). Kant on Intuition. The Philosophical Quarterly, 25(100): 247–65.
Wilson, M. (1978). Possible Gods. Review of Metaphysics, 32(4): 717–33.
Wilson, M. (1990). Leibniz’s Doctrine of Necessary Truth. New York: Garland Publishing.
Wippel, J. (1982a). The Relationship between Essence and Existence in Late-Thirteenth-
Century Thought: Giles of Rome, Henry of Ghent, Godfrey of Fontaines, and James of
Viterbo. In P. Morewedge (ed.). Philosophies of Existence: Ancient and Medieval. New York:
Fordham University Press, 131–64.
Wippel, J. (1982b). Essence and Existence. In N. Kretzmann, A. Kenny, and J. Pinborg (eds.).
The Cambridge History of Later Medieval Philosophy. Cambridge: Cambridge University,
385–410.
Witt, J. (2010). Essence and Existence. In H. Lagerlund (ed.). Encyclopedia of Medieval
Philosophy. New York: Springer, 304–9.
Wolff, M. (1995). Die Vollständigkeit des Kantischen Urteilstafel. Frankfurt a.M.: Vittorio
Klostermann.
Wolfson, H. A. (1947). Infinite and Privative Judgments in Aristotle, Averroes and Kant.
Philosophy and Phenomenological Research, 8(2): 173–87.
Wood, A. (1970). Kant’s Moral Religion. Ithaca, NY: Cornell University Press.
Wood, A. (1978). Kant’s Rational Theology. Ithaca, NY: Cornell University Press.
Wood, A. (1984). Kant’s Compatibilism. In A. Wood (ed.). Self and Nature in Kant’s Philosophy.
Ithaca, NY: Cornell University Press, 73–101.
Wundt, M. (1924). Kant als Metaphysiker. Stuttgart: Ferdinand Enke.
Wundt, M. (1939). Deutsche Schulphilosophie im 17. Jahrhunderts. Tübingen: J. B. C. Mohr.
Wundt, M. (1945). Die Deutsche Schulphilosophie im Zeitalter der Aufklärung. Tübingen:
J. B. C. Mohr.
Yong, P. (2014). God, Totality and Possibility in Kant’s Only Possible Argument. Kantian
Review, 19(1): 27–51.
Yong, P. (forthcoming). Hegel and the Given: A Phenomenological Interpretation of Hegel’s
Philosophy. Ph.D dissertation. University of California, San Diego.
OUP CORRECTED PROOF – FINAL, 16/2/2016, SPi
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

Index

Table and figures are indicated by an italic t and f following the page number.

Abaci, U. 81n2, 97nn39, 40, 107, 107n21, ground of formal possibility 178–9, 205–7,
108n22, 204n22, 206n28, 213n57 210, 269, 275–6, 278
absolute necessity, see necessity: absolute see also understanding
absolute positing 77–9, 86, 155, 163–6, 170, a priori 228–32, 264–5, 275, 298
211, 239n30, 269, 301, 308, 313, 321–2, epistemic meaning of 23–4, 28, 40n114,
324; see also existence 148n55, 154–5, 159–60, 170, 244–6,
accidents, see properties: accidental 249–50, 254, 257–8, 263, 283, 317
acquaintance 79 ‘from the grounds’ meaning of 23–6, 28–9,
actualism 35, 45–8, 55–6, 59; see also 84, 147–8, 173n59, 212–13, 217–18,
possibilism 243n40, 258n77, 282–3, 310; see also
actuality 20n31, 23, 27, 30, 55, 57–73, 87, 101, insight
103, 113, 118, 145, 174, 222n78, 220–4, semantic meaning of 120, 153, 156–7, 202
227, 231–3, 269–71, 276–9 see also cognition: how possible a priori
entails possibility 24, 104, 161, 174, 186, 318 Aquinas, Thomas 135n26
intuitive intellect and 310–12, 317–21 Aristotle 14, 14n9, 208n34, 238n27
modal concept 1, 211n46, 307–10, 318–19, attributes, see properties: attributes
323–5 Axioms of Intuition 210n39, 211nn43–4
real possibility grounded in 129–30, 149,
198–201, 203, 205, 207, 215–16, Bartholin, T. v–vi
218, 226, 241, 254–5, 261–2, 269, Baumgarten, A. G. 2–3, 13–14, 26, 29n79,
273–4, 279 35n96, 76, 83n11, 114n40, 115, 166, 169,
relation to existence 34–5, 57, 59, 238, 269, 173, 207, 236n25, 311
321–5 complete determination theory of
rigid vs. non-rigid conception of 70–3, 130, existence 34, 37, 42n117, 56–7, 62–5,
276n29 290, 292–3
see also existence Kant’s critique of 63–5, 72–3, 84–6
Adams, R. M. 20n34, 23n48, 50nn22, 24, 53, Metaphysica 2, 14, 34, 37, 57–9, 84–5, 311
54nn36, 41, 55–6, 72n71, 134n25, ontological argument 27, 30n84, 56–63
135n27, 262n5 views on possibility and necessity 15–18, 20,
aggregate 104n11, 130, 132–3, 138, 141n43, 295 25, 58–61
Allais, L. 163n35, 168n28, 177n74, 181n82 Bausch, J. L. vi
Allison, H. 181n83, 228n1, 241n35, 299n4, beings 26, 35, 44, 47, 57, 60, 76, 111, 153–4,
304–5, 305nn20–1 166–7, 169, 173, 195, 293–5; see also ens;
alteration 116, 191, 202, 207n28, 208, 212–19, objects; things
222–4, 226, 228–9, 231, 244, 258, 274, belief 83, 263, 280–8, 290, 299, 325
276–8, 284, 318; see also causation necessary theoretical belief 8, 262–3,
Analogies of Experience 210n39, 211n44, 287–8, 296
215n61, 229n2, 238 Blumenbach, H. vin7
analytic judgments 21, 39–40, 74, 88n30, 239, bodies 48, 81, 84n18, 91, 93, 103, 162, 224, 234,
270, 284, 322 243, 247, 249, 251, 253n68, 291
Anselm, Saint 28, 44
Anticipations of Perception 210n39, cancelation
211nn43–4 cancelation of existence 74–5, 122–3
appearance 176, 180–2, 186, 200, 206, 210n39, cancelation of predicates in real conflict, see
211nn43–4, 214, 220, 222, 224–5, 249, real conflict: predicate-canceling
264, 295, 297; see also phenomena cancelation of real possibility 123–35, 141,
apperception 196, 200, 204n22 148, 273, 276–8, 290, 294, 296, 327
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

 INDEX

cancelation (cont.) a priori 3, 7, 153, 159, 169n47, 174, 188,

hyperintensional relation 125–6, 148; see also 202, 265
hyperintensionality atomic vs. molecular 22, 50, 163n34
Cartesian circle 193, 193n112 clarity 16n16
Cartesian predication principle 45–7 completeness of 289
categories containment relations among 15–16, 22, 85,
cause-effect 164, 180n78, 202, 227–8, 91–2, 149, 156n11, 198–9
267, 322 distinctness 16n16, 21
community 210–12 empirical kind concepts 237
mathematical vs. dynamical 210n39, 211n46, empirical vi, 8, 37, 160, 174, 185–9, 186–7t,
263n11 237, 244, 245f, 247, 250, 265
modality, see modal categories extensions of xvii, 26, 37–8, 42, 59, 63–4,
objective validity 159, 206, 264–5 75–6, 162n29, 199
quality 167, 175, 210, 210n39, 211n43, intellectual 158, 159n23, 160, 194
263n11 intuitive vs. symbolic grasp of 21–3
quantity 167, 175, 210, 210n39, 211nn43, 46, logicist theory of 4–5, 21–6, 80
263n11, 268n17 ontological status of 16n16, 21n37, 26
relation to judgment 175–7, 232–3; see also vs. intuitions 3, 6–7, 77, 120, 156–7, 159–60,
schemata and schematism 162–71
relational 211, 268n17 conditions, see ground
substance-accident 176, 202 connection (nexus) 15n11, 85, 88, 90, 92,
Caterus 44, 46n11, 47, 49, 52, 62n57 103n8, 139, 164, 180, 220, 222; see also
causation causation; ground
category of 164, 180n78, 202, 227–8, 227n95, construction 174
267, 322 construction of the concept of matter 247–8,
governed by laws 7, 202–3, 202n19, 214–15, 254n71
215n62, 222, 224–5, 224n85, 228–32 mathematical 174, 207, 247, 265
interaction 137, 139–40, 142–4 metaphysical 247–8
logicist account of 81, 84–5, 93 procedure of MFNS 247–8
real grounding 81, 88–98, 146 contingency 15n11, 17n21, 88, 90n31, 124n4,
chemistry 242–3, 246, 257 135n27, 298, 306, 316n49, 321
Chignell, A. 93n37, 97n39, 102n6, 107–12, correlates, see noumenal correlates
108f, 113n36, 114nn39, 41, 129n13, cosmology, see worlds, Kant’s theory of
135–6, 145n49, 154n1, 161n27, 179n76, counterfactuals 33–4, 88n29, 125, 134–5, 256,
183n89, 204n22, 280n38, 283n51, 276–8, 283n51
287n61 counterpossibles 121, 125, 128, 134, 273
cognition Cover, J. 19n28
a priori 3, 24, 85, 162, 185, 190–6, 328 Crusius, C. A. 13, 42n117
empirical 174–5, 201, 201n11, 211n45, views on freedom 194n116, 214n60, 315–16,
248n56 315n47
how possible a priori 24, 162, 169, 171–83 views on ground 83–4, 83nn7, 11,
mathematical 154, 173–5, 185n91 84nn14–16, 139n39, 146n50, 187t, 191,
modal characterization (m-cognition) 169, 191n107, 192n110, 199n9, 207, 208n34
171–5 views on possibility 101, 101n2, 107n21, 113,
semantic characterization 160, 190–5, 193n113
(s-cognition) 171–4
sensible vs. intellectual 153 deduction, see Transcendental Deduction
transcendental 3, 168–9, 169n47 definition
vs. knowledge 24, 171–2, nominal 23, 25, 25n59, 30, 234–5, 237,
227, 263 265–7, 266n14, 270, 309, 319
compatibilism 124n4, 298, 315n45; see also real 23–7, 30–1, 49, 122–3, 127–8, 148,
freedom 234–7, 263–72, 309
conceivability 190–1 real vs. nominal 23, 234; see also essence:
concepts whether real or logical predicate
adequacy 21–2 role in philosophy 234–5, 265–6
analysis of 18n28, 22, 49–50, 67n64, 89–90, de re vs. de dicto possibility 18, 130, 221,
155–6, 164, 226, 234 258n77
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

INDEX 

derivative predicates, see predicates: derivative existence as quantifier 32, 34–5, 37–9, 41,
Descartes, R. 55–6, 76–7
Meditations 20, 43–9, 52–3 God’s, see God: existence of
not a logicist 5, 20, 31, 43 intersubjectivity and 210–11, 211n44
ontological argument 5, 31n87, 43–9, 76 judgments of 248–9
Principles of Philosophy 48–9 Kant’s argument that existence not a real
reply to Gassendi 45–7 predicate
views on clear and distinct perception 44 Beweisgrund 37, 65–73
views on existence 36n97, 43–9, 45n8, 53 Critique of Pure Reason 4, 74–7
views on modality 44–5, 113 Kant’s positive theory of 77–9, 163–6,
determination 321–5
Kant’s account of 36–42 Kant’s real definition of necessary
principle of complete determination 64–5, existence 122–8
73 necessary 1, 29–30, 52–5, 121–32, 134–5,
synonymous with real predicate 36 148, 271f, 273, 276, 278t, 279, 318–19,
see also predicates: real 323
determinism not logically necessary 42, 80, 97, 123
empirical 276 relation to actuality 34–5, 57, 59, 238, 269,
inconsistent with freedom 7 321–5
Kantian 217–18, 276–7 univocity of 54, 56, 62
Diderot, D. v vs. existence* 238–9, 269, 322–4
Dummett, M. viin8 whether real or logical predicate 36–7
whether there are objects that lack it, see
Eberhard, J. A. 14n9, 88 objects: non-existent
ens 14–15, 14n8, 18, 27, 35, 57, 166; see also see also determination; predicates: real
objects; things experience
ens imaginarium 167, 168f determinable forms of 202–3
ens perfectissimum, see God; perfection form of 200, 205n23, 208n33, 209–10, 213,
ens rationis 167–8, 168f 221, 223, 227
ens realissimum, see God; reality intellectual conditions of 179, 196, 205,
epistemology 275–6, 305; see also apperception;
basing relation 281–8, 283n50 categories
logicist 4, 21–6 matter of 190, 223–4, 225n89, 247, 264n12
modality, see modal epistemology possibility of non-actual form of 221–3,
see also belief; Fürwahrhalten; knowledge 224n83, 225n89, 227
essence possibility of non-actual matter of 221–3,
God’s, see God: essence of 225n89, 227
logicist view of, see logicism: theory of totality of 220, 263n7
essence extension, concept of 80–1, 93–4, 101–3,
matter 8, 229, 235–7, 242nn37, 39, 243–59, 108–11
245f, 252f externalism 189–90
real vs. logical 234–8
vs. nature 238–44 Fine, K. 122n2, 130, 130n17, 132
essential properties, see properties: essential force 81, 90–3, 96, 186, 189, 204n22, 250
excluded middle, principle of 42, 63–5, 73; attractive 235, 248n58, 291–2
see also determination: principle of essential to matter 243–4, 251–6
complete determination relation to forms of experience 253–4
existence inverse-square law of gravitation 7, 232, 237,
absolute necessity and 121–32, 134–5, 148, 244–6, 258
271f, 273, 276, 278t, 279, 318–19, 323 Forgie, W. 39n108, 45n10
category of 1, 38n105, 322n64 form of experience, see experience: form of
complete determination, see Baumgarten: free logic 32n89
complete determination theory of freedom
existence; determination: principle of God’s will and divine freedom 87, 113,
complete determination 124n4, 145–6, 310, 313–14
equivalent to absolute positing 77–9, 155, implied by ought 224, 224n85, 297, 306
155n9, 163–6, 170, 269, 301, 321–5 intuitive intellect 100, 165, 300–2
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

 INDEX

freedom (cont.) intellectual reading 114–16

apparent inconsistency and its Kant’s Critical attitude to 99–149
resolution 8–9, 297–9, 312 Kant’s failure to give positive account
Kant’s Critical view of 300–25 of 117–18
Kant’s pre-Critical view of 124n4, 315–16, Kant’s pre-Critical argument for 88–98,
315nn45, 47, 316, 325–8 122, 128–35: plurality objection to 122,
libertarian 194n116, 317 132–5; response to plurality objection
non-modal characterization of 9, 315–16, 132–44
318–20 powers reading 6, 9, 112–14, 116–18,
possibility of 145–6, 149, 327–8
empirical-causal 7, 224–5, 297–8 tension in Kant’s pre-Critical conception
noumenal 8–9, 224–5, 297–9, 302, 313–21 of 9, 100, 114n39, 118, 146, 149: how
ratio essendi of the moral law 208n33 resolved in Critique of Judgment
Frege, G. 37–8 298–302, 306–12
Friedman, M. 38n104, 207n29, 233, 246 self-cognition 115, 310–11
fundamental predicates, see predicates: simplicity of 27, 132–5, 294
fundamental uniqueness of 4, 35, 50, 66–7, 104, 126–7,
Fürwahrhalten 172, 261n4, 262–3, 280–8, 283f, 131–2
302n12, 317n53 will of 87, 113, 124n4, 145–6, 310, 313–14
belief 263, 280–7 gravity, see force: attractive
knowledge 281–3, 286–7 Griffin, M. 48n16, 52nn29–30
objective sufficiency of 280–1, 283–4, 288 ground
opinion 280–3, 285 antecedently determining 83–4, 104–6,
subjective sufficiency of 280–4, 287 116–18, 147, 213
Beweisgrund (ground of proof) vs. Beweis
GARP (ground of all possibility) 105, 105n15, (proof) 128–35
128, 130–3, 138, 140–2, 141f, 149, 294–6; causal, see causation
see also God consequently determining 148
Gassendi, P. 39n108, 45–7 determining vs. sufficient 214n60, 315–16;
geometry, see mathematics see also Crusius
God explanatory vs. epistemic 83–4
divine understanding 116n48, 139, 140n40, Kant’s Critical theory of 197–200, 207–8,
144n 48 259, 271f, 275
ens perfectissimum 43n1, 50–4, 58–9, 61; Kant’s pre-Critical theory of 80–98, 138–44,
see also perfection 199, 212–13, 289
ens realissimum 29n79, 51, 62n57, 93–7, 108, logicist theory of, see logicism: theory of
274; see also reality grounds
essence of 4, 27, 43, 49, 53, 56, 58, 61n57, 62, possibility, see possibility: grounds of
100, 115, 119, 145, 148, 313, 326–7 ratio essendi 208, 212, 236
existence of ratio fiendi 208
whether essential 27, 31–2, 56 real vs. logical ground 81, 83n11, 88–91
why necessary 30, 130–2, 272–80 GSRP (ground of some possibility) 105,
see also ontological argument 105n15, 130–4, 136, 138–40, 144;
his cognition of actuality 313 see also God
his cognition of possibility 310–11, 313 Guyer, P. 231nn6, 9, 232
intellect of 8, 82n5, 105n14, 114–17, 119,
192, 298 Hanna, R. 39n110, 205n23, 251n67, 287n62
logicist view of his necessary existence, 4–5, harmony, see pre-established harmony
14, 30, 43, 75–6, 80 Hegel, G. W. F. 82n6, 119n50, 184n90,
object of theoretical reason 149, 287–8, 189n98, 299
296, 299 Heidegger, M. 14n5
one-way causal influence 139n36 Heimsoeth, H. 101n2, 191n106
ontological argument for, see ontological Herz, M. see Kant’s works: letter to Herz
argument Hogan, D. xiii, 86n23, 88n30, 101n2, 124n4,
real ground of real possibility 6, 8, 99, 116, 214n60, 314n41, 315n45, 316n50,
121, 238 317n55
harmony reading 107–16 Honnefelder, L. 36n97
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

INDEX 

Hume, D. 161n25, 180, 180n78, 232 Critique of Pure Reason 1–5, 74–7, 197–265,
hyperintensionality 121, 121n1, 123n3, 124n5, 280–96
125–6, 125n7, 148, 204, 205n23 Inaugural Dissertation 6, 153,
156–8
impossibility 44n6, 120, lectures on logic 234
166n41, 192 lectures on metaphysics 29, 42, 166n41,
absolute 30 208n35, 234
concept of 3 letter to Herz 7, 153, 158–62, 175, 184–8,
see also nothing 191n107, 192
infinity 251 Metaphysical Foundations of Natural
insight 21–3, 25, 29n76, 85, 212–13, 217, 230, Science 8, 204n22, 229, 233, 238,
257–8, 258n77, 262, 268, 317n52 240–54, 257
Insole, C. 140n40 Mrongovius lectures 2, 15, 40n113, 208, 235
instantiation xvii, 16–17, 24, 82, 104, 107, 109, Negative Magnitudes 5, 81, 85, 88–9, 90–1,
119, 142, 144–5, 188–9 94, 108n22, 156, 212, 213n54
intellectual intuition 158, 299, 299n4, 300–2, Nova dilucidatio 28, 86n23
306, 312 Prize Essay 6, 153–6, 175
interaction, see causation knowledge
intuition distinct from cognition 24, 171
acquaintance 79 distinction between a priori and a
giving us objects 6–7, 120, 154, 157, 174 posteriori 23, 202–3
intellectual, see intellectual intuition see also a priori; belief; epistemology;
relation to existence 162–6 Fürwahrhalten
space and time as pure forms of 146n50, 156, Kreines, J. 228n2, 232n10
159n22, 167, 174–5, 177–9, 182, 196, Kripke, S. viin8
198n3, 202–4, 205–6, 221, 228
vs. concepts 6–7, 159, 162–3 laws of nature, see natural laws
see also perception Leech, J. 233n15, 301n11, 308n30
intuitive intellect 8, 82n5, 105n14, 114–17, 119, Leibniz, G. W.
192, 298, 310–12, 317–21 a priori/a posteriori distinction 23
freedom and, see freedom: intuitive intellect correspondence with Elizabeth, Princess of
modality and 306n23, 307, 312 Bohemia 24n54, 49n19, 50n21, 51n27
reception in German Idealism 299n3 correspondence with Wolff 21
whether distinct from intellectual critique of Descartes 49–50, 52
intuition 301n11 critique of Spinoza 52, 55
inverse-square law, see force: inverse-square law De Summa Rerum 50n23, 52n28, 54n40
of gravitation Discourse on Metaphysics 22n41, 24n54,
49n19, 51n25
judgment doctrine of complete concepts 18n28,
infinite vs. negative 64 67–8, 90
logical functions of 175–7, 179, 232–3 doctrine of the identity of
modal functions of 232, 233n15 indiscernibles 50–1
relation to categories, see categories: relation Kant’s critique of vi, 29, 29n79, 65–71
to judgment logicism 4, 27, 50, 84, 88
see also existence: judgments of; “Meditations on Knowledge, Truth, and
Fürwahrhalten Ideas” 21, 23–5, 49, 51, 53
Monadology 13n2, 27n63, 50
Kant’s works monads 115–16, 173
Beweisgrund vi, 5–6, 8–9, 28, 29n80, 30, New Essays 17n22, 18nn26, 28, 21n38,
36–7, 42n116, 43, 51, 63, 65–73, 75, 77, 22nn41, 44, 23nn50–1, 51n27, 54n37
80–1, 93, 97–100, 106–22, 124n4, 126–9, “On the Ultimate Origination of
139, 145–53, 163, 197n1, 273, 294–6, Things” 54n40
325–8 ontological argument 5, 27n63, 31n87, 49–56
Critique of Judgment 8, 147, 297–302, possible worlds 34, 70
305–10, 312, 316, 321–2, 325 pre-established harmony doctrine 84n18,
Critique of Practical Reason 208n33, 224n85, 115–16
225, 227n95, 297, 305, 319n59 Protogaea v
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

 INDEX

Leibniz, G. W. (cont.) metaphysica generalis 1–2, 7, 9, 153, 166–7

texts available in eighteenth century 13n2 metaphysica specialis 1–2
Theodicy 317n57 ontology 1–4, 14–15, 20
views on concepts 18n28, 19, 21–5, 49–50, modal categories 182, 211n46, 214, 219, 232–3,
76–7 305–7
views on definition 23–4 not determinations 165
views on existence 27, 51n27, 53–6 schematized vs. unschematized 264, 267–72,
views on necessity 19, 21 309–10, 312, 323–4
views on possibility 15, 17–18, 20–1, 20n34, see also possibility: category of; existence:
25–6, 35 category of; necessity: category of
views on truth 18n28 modal epistemology
work on universal symbolic language 17 divine illumination theories of 160
Lewis, D. 34n92, 68n65, 125n7, 138n35, 215n64 circularity objection to 160–1
logic, formality of 76 Kant’s epistemology of real
logicism possibility 153–96
doctrines of modality 13–26, 50, 84, 148–9 logicist, see logicism: doctrines of modality;
epistemology 4, 21–6 logicism: epistemology
explanation of divine necessity 4, 27, modal logic 134, 137
30–1, 42 modal pluralism 132, 135n26, 140–2, 145–6
theory of concepts 4–5, 21–6, 80 modality, see actuality; existence; modal
theory of essence 4, 13–14, 18, 21, 24, 26–7, categories; necessity; possibility
31–6
theory of grounds 18, 21, 24–30, 84–9 narwhals v–vii, 37–9, 79
theory of necessity 4, 25–7, 30–1, 80, natural laws
86–7, 155 causation and 87, 215–18, 224n85
theory of possibility 4, 25–6 definition of 241
theory of properties 5, 14–21 essences and 228–34, 241–2, 246, 251n67,
Longuenesse, B. xiii, 87n28, 299n4 254–9, 277–8
modal status of, see necessity: nomic
Malebranche, N. 158, 160, 187t, 190–5 nature and 228–30, 240–54
mammoths vi natural kinds
Manicheanism 140n42 empirical concepts of 240, 240n33
manifolds 175–9, 210–11, 213, 230, 248n56, essences of 8, 229, 237–8, 241–2, 246, 254–9,
257, 275, 309–10 277–8
Mates, B. 16n16, 18n28, 20n34, 21n37, 25n58, see also essence
56n45 nature
mathematics 243, 247 formal vs. material meaning of 238
a priori cognition of pure quantities 175 law-governed 228–30, 240–54
concepts of 154 vs. essence 238–40
construction in, see construction: necessary truths 4, 8, 17n21, 25, 42, 53, 123–6,
mathematical 155, 283
distinct from philosophy 154 necessity
geometry 48, 203–4, 207, 238 absolute 86, 122–8, 204, 276, 278t, 279,
grounding in 204 318–19, 323
mathematical objects 48, 146n50, 174–5, cancelation of real possibility 121, 123–35,
239, 322 141, 148, 273, 276–8, 290, 294, 296, 327
matter necessity of God’s existence 148, 273
essence of 8, 235–7, 242n37, 243, 246, 250–6, understood counterpossibly 121, 125–6,
252f, 258–9, 278 128, 134, 273
laws of 8, 245f, 278 category of 1, 233, 263, 305
see also experience: matter of demonstrability and 155
McLear, C. 162n28, 163n35, 176n72 empirical-causal 214
Metaphysical Deduction 180 absolute 274, 276–8, 278t
metaphysics co-extensive with empirical-causal
distinct from transcendental philosophy 1–4, possibility 215–18
166, 169, 209 determinism and 216, 276–7
eighteenth-century textbooks 2 freedom and 224–5, 306
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

INDEX 

definition of 216 non-actual 34, 34n92

necessity of all alterations 215–17, non-existent vii, 4, 14, 34–5, 40–2, 43, 55,
231, 276 60, 77, 148, 165
existence, see existence: necessary object as highest concept of transcendental
formal 203–4, 207, 214, 228, 231 philosophy 2–3, 166–71, 244, 245f
absolute 275, 279n37 quantificational concept of (q-objects) 26,
definition of 203 35n95, 166, 170–1, 290, 313, 322
necessity of causal principle 231, 246 relation to beings 3–4
necessity of mathematics 284 relation to intuition 6–7, 120, 154, 157, 174
in se vs. hypothetice 15, 20n31, 86 representation in general (r-objects) 159–60,
interdefinable with possibility 19, 203, 220, 170–1, 179
220n72, 226n90, 255, 270, 279, 279n37 see also beings; ens; things
logical 5, 75, 122–3, 127, 191, 231, 233, ontological argument
256, 259 Baumgarten’s 27, 30n84, 56–63
nomic 7–8, 228–59 Descartes’ 5, 31n87, 43–9, 7
definition of 229 explanation of divine necessity 4–5, 13,
distinct from other kinds of 27n63
necessity 230–3 Leibniz’s 5, 27n63, 31n87, 49–56
essence and 234–44 Kant’s criticisms of 4, 14, 28–31, 37,
grounds and 235–6, 239, 241–2, 252–3f, 65–77, 164
255–7 ontology
puzzle about 246–54 distinct from transcendental philosophy 1–4
non-contradiction and 30, 74, 123 highest concept of 2, 166
noumenal 219–27, 231, 288, 318–19, 323 metaphysica generalis 1–2
real vs. logical 123 ontotheism 41–2, 65, 69, 77
negation 4, 16, 50, 64, 110–12 relation to logicism 4–5, 13–14, 26–31
Newlands, S. 55n42, 114n39, 145n49 ontotheology 27–8, 43, 147
Nolan, D. 125n7, 126n9
Nolan, L. 48n17 parts and wholes 176n72, 210
nomic necessity, see necessity: nomic perception 164, 210–11, 210nn38–40,
non-contradiction, principle of, see Principle of 211nn43–6, 220–2, 247–9, 248n58
Non-Contradiction clear and distinct 43–4, 49–50, 53
nothing sensory manifolds 210, 210nn38–9,
concept of 167 211nn44, 46, 213, 245n56, 310
deprivation vs. lack 45, 96 see also intuition; phenomena; space; time
logical impossibility and 75–6, 122–3, 169 perfection 45–8, 51, 108n22, 147
real impossibility and 2, 99 existence as 51n27, 53–4, 62
see also ‘Table of Nothings’ God’s essence 53–5, 58
noumena limited 50
causal powers of 224–7, 271f, 322 unlimited 50–1
grounds of the appearance of empirical see also reality
objects 200, 270–1 phenomena
grounds of possibility 200, 223, 272, a priori cognition of 171–83, 190, 263
292–3, 323 causal relations among 216–19, 228
lack of modal properties 8, 203, 298, 299n4, instantiation of categories 7, 189
303–4, 327 objects of sensible intuition 7, 171, 182n86,
negative vs. positive 302–3, 303f 183, 190, 195–6
no cognition of 172, 182–8, 197, 223, phenomena/noumena distinction 7, 182–3,
270–1, 296 196, 263–5
see also freedom; intuitive intellect; real grounds of 196–7, 198n5, 200–1, 206,
possibility: noumenal; things in 209, 213, 270–2, 271f, 275–6
themselves physics 8, 204n22, 229, 233, 238, 240–54, 257
noumenal correlates 8–9, 312–15, 317–21, 320t, empirical vs. rational 244–6, 245f
324–7 Plato 160, 190n101
pluralism, see modal pluralism
objects possibilia 35, 48–9
actual vs. possible vii, 60, 323–4 possibilism 14, 31–6, 38, 43, 52, 56, 60–1
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

 INDEX

possibility real possibility 194, 198n6, 224, 226, 273

category of 1, 17, 267–9, 271f, 272–3, 278, real 97–8
305, 307, 309–10, 319n59 distinct from logical possibility 5, 101,
concept of 3, 8, 24, 51, 100–7, 198–200, 149, 169–70, 197–9
261–2, 268–70, 272–3, 278, 292–3, 296, unschematized category of 268–9, 271f,
317, 319n59 272, 278, 309–10, 327–8
empirical 8, 215n63, 216, 218, 223, 226–7, see also God: real ground of real possibility
271f, 277 real grounds of 8, 146n50, 156, 200,
empirical-causal 201f, 214–19, 276, 278t, 207, 270
292, 318 relation to thinkability 6, 103, 119–20,
co-extensive with empirical-causal 132–3
necessity 215–18 possible worlds 33–4, 66–71, 125, 137, 205n23,
definition of 215, 218 215, 220–1, 256, 276
freedom and 7, 224–5, 297–8 God’s choice in creation 54
grounds of 200, 216–18, 223, 270, 277 Kantian world (k-world) 137–8,
real possibility 218, 267 137n34
relation to a priori 214 maximally perfect 17n21, 20n31, 54, 89,
formal 8, 71, 200–13 124n4
definition of: in terms of modal indices 137–8
compatibility 204–5 in terms of non-actual 52, 59
grounding 203, 205 Postulates of Empirical Thinking 7, 203–4, 214,
grounds of 204, 206–7: immanent 200, 219, 293, 322n64
201f, 223, 270, 271f; intellectual 275–6; powers
real 207–8; sensible 275 causal powers of phenomena 198, 216–18
transcendent 274–5 freedom 271, 316–17, 317n52, 326
real possibility and 206–7 God’s powers as real ground of real
relation to a priori 206 possibility 6, 9, 112–14, 116–18, 145–6,
vs. causal possibility 7 149, 327–8
see also necessity: formal grounds of possibility 217–18, 224,
givenness of 6, 104–5 226–7
God as the ground of, see God non-modal noumenal correlates 9, 312–15,
immanent grounds of 200, 201f, 223, 270, 317–21, 320t, 324–7
271f, 278t noumenal causation 224–7, 271f, 322
in se vs. hypothetice 15, 20n31 see also Crusius
interdefinable with necessity 19, 203, 220, predicates
220n72, 226n90, 255, 270, 279, 279n37 compatibility of 5, 20, 27, 80–2, 93–9, 101–2,
logical 64, 93, 97–8, 104, 123–4, 157, 168–71, 107–11, 120, 135–6, 149, 156
195, 201f, 203, 207, 226, 266, 268, 270, conjunctions of 58, 107, 110, 112
271f, 279 derivative 99–100, 104–13, 106f, 108f,
distinct from real possibility 5, 101, 149, 117–18, 121, 131, 136, 141–5,
169–70, 197–9 141f, 292
logicist theory of 4, 25–6 derivative vs. fundamental 99–100
material requirement on 101–4, 107, fundamental 99–100, 104–14, 106f, 108f,
109, 129 117–18, 121, 131, 135–6, 138–45, 141f,
nomic 8, 201f 148, 292
definition of 255 Kant’s argument for real incompatibility
grounds of 271f of 93–9
real possibility 254–7, 267 ontological status of 16n16
non-contradiction and 20, 123, 169, 191 real 4–5, 36–42
not a real predicate 165 distinct from logical 36
noumenal 8–9, 219–27, 271–2, 271f, 272–80, equivalent to determinations 36
287–8, 288–96 whether existence is one, see existence:
definition of 318 Kant’s argument that existence not a real
freedom and 8–9, 224–5, 297–9, 302, predicate
313–21 real conflict of, see real conflict
grounds of 200, 223, 226, 272, 292–3, 318, pre-established harmony 84, 84n18,
323: God 8, 274, 321 115–16, 190
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

INDEX 

properties schemata and schematism 176–7, 179,

accidental 18, 27, 57–8, 60–4, 73, 139, 142, 213, 266
176–7, 191, 202, 209, 212, 219 schematized modal categories as rules of time
attributes 18, 27, 48n17, 57–8, 60, 93, determination 309–10
110n27, 136, 234, 236–8, 256 schematized and unschematized
essential 18, 27, 32–3, 57–8, 60–1, 86, 234, categories 264, 267–72, 309–10, 312,
236, 239, 243–4, 253–6, 258n77 317–20, 322–5, 327–8
propositions 16–21, 24–6, 125n7 unschematized category of the real 268–9,
disjunctive 2 271f, 272, 278, 309–10, 327–8
hyperintensional relations 121n1, 125 Schneeberger, G. 223n81
indemonstrable 155 Schönfeld, M. 13n3, 101n2, 129n15, 139n39
ontological status of 16n16, 21n37 science, Kant’s system of 8, 204n22, 229, 233,
principle of Non-Contradiction (PNC) 26, 30, 238, 240–54, 257
74–5, 85, 86n23, 88, 169, 170n48, 191 sensation 93, 180, 184, 188, 201–2, 210n40,
highest principle of ontology and logic 4, 20, 211n43, 223, 309
75, 123 ‘silent decade’ 156, 161
principle of Sufficient Reason (PSR) 20, 54n38, Smit, H. 162n32, 171n50, 207n32, 302n13
86n23, 87–9, 141, 214n60 something 2, 88, 101; see also nothing; things
Proops, I. 30n82, 36n98, 57n48, 61n57 space 102–3
absolute 249
quantifiers, see existence: as quantifier Euclidean 202, 202n18, 204
Quedlinburg unicorn v–vi form of intuition 146n50, 156, 159n22, 167,
Quine, W. V. 26n61, 38n106, 55n44, 170n48 174–5, 177–9, 182, 196, 198n3, 202–4,
205–6, 209–12, 221
ratio essendi, see ground: ratio essendi Spinoza, B. 27, 52, 55, 78
ratio fiendi, see ground: ratio fiendi spontaneity 308, 316n49
real conflict Stephenson, A. 162n28, 163n35
argument against logicism 81, 91–2 substance
definition of 81 complete concept of 18n28, 66, 67n64, 69
Kant’s account of 94–7, 111 contingent existence of finite substances 26
predicate-canceling 110n30 finite 15, 66–7, 115, 139–40
subject-canceling 110, 110n30 infinite 66; see also God
real incompatibility, see predicates: Kant’s instantiation of predicates 104, 109
argument for real incompatibility of intellectual concept of 154, 160, 177, 189
real predicates, see predicates: real interaction 110n29, 137, 139–40, 142–4,
reality 142f, 143f
conflict of, see real conflict Leibnizian monads 115–16, 173
gradable degree of being 51, 51n26, 54 plurality thesis 122, 132–5, 137–8, 140,
metaphysical structure of grounds and 142, 144
consequences 84 simple 13, 132–5
mind-independent, see noumena Spinozistic doctrine of modes of
neutral term for perfection 51 substance 55
relation to an object 158–62; see also cognition thinking 2
reason world-mates (members of a common
ends of theoretical reason 288–96 world) 115–16, 137, 139, 142
logical principle of 289, 291 see also alteration; categories: substance-
postulates of 280–8 accident; essence; predicates; properties
pure principle of 290 subsumption and judgment 78–9
rational demand for a ground of
possibility 8, 261–3, 272–3, 288n63, ‘Table of nothings’ 166–9, 168f
296, 325–6 theology 2, 27–8
receptivity 184, 308 see also God; ontological argument;
recombination, principle of 122, 138, 140, 143 ontotheology
Rosefeldt, T. 68n65 things 83, 293
things in themselves 180–4, 201, 206, 210n39,
S5 axiom 134–5 211n44, 223–7, 227n92, 249, 287n62,
Schafer, K. 161n27, 171n50, 183n89 297, 308
OUP CORRECTED PROOF – FINAL, 17/2/2016, SPi

 INDEX

thinkable, see possibility: relation to thinkability Watkins, E. 13n3, 15nn12–13, 83n11, 84n18,
time 86n23, 115nn44–5, 135n27, 139n39,
form of intuition 146n50, 156, 159n22, 167, 140n40, 208n34, 214n59, 225n88,
174–5, 177–9, 182, 196, 198n3, 202–4, 228nn1–2, 242n37
205–6, 221, 228 wholes, see parts and wholes
schematism and 309–10 Wolff, Christian 1, 3–5, 13, 13n3
Tolley, C. 164n36, 210n39, 263n8 Deutsche Logik 14n9, 23n46, 24n53
Transcendental Deduction 159, 177, Deutsche Metaphysik 14n8, 15, 17n21,
180, 184 24n53, 36n97
transcendental idealism 7, 180–2, 217, 252, logicism 15, 19–21, 27, 84, 260
304–5, 305n20 Ontologia 14
transcendental philosophy 1–4, 154 ontological argument 43, 43n1, 56, 262
a priori concept of objects in general 3 pre-established harmony doctrine 84n18
object in general as the highest concept views on existence 36n97, 56n46, 111,
of 2–3, 166–71 244, 245f 153, 293
successor to, and improvement on, views on grounds 83–4, 84n17, 86, 86n23,
ontology 166–7, 209 88, 207
views on necessity 87
understanding 81, 88, 93, 101n2, 111, 136, views on possibility 16–18, 16n20, 25, 35,
148n55, 155, 157–8, 175–80, 184–5, 191, 166–7, 169
195, 202, 205, 221–2, 230, 257, 278t, worlds, Kant’s theory of 137–8, 137n34
288–9 see also possible worlds
unicorns v–vii, 37, 203 Wundt, M. 13n3, 21n38,
101n2
Valentini, M. B. v–vi
Van Cleve, J. 39n110, 181n81, Yong, P. 82nn5–6, 107, 107n21, 108n22, 109,
193, 305n21 110n27, 119n50, 136–8, 140n41,
von Guericke, O. v 142n45, 144n47, 326