ILKKA NIINILUOTO

'/?HE A I M A N D S T R U C T U R E OF
APPLIED RESEARCH

ABSTRACT. The distinction between basic and applied research is notoriously vague,
despite its frequent use in science studies and in science policy. In most cases it is based
on such pragmatic factors as the knowledge and intentions of the investigator or the type
of research institute. Sometimes the validity of the distinction is denied altogether. This
paper suggests that there are two ways of distinguishing systematically between basic and
applied research: (i) in terms of the "utilities" that define the aims of inquiry, and (ii)
by reference to the structure of the relevant knowledge claims. An important type of
applied research aims at results that are expressed by "techical norms" (in von Wright's
sense): if you wish to achieve A, and you believe you are in a situation B, then you
should do X. This conception of "design sciences" allows us to re-evaluate many issues
in the history, philosophy, and ethics of science.

1. BASIC AND A P P L I E D R E S E A R C H -- A N E G L E C T E D
DISTINCTION

Philosophers have mostly been concerned with sciences which explain

and interpret the world; now it is time to pay attention also to sciences
which change the world.
This remark may sound a little pathetic, but it conveys an important
truth. The most influential philosophers of science in our age - both
within the analytical and hermeneutical traditions - have usually
grounded their analyses of the aims and methods of inquiry upon
models provided by such basic sciences as mathematics, physics, bi-
ology, history, psychology, and sociology. Much less attention has been
devoted to fields like applied mathematics, computer science, aeroplane
engineering, forest technology, dairy science, agricultural chemistry,
veterinary medicine, sport medicine, pharmacy, nursing science, logop-
edics, didactics, homiletics, household economics, social policy studies,
library science, military science, peace research, and future studies.
This neglect by philosophers - of course with some notable excep-
tionsl,-is both surprising and harmful in many ways.
In the first place, the students of the more exotic "practical sciences"
may of course learn important lessons about science by reading Carnap,
Hempel, Popper, Kuhn, Lakatos, Feyerabend, Laudan, Stegmtiller,

Erkenntnis 38: 1-21, 1993.

© 1993 Kluwer Academic Publishers. Printed in the Netherlands.
2 ILKKA NIINILUOTO

Ruse, Apel, and other important authorities. But it is by no means

clear that conceptions founded upon the model of the basic sciences
would do justice to the peculiar characteristics of the applied sciences.
Secondly, the converse error has been committed by the instrumen-
talist philosophers (represented by many pragmatists and Marxists,
among them Dewey, Bernal, and Habermas) 2 who treat all science as
if it were applied. An extreme expression of this trend is the claim that
the Scientific Method is to be identified with the techniques of practical
problem-solving in Operations Research. 3
Thirdly, despite its frequent use, the distinction of basic and applied
research is often presented in a vague and confusing way within science
policy. This fact - together with the fashionable instrumentalism among
policy makers - has led to an "epistemic drift", 4 whereby the category of
fundamental research seems to fading away or melting into that of
applied research.
Fourthly, the failure (or refusal) to distinguish basic and applied
research has led many historians and sociologists of science astray
in their criticisms of what they regard as the "ideology" of "pure
science" .5

2. PRAGMATIC OR SYSTEMATIC DIFFERENCE?

The standard distinction between basic and applied research was codi-
fied by OECD in 1966. 6 First, within R&D, research is defined as the
pursuit of knowledge, while development uses the results of research to
develop "new products, methods, and means of production". Secondly,
basic research is defined as "the systematic pursuit of new scientific
knowledge without the aim of specific practical application", and ap-
plied research as "the pursuit of knowledge with the aim of obtaining
a specific goal".
The distinction between research and development is systematic in
the sense that it is couched as a difference in their products: knowledge
vs. artefacts. But the basic-applied distinction is vaguely based upon
the "aims" of research - without any specification whose aims are in
question.
Most of the attempts to separate basic and applied research appeal
to pragmatic factors, i.e., to contextual features that may vary in time
and location. Examples of such factors include the knowledge of an
individual scientist ("the applied researcher has in his or her mind a
THE AIM AND STRUCTURE OF APPLIED RESEARCH

possible practical application"), personal motives ("the motive of basic

research is pure curiosity, that of applied research utility"), the inten-
tions of the funding institutions ("applied research is financed because
of its economic utility"), research sites ("basic research is done in the
universities, applied research in the polytechnics, business schools, and
industrial laboratories"), and the speed of utilization ("strategic basic
research leads to practical applications in the long run, mission-oriented
research in the short run").
Criteria of this sort are vague, ambiguous, and incompatible with
each other. 7 The same activity may be classified as basic research on
one criterion, applied research on another. It is no wonder that the
validity of these distinctions if often doubted or denied. It is, therefore,
important to ask whether they could be replaced by some non-pragmatic
division. In this paper, I look for a systematic distinction in two di-
rections: (i) the "utilities" that define the aims, progress, and rationality
of inquiry, and (ii) the structure or logical form of the knowledge
claims.

3. BASIC SCIENCE AND EPISTEMIC UTILITIES

Basic or fundamental research can be understood as the activity of the

scientific community to produce new scientific knowledge by means of
the scientific method. This knowledge should provide answers to cogni-
tive problems: it should describe, with as good justification as possible,
what the world (i.e., nature, man, culture, society) is like. The success
of this activity thus depends on the amount and the correctness of the
obtained information about the world. Basic science can thereby be
characterized as the attempt to maximize the "epistemic utilities" of
truth and information - or, as their weighted combination, truthIike-
ness. 8.
This description of basic research follows the course of scientific
realism. 9 According to realism, the primary task of basic science is
cognitive: the so far best results of science give us the elements of a
dynamically developing world view. Knowledge about the current state
and the regularities of the world also allows us to explain and understand
reality. 10
In treating truth, information, and explanatory power as epistemic
utilities, a scientific realist does not claim that they constitute the inten-
tional goals of individual scientists or the motives of funding institutions.
4 ILKKA NIINILUOTO

Rather, they are the cognitive virtues which define the success and
progress of inquiry. They also have normative force in the sense that
the arguments for or against a scientific hypothesis or theory may appeal
only to its cognitive status in the light of the available evidence - not,
e.g., to moral, political, religious, or economic factors. Indeed, the
standard methods of scientific inquiry (e.g., sampling and experimental
techniques, statistical methods) have been designed so that they tend
to promote the cognitive goals of realism.
Many historians and sociologists of science have been keen to show
that reaMife scientific work does not always fit the picture drawn by
realism. 11 These studies have given us valuable (in the realist's sense!)
new information about scientific practice. But, as a criticism of scientific
realism, they seem to involve two flaws.
First, examples of behavior violating a normative command do not
disprove the norm: for example, criminal acts do not disprove criminal
law. A norm is shown to be invalid only if its violation is not punished
by the associated sanction. 12 To disprove the normative force of the
epistemic utilities of basic science, it should be shown that their viol-
ation, when uncovered in public, does not lead to any sanction. But
this is not the case with the methodological and ethical norms of science:
if someone is found to be guilty of fraud, manipulation of data, or bias
in favor or against a hypothesis on political, racist, sexist, nationalist,
religious, or economic grounds, the credibility of his or her arguments
will be demolished or at least seriously weakened within the scientific
community.
Secondly, the arguments against the "false ideology" of "pure" or
"objective" science often involve examples of the value-laden choices
between medical technologies or social policies. ~3 However, the study
of such topics belongs to applied rather than basic science - and, as we
shall see (Section 9), there is a sense in which applied research is
not "value free" (i.e., normatively restricted to epistemic utilities and
descriptive languages) in the same way as basic research.

4. TECHNOLOGY AND PRACTICAL UTILITIES

By technology I mean the design and use of material and social artefacts
which function as tools in the interaction with and the transformation
of reality. 14 The word 'technology' may also refer to the products of
THE AIM AND STRUCTURE OF APPLIED RESEARCH 5

this activity. In this sense, 'development' is a name for science-based

technology.
Unlike the linguistic products of scientific research, tools and artefacts
are not true or false. Rather, they are intended to create new powers
and possibilities of action, and thereby to increase man's positive free-
dom. Hence, the basic "technological utility" is effectiveness relative to
the intended use (e.g., the power of an engine). Besides creating new
possibilities, the use of tools consumes resources and has intended
and unintended effects on the material and social reality. Therefore,
technologies can (and should!) be assessed also in terms of their eco-
nomical efficiency (relation of costs and effects) and their ergonomical
(man-tool relations), ecological (man-nature relations), aesthetic, ethi-
cal, and social aspects.
It is important to emphasize here the crucial difference between the
decisions to "accept" a scientific hypothesis or a new technological
tool. The decisions to develop and use, e.g., nuclear power plants,
agricultural fertilizers, or missiles means in effect the introduction of
new artificial entities in the world - and therefore they are inherently
value-laden, i.e., their rationality depends on the balance of their practi-
cal utilities and disutilities. On the other hand, even though the accept-
ance of a scientific theory (e.g., theory of evolution, theory of relativity)
may also have indirect social effects, such a theory has a truth value
independent of our opinions, interests, and negotiations.

5. APPLIED SCIENCE

Engineering sciences, agricultural and forestry sciences, medical sci-

ences, and practical social sciences are often mentioned as examples of
applied sciences. Falling between basic science and technology, they
produce new knowledge which is intended to be useful for the specific
purpose of increasing the effectiveness of some human activity. The
produced knowledge functions as a tool. Hence, the value of the results
of such applied sciences can be evaluated both in terms of epistemic
and practical utilities.
As applied science aims at knowledge, its products should be assessed
for their correctness, informativeness, and truthlikeness. But the re-
quirement of practical applicability suggests that - besides the epistemic
utilities relevant to basic science - applied science is also concerned
with simplicity or manageability. 15 The choice of the simplest among
6 ILKKA NIINILUOTO

equally well supported hypotheses, if it will be used as a resource in

human action, is often assumed to be justifiable by economic reasons.
Further, to make the calculations easier or practically possible, approxi-
mations and simplifications may be introduced to quantitative laws -
even at the expense of their truthlikeness.16
Beside epistemic utility, the knowledge provided by applied science
is expected to have instrumental value for the associated human activity.
Applied science is thus governed by what Habermas calls the "technical
interest" of controlling the world. But to extend this conception to
basic natural science as well, and thereby to treat all natural science as
it were applied, is to commit oneself to instrumentalism. 17
It might be objected that the double assessment - by epistemic and
practical utilities - is superfluous here, since there is a conceptual
connection between them. It is true that "practice is a criterion of a
theory", i.e., the pragmatic success of a theory in guiding human action
is a fallible indicator of its cognitive virtues. But the truth of a theory
of applied science implies only its potential pragmatic success: e.g., it
may happen that someone makes a theoretical proposal for an edu-
cational reform, which would have its claimed effect, but this proposal
is actually never implemented. In this sense, it is possible that there is
cognitive progress in applied science which is not, nor will be, cashed
out in practice.
The interplay of epistemic and practical utilities can also be illustrated
in decision-theoretic terms. Suppose we are interested in the health
risks of radiation, and a choice has to be made between linear and
quadratic dose-response models. 18 If the problem is purely theoretical,
then the loss of a mistaken model could be equated with its distance
from truth. But if the model will be implemented in the adoption of
safety standards, with the interest of protecting the public and workers
in nuclear facilities, then the loss function could be transformed to give
higher penalties in the direction of lower risk estimates.
Our discussion so far has not yet helped to understand the relations
between basic and applied research, and it has not clarified the crucial
issue of how a piece of knowledge may have instrumental value. To
approach these questions, the analysis of applied science in terms of its
aims or utilities is not sufficient, but it is necessary to try to uncover
the logical form of its typical products.
THE AIM AND STRUCTURE OF APPLIED RESEARCH 7

6. DESCRIPTIVE SCIENCE: EXPLANATION AND PREDICTLON

Basic sciences are descriptive in the sense that they primarily describe,
with sentences in the indicative mood, singular and general facts about
the world. They seek to establish theories which express true and
lawlike (nomic) connections between properties or types of events. A
typical result of basic research is a deterministic or probabilistic causal
law of the form
(1) X causes A in situation B
or

(2) X tends to cause (with probability p) A in situation B.

By fulfilling its descriptive function, basic research also provides the
opportunity to give scientific explanations. Assume that event of type
A has occurred in a certain situation b. Then laws (1) and (2) allow us
to construct deterministic or probabilistic explanatory arguments with
A as the explanandum:
(3) X causes A in situation B
X occurred in situation b
The situation b is of type B
Hence, A occurred in b.
Laws of form (1) and (2) can also be used for prediction. Suppose
an event of type X has occurred in a situation b of type B. Then the
occurrence of A in this situation can be predicted (certainly or with
probability p). The structure of this predictive argument is again (3).19
For example, the same laws of celestial mechanics can be used for both
explaining past eclipses and predicting future ones.
Predictive sciences are often considered the basic type of applied
science. Human cultures have been interested in successful prediction
for various practical reasons. A scientific theory, which is able to pro-
duce reliable predictions about future events, has predictive power.
Practical astronomy, meteorology, and social statistics are examples of
applied sciences which have predictive power as their central epistemic
utility.
In spite of the structural similarity of explanatory and predictive
arguments (cf. (3)), there are good explanatory theories without much
predictive power (e.g., theory of evolution). Further, some useful
8 ILKKA NIINILUOTO

"models" for prediction may be so simplified or idealized that they are

not taken seriously as premises of explanation. Still, explanatory and
predictive sciences are two subtypes of descriptive science - with a
difference which is only pragmatic in nature.

7. DESIGN SCIENCE
It would be a serious mistake to generalize the observations of Section
6 to the claim that the basic-applied distinction coincides with the
explanatory-predictive distinction. This move is seductive, since the
descriptive account has almost universally been accepted to be the
model of science.
Some - not very successful - attempts have been made to formulate
alternative views, which would allow also for a special kind of "norma-
tive science" or "critical science". 2° But the supporters of the descrip-
tive ideal have liked to knock out these proposals, since they seem to
openly bring or to smuggle into inquiry moral or political valuations.
An important exception is Herbert Simon's insightful book The Sci-
ences o f the Artificial (1969). He argues that the traditional model of
science (and science education) gives a misleading picture of such fields
as engineering, medicine, business, architecture, painting, planning,
economics, education, and law, which are concerned with design - i.e.,
not how things are, but "how things ought to be in order to attain goals,
and to function". 21
Let us say that design in the broad sense includes all "artificial"
human activities, i.e., the production, preparation, or manipulation of
natural systems (e.g., human body, forest) or artefacts (e.g., an aero-
plane, city, legal order). 22 This concept of design thus ranges from
environmental, economic and social planning to engineering, architec-
ture, industrial design, crafts, and the fine arts. Then research aiming
at knowledge that is useful for the activity of design - i.e., enhances
human art and skill (Greek techne, Latin ars) 23 - may be called design
science.
Before discussing the nature of design science in more detail, it is
important to distinguish it clearly from scientific design. (I think Simon's
account is ambiguous here.) Scientific design is a species of design, i.e.,
the activity of solving design problems by using scientific methods and
scientific knowledge. Operations Research (OR) provides methods for
finding optimal or satisfactory solutions to design problems (e.g., game
THE AIM AND STRUCTURE OF APPLIED RESEARCH 9

PROFESSION PRACTICE ART [ SCIENCE

medicine man therapy medicine medical science
physician healing
nurse nursing art of nursing nursing science
pharmacist preparation of medical pharmacy pharmacology
drugs
farmer farming art of f a r m i n g agricultural science
agriculture
engineer design of mechanical engineering engineering science
works
soldier warfare strategy military science
art of war
? peace-making ? peace research
politician politics politics political science
administrator
social worker social service social policy policy science
merchant commerce art of trading economics
tradesman trade
teacher teaching didactics didactics
athlete sporting athletics sport science

Table 1.

theory, decision theory, linear programming). In this sense, scientific

design is the result of the "scientification" of art, technology, manage-
ment, or development.
On the other hand, design science is the activity of generating instru-
mental knowledge for the production and manipulation of natural and
artificial systems. Design science produces knowledge which may then
be applied within scientific design.
More generally, it is important to distinguish applied science from
the applications of science. The former is a part of knowledge produc-
tion, the latter is concerned with the use of scientific knowledge and
methods for the solving of practical problems of action (e.g., in engin-
eering or business), where a scientist may play the role of a consultY
These distinctions can be clarified by Table 1, which separates from
each other a profession (e.g., a farmer), the related practice (e.g.,
farming, agriculture), art or skill needed in this practice (e.g., the art
10 ILKKA N I I N I L U O T O

"scientificafion" /4 activity
practice
art

. . . ~ - - - - - 7---everyday ]
J
" / experien'cex', I
/
/
I
I !
~' design /
x ,, science / "mechanization"

Fig. 1.

of farming), and a design science aiming at improving the art (e.g.,

agricultural science).
Sometimes one and the same person may simultaneously act in a
profession and do research about his or her own practice: a physician
may heal patients, keep record of this activity, and write a doctoral
dissertation using its results. But it is nevertheless possible to make a
conceptual distinction between the practice of an art and the research
of the associated design science. The fact that today sport science
studies the sliding properties of skies, the effective training of male and
female skiers, and the best skiing styles, does not mean that the skiing
champions suddenly become "scientists". 25
Another way of illustrating the conception of design science is given
in Fig. 1. Already Plato was well aware that some arts require more
background knowledge and exactness than others. In the dialogue Phi-
lebus (55d-56c), he pointed out that some crafts use "numbering, meas-
uring, and weighing" (e.g., the building of ships and houses employs
instruments like straight-edge, compass, and plummet), while others
are based on "guesswork" or "experience and rule of thumb, involving
the use of that ability to make lucky shots which is commonly accorded
the title of art or craft, when it has consolidated its position by dint of
industrious practice". The latter class of arts include, Plato said, medic-
ine, navigation, warfare, and music.
Originally all human arts or technologies have been based upon
expertise consisting of the practical skills and "rules of thumb" of the
THE AIM AND STRUCTURE OF APPLIED RESEARCH 11

masters: they have presupposed or employed knowledge only at the

level of everyday experience. Their later development has followed two
patterns which may be called the mechanization and the scientification
of practices or arts. 26 First, human activities have become more and
more effective through new mechanical inventions (e.g., tools of war-
fare, agriculture, architecture, sport). Secondly, knowledge which
serves some art has been collected into systematic bodies of rules. This
process started already in the ancient times with the arts of counting,
measuring, warfare, medical care, arguing, building, and judging - with
the emergence of the first guide books of arithmetic, geometry, military
strategy, medicine, logic, architecture, and law. Later the operation of
such rules is put in scientific tests - and a design science is created
through the "scientification" of the background knowledge serving a
practice (cf. Table 1). In this sense, the treatment of rules for geomet-
rical constructions in Euclid's Elements is a classical exposition of design
science. Pharmacology is an example of an empirical design science
which tests the efficacy of medical drugs.
These observations give us new insight into the history of science.
The standard view that all scientific disciplines have emerged from
philosophy may be true for some basic sciences (physics, biology, psy-
chology, sociology), but many practical disciplines have been created
through the scientification of professional activities. 27 This process con-
tinues in our age with new disciplines - such as nursing science - which,
I believe, gain illumination about their identity by the conception of
design science.

8. TECHNICAL NORMS

If a design science is expected to contribute to the scientification of

human practices, its results should be some kinds of rules of action
(Fig. 1). But rules, as normative statements in the imperative mood,
are usually thought to lack truth values. How could they, then, at the
same time constitute knowledge?
The solution can be found in G. H. yon Wright's concept of technical
norm, which is a factual statement about the relation between means
and ends:
(4) If you want to make a hut habitable, you ought to heat it.
Both the antecedent and the consequent of such a conditional norm
12 ILKKA NIINILUOTO

are descriptive statements, the former about the wants or preferences

of a person, the latter about the existence of a "technical ought" for
him. The technical norm (4) is true if and only if heating the hut is
necessary condition for making it habitable. 2s
More generally, a technical norm is a statement of the form
(5) If you want A, and you believe that you are in a situation B,
then you ought to do X.
The categorical normative statement, which corresponds to the conse-
quent of (5), may have stronger or weaker forms
(6) You should (ought to) do X.
It is rational for you to do X.
It is profitable for you to do X.
The conditional norm or recommendation (5) (with alternative conse-
quents given by (6), respectively) can be defined to be t r u e if and only
if doing X is
(7) a necessary cause of A
a sufficient cause of A
probabilistic cause of A
in situation B. 29
If (6) is read in the imperative mood, it has no truth value. If (6) is
treated as a descriptive statement, which is a consequence of "practical
inference", its truth presupposes some valuation as a premise ('you
want A'). Therefore, singular statements of the form (6) can hardly be
proposed to be the theorems of any science. However, the case with
technical norms is different: the truth or falsity of (5) is an "objective"
and general feature of the world, which does not presuppose any com-
mitment to the valuation in the antecedent of (5).
Hence, we may propose that technical norms of the type (5) express
the typical structure or logical form of the knowledge provided by
design science. Examples from sciences of Table 1 are easy to find:
If you want to heal a patient with these symptoms, you
should use this treatment.
If you want to increase the productivity of fields, you should
use these fertilizers.
If your aim is to build safe aeroplanes, use this material.
THE AIM AND STRUCTURE OF APPLIED RESEARCH !3

If we wish to increase the probability of peace (reduce the

risk of war), a disarmament programme ought to be ac-
cepted.
If we want to avoid unemployment, we should lower interest
rates.

There are two ways in which a technical norm may be supported, so

to speak, from above and from below.
Support "from above" means the derivation of a technical norm from
descriptive statements provided by basic research. Bunge (1966) has
given a good example of this process:

Magnetism of iron disappears above the temperature 770°C.

$
If the temperature of iron exceeds 770°C, it is not magnetic.
+
If a magnetized piece of iron is heated above 770°C, then it
is demagnetized.

In order to demagnetize iron, heat it above 770°C.

More generally, assume that basic research has established a causal

law of the form (1) or (2), 'X causes A in situation B'. If the cause
factor X is not manipulable by us, but is chosen by nature, then this
law can be used for predictions: if we observe X in situation B, we may
expect A as well. But if X may be chosen by us, then the causal law
can be converted to a technical norm (5): if we want to achieve the
aim A, and the situation is of type B, then we should bring about the
cause X.
Design science can, in this precise sense, be applied science: its
knowledge is derivable from the results of the descriptive basic sciences.
In many cases, however, there is not available any general theory
from which a technical norm can be deduced. Then technical norms
are supported "from below" by building up a simplified model of the
situation, using trial-and-error procedures and experimental tests to
investigate the dependences between the most important variables, and
trying to find the optimal methods of producing the desired effects.
When the result is expressed as a general rule, a technical norm with
some empirical support is obtained.
14 ILKKA NIINILUOTO
descriptive science

technology
predictive design development
science science scientific problem-solving
I I I I
basic
research applied
research Fig. 2.

9. DESIGN S C I E N C E , ETHICS~ POLITICS

Design sciences differ from explanatory basic sciences and predictive

applied sciences in a systematic way, because technical norms are not
descriptive statements about the world. They don't tell us what is, was,
or will be, but what ought to be so that we can attain given goals.
Nevertheless, they are true or false statements, and can, in principle,
be supported by theories and experiments in an objective fashion. These
conclusions are summarized in Fig. 2, which in a sense suggests that
the old basic - applied dichotomy is less fundamental than the descrip-
tive - design distinction.
It should be emphasized that the border between descriptive and
design science splits many scientific disciplines, a° Let S be some activity
which can be studied by science, e.g., S might be farming, nursing - or
science itself which is the object of "science studies" .31 Then descriptive
research of S includes at least the history of S, the psychology of S, the
sociology of S, and the economics of S. Basic research about S tries to
describe the present state of S and to establish some systematic regu-
larities about S - in this way, we may speak about basic research
within technical sciences, life sciences, medicide, social sciences, and
jurisprudence. Design science contains only a part - the practical kernel,
so to speak - of these discipines.
In one sense, however, we have found that the border between
predictive and design science may depend on the pragmatic question
about the human manipulability of causal factors. Astronomy and
meteorology are today predictive sciences, since their regularities can-
not be transformed to useful technical norms - the rule 'If you want
an eclipse, place moon between the Sun and Earth' is irrelevant in
relation to human possibilities. But scientists are already speculating
about making planets habitable. Whether the causal factor X in a law
THE AIM AND STRUCTURE OF APPLIED RESEARCH 15

'X causes A in B' is manipulable by us depends on the stage reached

by human technology.
This conclusion already suggests that, unlike basic research, design
science has to satisfy a special condition of social relevance. A technical
norm (5), where X can be brought about by us, is useful for practical
purposes only upon two further conditions: (i) there are in fact situ-
ations that are (excatly or at least approximately) of type B, and (ii)
the goal A should be at least potentially acceptable for some social
group or "auditory".
As technical norms contain evaluative and normative terms, design
science seems to radically differ from the "positivistic" ideal of "value-
neutral" science. But as a conditional statement, a technical norm does
not require a commitment to the value premise of its antecedent. For
example, a militarist and a pacifist may both agree on the truth of the
conditional recommendations of military studies, even if they sharply
disagree about their relevance. A technical norm is binding only for
those who accept their conditional value premise. 32
At the same time, in spite of value-neutrality in this sense, a person
contributing to applied science is morally responsible for the tools (tech-
nical norms) he or she has created.
How is the value A in a technical norm chosen? In the case of
some design sciences, the answer is straightforward: medicine aims
at promoting health, business economics at maximizing profits, peace
research tries to reduce the risk of military conflicts, and social policy
studies promote the welfare of society. Aims of this kind are so tightly
fixed with the professional goals and the self-identity of these sciences
that they often present their conclusions categorically, as if forgetting
the overall value commitment. Still, the content and desirability of
these aims may be put into question within philosophy and ethics:
the philosophy of medicine discusses the definition of health, social
philosophy the "health" of society, business ethics and environmental
ethics the legitimate goals of human actions towards other agents or
nature.
For example, as medicine takes it for granted that a patient must be
healed, technical norms of the form 'If you want to heal a person
with symptoms B, use treatment X' may be expressed simply by 'Use
treatment X for patients with symptoms B'. Such a move becomes
problematic only if it is denied that the "symptoms" B constitute a
"disease" (e.g., drug use, mental abnormality).
16 ILKKA NIINILUOTO

In many cases, the choice of A involves a political debate. Policy

issues, where the experts differ in their recommendations, can be ana-
lyzed into their elements by the concept of technical norm. If two
groups of experts advocate conflicting policies 'Do XI!' and 'Do X2!',
respectively, their disagreement may be due to at least three separate
reasons:
(a) disagreement about the relevant goals A1 or A2
(b) disagreement about the current situation B~ or B2
(c) disagreement about the underlying causal mechanism be-
tween X & B and A.33
Here (b) and (c) are factual disputes, in principle solvable by empirical
research, but (a) is a difference about valuations. Disputes about poli-
cies concerning energy, environment, and society often contain both
factual and evaluative assumptions - and the latter tend to be concealed
under the guise of "neutral" experts. 34
There is one further, rather subtle way in which a technical norm
may involve valuations. In making a distinction between the manipul-
able factors X and the unchanging situation B, we in fact bring in an
assumption that B must be kept constant. 35 For example, an economic
policy recommendation in a capitalist (resp. socialist) country may
presuppose that the economic system itself must not be interfered with.
Therefore, to eliminate such hidden valuations, a technical norm should
be formulated so that the goal A includes all the relevant value assump-
tions (e.g., A = t o improve gross national product, to preserve the mar-
ket economy, and to avoid such and such side effects).

10. CONCLUDING REFLECTIONS

As it is not true that "the aim justifies the means", a technical norm
should include among its antecedent A all the relevant valuations that
concern the direct and indirect consequences of the recommended ac-
tion X. That this requirement has not always been respected, and
applied science has been utilized with a very narrow scope and distorted
content of human valuations (usually only technical and economic effi-
ciency in the short run), has led to a justified criticism of the dangers
of "instrumental reason". 36
However, the "legitimation crisis" of modern science does not imply
the need to reject the idea of basic science or to invent a new type of
THE AIM AND STRUCTURE OF APPLIED RESEARCH 17

science, with aims and standards differing from other forms of inquiry. 37
The concept of design science covers in fact a whole variety of different
possibilities. They include cases of blind "technocraticapproach (where
the scientist accepts uncritically the goal A, without questioning it or
without understanding his or her moral responsibility in producing
tools for reaching A), piecemeal social engineering (where the goal A
proposes only small reforms to the social system), and emancipatory
research (where the goal A is critical of the status quo and proposes
radical, even utopian changes in the prevailing order).
The last point illustrates the fact that, in a social design science, the
goal A need not be a demand imposed "from above" by a bureaucratic
planning officer over the citizens. Instead, it may express the "we-
intention" of a democratic community. The concept of design science
thus covers also the so-called participatory planning or action research,
where the researcher goes to live with his or her "clients" and helps
them elicit their own preferences. It also thereby shows that the "in-
crementalist" criticism of planning theories (i.e., the alleged impossi-
bility of separating subject and object) is not a sufficient reason for
rejecting the idea of instrumental rationality. 3s
But, on the other hand, the concept technical norm shows how
extremely difficult applied science may be. For many systems (e.g.,
economy, technosystem) it may be very hard to find any approximately
true and lawlike regularities. For some systems involving both material
and human elements (e.g., a city), it may be an immensely complex
affair to give a correct and sufficiently detailed description of its present
state. And for many situations (e.g., animals, human patients) it may
he highly controversial what goals and means are legitimate from a
moral point of view.
Is the attempt to establish design sciences, in spite of all these diffi-
culties, still a worthwhile enterprise? Many professions think so today,
obviously with hope that the making of their practice and education
"scientific" would given them a higher status in society.
The real challenge to design science seems to come from the argu-
ments of Hubert L. Dreyfus, whose criticism of artificial intelligence
and expert systems is directly applicable to know-how represented by
technical norms. Dreyfus argues that skill based upon rules be!ongs
only to "novices" and "advanced beginners", while the true "expert"
acts by "intuitive intelligence" without reliance on action-guiding
rules. 39
18 ILKKA NIINILUOTO

The important issue, which remains to be settled in further philosoph-

ical and empirical work, is to analyse and to classify human skills into
those which can, or cannot, be improved by their "scientification".

NOTES

1 In my view, the most interesting account is due to Herbert Simon (1969) (cf. Section
7 below). Useful discussions have been presented by Polish praxeologists (see Gasparski
et al., 1983) and analytical philosophers of technology (see Rapp, 1974). A special issue
on applied science was published in Synthese 81:3 (1989). The work on the Finalization
Thesis (cf. Schiller, 1983) seems to me inconclusive (cf. Niiniluoto, 1984a, Ch. 10). My
own ideas about applied science have been developed since 1983. See Niiniluoto (1984a),
Chs. 10-12, (1984b), (1985b), and the Finnish papers (1985a) and (1987b).
2 For Bernal, see Niiniluoto (1990a).
3 See Ackoff (1962) and the criticism in Niiniluoto (1984a), Ch. 11.
4 This nice term is due to Elzinga (1985). An example of this drift is the distinction
within basic research that Irvine and Martin (1984) make between pure curiosity-oriented
research and strategic research, where the latter provides the knowledge base for tomor-
row's technologies. This terminology does not appreciate the fact that curiosity-oriented
basic research is also "strategic" relative to the cognitive goals of inquiry.
5 A recent example is Latour (1987), who operates with many - surprisingly sharp -
Janus-faced dichotomies, but finds no difference between science and technology.
6 Cf, Sintonen (1990).
7 After listening to my queries about the basic-applied distinction, Commission of Basic
Research (appointed by the Ministry of Education in Finland in 1989) decided to include
within "basic research" all publicly funded research (about 40 percent of the R&D
volume in Finland).
8 See Levi (1967), Niiniluoto (1987a).
9 See Popper (1963), Niiniluoto (1984a).
10 Some realists would take explanation to be more fundamental notion than truth. Cf.
Leplin (1984) and Tuomela (1985).
11 See, e.g., Mulkay's (1979) summary of the criticism against the Mertonian norms of
science.
12 For norms, see von Wright (1963, 1983).
13 See, for example, the interesting example of research on drug use in Restivo and
Loughlin (1987).
14 Cf. Skolimowski (1966), Rapp (1974), Mitcham and Mackey (1983), Niiniluoto
(1984a), Ch. 12.
15 See Rescher (1990) and Niiniluoto (1992).
16 See Niiniluoto (1984a), p. 262.
17 Cf. Habermas (1971) and Niiniluoto (1984a), p. 221.
18 See Longino (1989).
19 Cf. Stegmiiller (1969) on the symmetry between explanation and prediction.
THE AIM AND S T R U C T U R E OF A P P L I E D R E S E A R C H 19

20 Cf. Habermas (1971) and Held (1980). Note that a science a b o u t the norms valid in
a given society (e.g., legal dogmatics) may be descriptive. See Niiniluoto (1985b).
21 See Simon (1982), p. 7. For comments on "design" in the narrower sense of industrial
design, see Niiniluoto (1984b).
22 For such a broad concept of design, see Bunge (1979). However, I don't assume with
Bunge that technology always is based upon science.
23 For the concepts of techne and ars, see Mitcham (1979).
24 This distinction is denied by the view L. J. Savage called "behaviouralism": to accept
a scientific hypothesis is always a decision to act as i f the hypothesis were true. For a
criticism of behaviouralism, see Levi (1967).
z5 But it might be mentioned that the gold medalist of javelin at the Olympic Games in
Soeul (1988), Tapio Korjus, wrote his Master's Thesis in physical education about javelin
throwing.
26 Cf. Niiniluoto (1984b).
27 I am not suggesting that such practical disciplines became sciences before the birth of
theoretical science in Greece: the scientification of a practice presupposes that science
(as a method) has already been invented.
28 See yon Wright (1963, 1988).
29 A complication with the weaker conclusions in (6) arises from the problem that
concepts such as 'rational' and 'profitable' can be defined by several different formal
criteria. Sharper formulations of (6) and (7) could be given in (Bayesian) decision theory,
where a technical norm is generalized to a statement of the form 'If your value system
is V, your belief system is B, and your favourite decision criterion is C, then you ought
to do X'.
3o My own experience is that some scientists fear that the idea of design science is too
much tied with instrumental rationality or technocratic values. I am n o t by any means
suggesting that disciplines such as future studies, library science, or nursing science are
entirely reduced to design science, since these activities should also be investigated by
philosophy, history, psychology, sociology, etc.
31 The design science corresponding to scientific research (as a subsystem of "science"
as a whole) is nothing else than "methodology". Laudan (1987) has suggested that
methodological norms in science are "hypothetical imperatives" (i.e., technical norms).
Another design science related to science is science policy studies.
32 Cf. Niiniluoto (1985b) for the relevance of this observation to auditory-relative concep-
tions of truth.
33 A fourth factor is the choice of the criterion of rationality (e.g., Bayes-rule, minimax).
See Levi (1980), Appendix.
34 For the suggestion that the so-called "technological imperatives" are really technical
norms, with a hidden value antecedent, see Niiniluoto (1990b).
3s A remark to this effect has been made, in the context of ecological theories, by Dr.
Yrj6 Haila.
36 See Held (1980) and yon Wright (1986).
37 See, e.g., Restivo and Loughlin (1987) on the new standards of "validity" in the
"science for the people" movements.
38 Cf. Lindblom and Cohen (1979).
39 See Dreyfus and Dreyfus (1986).
20 ILKKA N I I N I L U O T O

REFERENCES

Ackoff, R. L. (with collaboration of S. K. Gupta and J. S. Minas): 1962, Scientific

Method: Optimizing Applied Research Decisions, John Wiley, New York.
Bugliarello, G. and Doner, D. B. (eds.): 1979, The History and Philosophy of Technol-
ogy, University of Illinois Press, Urbana.
Bunge, M.: 1966, 'Technology as Applied Science', Technology and Culture 7, 329-349.
(Reprinted in Rapp (1974), pp. 19-36.)
Bunge, M.: 1979, 'The Philosophical Richness of Technology', in Bugliarello and Doner
(1979), pp. 262-281.
Dreyfus, H. L. and S. E.: 1986, Mind over Machine: The Power of Human Intuition and
Expertise in the Era of the Computer, Blackwell, Oxford.
Elzinga, A.: 1985, 'Research, Bureaucracy and the Drift of Epistemic Criteria', in B.
Wittrock and A. Elzinga (eds.), The University Research System, Almqvist & Wiksell,
Stockholm.
Gasparski, N. et al.: 1983, 'Contemporary History of Design Science', Praxeology Year-
book 3, 147-156.
Habermas, J.: 1971, Knowledge and Human Interest, Beacon Press, Boston.
Held, D.: 1980, Introduction to Critical Theory, University of California Press, Berkeley.
lrvine, I. and Martin, B.: 1984, Foresight in Science: Picking the Winners, Francis Pinter,
London.
Latour, B.: 1987, Science in Action, Open University Press, Milton Keynes.
Laudan, L.: 1987, 'Progress or Rationality', American Philosophical Quarterly 24, 19-
31.
Leplin, J. (ed.): 1984, Scientific Realism, University of California Press, Berkeley.
Levi, I.: 1967, Gambling With Truth, Alfred A. Knopf, New York. (2nd ed. The MIT
Press, Cambridge, Mass. 1973.)
Levi, I.: 1980, The Enterprise of Knowledge, The MIT Press, Cambridge, Mass.
Lindblom, C. E. and Cohen, D. K.: 1979, Usable Knowledge: Social Science and Social
Problem Solving, Yale University Press, New Haven.
Longino, H. E.: 1989, 'Biological Effects of Low-Level Radiation: Values, Dose-Re-
sponse Models, Risk Estimates', Synthese 81, 391-404.
Mitcham, C.: 1979, 'Philosophy and the History of Technology', in Bugliarello and Doner
(1979), pp. 163-201.
Mitcham, C. and Mackey, R. (eds.): 1983, Philosophy and Technology, Macmillan, New
York.
Mulkay, M.: 1979, Science and the Sociology of Knowledge, Allen & Unwin, London.
Niiniluoto, I.: 1984a, Is Science Progressive?, D. Reidel, Dordrecht.
Niiniluoto, I.: 1984b, 'Philosophical Perspectives on Design', in S. Vihma (ed.), De-
signforschung, Design Research, University of Industrial Arts, Helsinki, pp. 13-31.
Niiniluoto, I.: 1985a, 'Edistyminen soveltavissa in tieteissfi', in E. Kaukonen et al. (eds.),
Tieteen historia ja tieteen edistyminen, Suomen Akatemian julkaisuja 2, Helsinki, pp.
169-192.
Niiniluoto, I.: 1985b, 'Truth and Legal Norms', in N. MacCormick et al. (eds.), Con-
ditions of Validity and Cognition in Modern Legal Thought, ARSP, Beiheft 25, Franz
Steiner Verlag, Wiesbaden, pp. 168--190.
THE AIM AND S T R U C T U R E OF A P P L I E D R E S E A R C H 21

Niiniluoto, I.: 1987a, Truthlikeness, D. Reidel, Dordrecht.

Niiniluoto, I.: 1987b, 'Soveltavat tieteet tieteenfilosofian n~ik6kulmasta', in Academia
Scientiarum Fennica - Year Book 1986-1987, Helsinki, pp. 137-142.
Niiniluoto, I.: 1990a, 'Science and Epistemic Values', Science Studies 3:1 (1990), 21-25.
Niiniluoto, I.: 1990b, 'Should Technological Imperatives be Obeyed?', International Stud-
ies in the Philosophy of Science 4-2, 181-189.
Niiniluoto, I.: 1992, 'Descriptive and Inductive Simplicity', forthcoming the Proceedings
of the Carnap-Reichenbach Symposium, Konstanz, May 1991.
Popper, K.: 1963, Conjectures and Refutations, Routledge and Kegan Paul, London.
Rapp, F. (ed.): 1974, Contributions to a Philosophy of Technology, D. Reidel, Dordrecht.
Rescher, N. (ed.): 1990, Aesthetic Factors in Natural Science, University Press of America,
Lanham.
Restivo, S. and Loughlin, J.: 1987, 'Critical Sociology of Science and Scientific Validity',
in Knowledge: Creation, Diffusion, Utilization, vol. 8:3, pp. 486-508.
Schiller, W. (ed.): 1983, Finalization in Science, D. Reidel, Dordrecht.
Simon, H.: 1969, The Sciences of the Artifciat, The MIT Press, Cambridge, Mass. (2rid
ed. 1982.)
Sintonen, M.: 1990, 'Basic and Applied Sciences-Can the Distinction (Still) Be Drawn?',
Science Studies 3:2 (1990), 23-31.
Skolimowski, H.: 1966, 'The Structure of Thinking in Technology', Technology and
Culture 7, 371-383. (Reprinted in Rapp (1974), pp. 72-85.)
Stegmiiller, W.: 1969, Wissenschafiliche Erkliirung und Begri~ndung, Springer-Verlag,
Berlin.
Tuomela, R. : 1985, Science, Action, and Reality, D. Reidel, Dordrecht.
yon Wright, G. H.: 1963, Norm and Action, Routledge and Kegan Paul, London.
von Wright, G. H.: 1983, Practical Reason, Blackwell, Oxford.
von Wright, G. H.: 1986, Vetenskapen och fOrnufiet, S/Sderstr~m, Borg~.

Manuscript submitted September 24, 1991

Department of Philosophy
University of Helsinki
Unioninkatu 40 B
00170 Helsinki
Finland