Jens Høyrup

The Art of Knowing

An Essay on Epistemology in Practice
Lecture notes

Filosofi og Videnskabsteori
på Roskilde Universitetscenter
1. Række: Enkeltpublikationer 1995 Nr. 1
 CONTENTS

I. INTRODUCTORY OBSERVATIONS . . . . . . . . . . . . . . . . . . . . . . 1
Philosophy and the problem of knowledge (3)

II. A PIAGETIAN INTRODUCTION TO THE

GENERAL PROBLEM OF KNOWLEDGE . . . . . . . . . . . . . . . . . 6
Schemes and dialectic (9); The periods (13); Supplementary
observations (23); The status of schemes and categories (28)

III. THE NATURE AND DEMARCATION

OF SCIENTIFIC KNOWLEDGE . . . . . . . . . . . . . . . . . . . . . . . . . 31
A pseudo-historical introduction to some key concepts (33);
Empiricism and falsificationism (39); Instrumentalism and truth
(45); Instruments or models? (50)

IV. A NEW APPROACH: THEORIES ABOUT THE SCIENTIFIC

PROCESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Popper and Lakatos: theories or research programmes? (54);
Theories falsified by theories (66); The limits of formalization
(71); Kuhn: Paradigms and finger exercises (75); The structure
of scientific development (82); Collective and individual know-
ledge (90); Two kinds of “logic” (93); Objections and further
meditations (94)

V. TRUTH, CAUSALITY AND OBJECTIVITY . . . . . . . . . . . . . . . . 101

Truth (102); Causality (107); Objectivity, subjectivity, and parti-
cularism (115)

VI. THE ROLE OF NORMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

Logic and norms (120); Explanations of morality (123); Morality,
language and social practice (126); Knowledge, norms and
ideology (129); Value relativism and value nihilism (132);
Institutional imperatives (134); Theoretical versus applied
science (140); Further norms, contradictions, contradictory
interpretations (142)

v
VII. THE THEORY OF INTERDISCIPLINARY
AND APPLIED SCIENCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
Know-how and know-why (147); The acquisition of theoretical
knowledge (149); The “Scientific-Technological Revolution”
(154); Interdisciplinarity (158); Interdisciplinarity in basic
research (163)

VIII. ART AND COGNITION . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

Knowing about art (166); Knowing in art (168); Fresh eyes (171);
Form versus contents (174); Gelsted and Epicurus (176); Art as
thought experiments (180); “Realism” (183); Synthetical under-
standing and practical knowledge (186)

IX. REFERENCES AND BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . 191

vi
 I. INTRODUCTORY OBSERVATIONS

The following pages are lecture notes for the second part of a university
course on the philosophy of human science. The first part of the course
dealt with the distinctive characteristics of the humanities – such features
as constitute their particular identity. Citations in the following of [Høyrup
1993] refer to the notes for this first part. The present, second part
concentrates on the complementary aspect: those features which the
humanities share with other sciences, the features which characterize the
humanities as well as biology, physics, economics and sociology as
sciences1. But even scientific knowledge and cognition share a number of
characteristics with other kinds of human knowledge and cognition – not
least the quality of being a less direct rendition of that reality which we
know about than we tend to think in naïve everyday life2. Critical
understanding of the general properties and categories of knowledge,

1
I remind of the definition of the concept of a “science” which was set forth in
[Høyrup 1993: 4]: A socially organized and systematic search for and transmission of
coherent knowledge in any domain. Speaking in the following of the humanities as
sciences is thus not meant as an implicit claim that they should emulate the natural
sciences – but rather that all sciences share a number of qualities of which some
have mostly been discussed in relation with the natural sciences and others in the
context of the humanities or the social sciences. Illustrations will, accordingly, be
taken from all three domains.
2
It should be noticed that this distinction between scientific and other kinds of
knowledge presupposes that knowledge is more than explicitly formulated theory.
Knowledge, we may say briefly, is any kind of conscious or subconscious acquaintance
of the surrounding world allowing some kind of adequate action – communicative action
included.

1
moreover, will throw light on certain problems belonging more or less
specifically to the humanities.
In brief survey, the notes are built up as follows: Chapter II introduces
to the general problem of what knowledge is via a presentation and
discussion of Piaget’s “genetic epistemology”. Chapters III to V develop
what could be called a “philosophy of science” in the habitual sense,
starting in Chapter III with a general presentation of some basic categories
and some classical points of view – Platonic and Aristotelian realism,
empiricism and Popperian falsificationism, instrumentalism, and the
“demarcation problem”. Chapter III also comprises a first confrontation
with what I have chosen in accordance with the philosophical tradition
to designate a “materialist” notion of truth (but which many contemporary
philosophers would call instead a “realist” concept). Chapter IV contains
a presentation and critical discussion of two main approaches to the
problem of scientific development: Lakatos’s concept of dynamic “research
programmes”, and Kuhn’s theory of progress through a sequence of
“normal science” phases separated by “scientific revolutions”. Chapter V,
footing on the discussions of Chapters III and IV, confronts and connects
three classical core problems of the epistemology of science (and of
epistemology in general): the questions of truth and objectivity, and the
notion of causation.
Already Chapter IV considers the acquisition of scientific knowledge
as the product of a scientific community. Chapter VI unfolds some
consequences of this approach, in particular under the perspective of norm
or value systems (“morality”) as regulators of the functioning of social
communities. In the process of doing so, it develops a general view on what
norm systems are.
Chapter VII takes up the historically-concrete making of scientific
knowledge under the conditions of the “scientific-technological revolution”.
Starting out from a discussion of the relation between “theoretical” and
“applied” knowledge, it returns to the Kuhnian cycle, viewing it now
specifically as a description of the development of scientific disciplines,
and contrasts it with the inherent interdisciplinarity of applied knowledge.
Chapter VIII appears to abandon the philosophy of science altogether.
Its central problem is indeed whether, and in which sense, art constitutes

2
a way of knowing, drawing for this on the epistemology which was
developed in previous chapters. The conclusion, however, is that art plays
a central role as training of the skill in synthetical judgment without which
analytical thought is useless, in science as in any other domain.
It goes by itself that the exposition owes much to many precursors,
from Aristotle and Kant onwards. Some of those by whom I am inspired
I have read in original, from others I borrow indirectly (and certainly often
without knowing so). It must be emphasized, however, that the essay is
not meant as a survey of the views of select philosophers and philosophi-
cally minded historians, psychologists and sociologists. Instead I have
attempted to formulate a personal synthesis, while keeping it so open that
readers will still be allowed to get a broader view of influential opinions
and important problems and to agree or disagree with the single strands
of the argument.
The argument is indeed a complex network containing many threads
and open suggestions. Even though the underlying thought is certainly
rationalist, the ideal of rationality which forms its basis is that of dialogue
and not the absolutist ideal of the strict proof. As a hint to the reading it
may therefore be added that the footnotes are just as important as the main
text. They contain further reflections, objections, qualifications, or they
function as a device which allows a branching of the argument. Some of
them contain material which is essential in subsequent parts of the essay.
They should not be skipped.
All translations into English are mine and made from the original
language if nothing else is stated. If a translation into another language
is referred to, I am responsible for the retranslation.

Philosophy and the problem of knowledge

At least since the pre-Socratic philosophers Parmenides and Zeno, the
“problem of knowledge” has haunted philosophy; Zeno, in particular, is
famous for a number of paradoxes meant to show that our naïve everyday
“knowledge” cannot correspond to genuine reality – we “see” the arrow
reach its target and “know” that Achilles takes over the tortoise. But the
intellect demonstrates clearly, according to Zeno, that this cannot possibly

3
be true. Plato, probably following Socrates on this account, argued that we
cannot come to learn what we do not know already, and developed his
doctrine of ideas on this foundation. Aristotle tried to put things straight
by distinguishing different kinds of reality (“particulars” and “universals”,
in the language of his Medieval followers – to be explained in more detail
below) and different kinds of knowledge. Thereby Aristotle set the stage
for the discussion as it took place until Kant, in his “critical” approach,
introduced a distinction between the conditions which delimit and determine
our possibilities of knowing and the properties of that reality which we strive
to know about, the famous “thing in itself” as distinct from “the thing as
it appears to us”: We cannot know (material) reality without categorizing
it into objects, time, space, and causality. Whether reality in itself is
structured that way will forever remain undecided and undecidable.
Epistemology (the theory of knowledge, of episteme; alias gnoseology,
theory of gnosis) does not end with Kant; nor does Aristotle’s “setting the
stage” imply that philosophers followed his doctrines until the late
eighteenth century. But some way or other all philosophical discourse in
the field has been concerned with the relevance or irrelevance of Plato’s,
Aristotle’s and Kant’s categories, concepts and doctrines (cf. [Casini 1988])3.
Mostly, it has also been “philosophical” – i.e., it has been highly sophisti-
cated in its relation to earlier contributions to the philosophical tradition
but at the same time commonsensical and often naïve4 in its appeals to

3
An introduction to epistemology based on the historical discussion is Losee [1972].
One critical observation should be made in connection with this otherwise
recommendable work: Hegel, and all approaches somehow derived from Hegel
(including all Marxist views), are absent from Losee’s universe.
4
Here as in the following, I use the term “naïve” in opposition to a generalization
of Kant’s notion of a “critical” approach, and not broadly as “gullible”. The “naïve”
attitude is the one which accepts things for what they seem; the “critical” approach
is the one which investigates whether, why and in which sense the naïve attitude
is justified – “examination of its possibility and limits”, in Kant’s words (Kritik der
Urteilskraft, B III).
Evidently, most of our practical life has to build on “naïve” foundations and
does so without problems. As formulated by Ogden Nash (quoted from Thorkild
Jacobsen [1988: 123]): “O, Things are frequently what they seem/ And this is
Wisdom’s crown:/ Only the game fish swim upstream./ But the sensible fish swim
down.” Often, critique does not tell us that we were wrong when being naïve but

4
empirically established knowledge about the processes of knowing. Aristotle
himself, it is true, based his doctrines upon profound understanding of
the best standards of scientific knowledge of his times; even philosophers
like Locke and Hume, however, were far from understanding the real
intricacies in Galileo’s and Newton’s scientific methods and standards,
notwithstanding their claim and belief that they expounded the true sense
of the feats of these heroes.
Practical scientists, on their part, however sophisticated the methods
they apply in the acquisition of knowledge about the field they investigate,
are often highly naïve when it comes to understanding the philosophical
implications of these “critically”. As formulated by Imre Lakatos [1974a:
148 n.1], “scientists tend to understand little more about science than fish
about hydrodynamics”.
For this double reason, a direct and immediate dialogue between
classical epistemology (or, more specifically, classical “philosophy of
science”) and actual scientific practice is not likely to be very fruitful. An
introduction to the problems of knowledge which intends to further critical
reflection among “practitioners of knowing” should therefore rather take
its starting point in approaches developed during the twentieth century,
based on thorough empirical observation of the processes of knowing and
requiring that their philosophical framework should be able to grasp these
empirical observations5.

only that we were right or as right as could be without knowing why; this is, e.g.,
the outcome of Kant’s “critique of pure reason”.
No critique is ever definitive. What seemed at one moment to be an absolute
underpinning (be it Euclid’s proofs that the methods of practical geometers were
right, Kant’s critique of Newtonian physics or Marx’s critique of Smith-Ricardian
political economy) turns out with historical hindsight to make other “naïve”
presuppositions which in their turn can be “criticized”. We choose a misleading
metaphor when we speak about establishing “a firm foundation” on which we can
build safely. We always build on swampy ground; what criticism does is to hammer
the piles on which we build through so many layers of mud and clay that they
are not likely to move significantly.
5
This preliminary description, “naïve” as anything could be, should not be read
as a claim that these approaches are nothing but empirical. One of the main points
of the following (derived, in fact, from the empirically oriented studies, and agreeing
in this respect with Kant) will be that no knowledge is based on empirical

5
 II. A PIAGETIAN INTRODUCTION TO THE
GENERAL PROBLEM OF KNOWLEDGE

One approach to the problem of knowledge is through individual

cognition. This is even, one might reasonably claim, the most obvious
approach, since knowledge is always knowledge known by somebody,
however socially conditioned and organized it may be. Even knowledge
embodied in books or databases has been put down by somebody and,
more decisively, is only transformed from black dots or magnetic traces
into knowledge when interpreted by a mind.
A number of twentieth century psychological schools have set forth
doctrines or theories about the nature and construction of human cognition.
Some, like the claims of behaviorist school6, are in the main postulates
about how cognition should be explained. Of interest in the present
connection are in particular the Soviet Vygotsky-Luria-Leontieff-school,
and the Swiss Piaget (1896-1980) with collaborators, on whose work I shall
base the present chapter (while being to some extent inspired by the former

observation and experience alone – a fortiori not knowledge about knowing, which
certainly remains philosophy, asking so to speak about the conditions for its own
existence from within the horizon it tries to describe. The “empirically oriented”
approaches are (or become when read as philosophy), to speak with Garroni [1992:
262), approaches which “refuse to be in a place where in fact they are not [viz
outside the world of experience] but which claim, if wanted and possible, to move
through the whole region of common experience and then to land in other, rarely
visited places, yet, at least supposedly, not utopian and still, somehow, belonging
to an experience”.
6
According to which, briefly spoken, all human knowledge is (like behaviour in
general) to be explained as a web of conditioned reflexes – cf. [Høyrup 1993: 189ff].

6
school in my own reading of Piaget).
Piaget is most widely known as a “child psychologist”, which is true
in the sense that he contributed decisively to the understanding of child
development (and has been amply used/misused in the planning of
curriculum reforms) but is otherwise a misleading statement. He started
out as a biologist specializing in mollusks but with a strong interest in
metatheoretical questions, not least in the logic and philosophy of science.
This led him to accept in c. 1920 the task to standardize Cyril Burt’s IQ
test for French children, which again led him to discover his own ap-
proach7.
The principle of the IQ test can be described as kinematics, “movement-
description”. A number of problems are presented to the experimental
subject, and correct and wrong answers are taken note of. At the same time
it is known (from the “standardization”) which tasks are solved by average
children at a given age. In this way, the “mental age” of the subject can
be determined, or the “intelligence quotient” understood as “percentage
problem-solving capacity” in comparison with the average subject of the
same age. The central concept is thus the dichotomy “correct”/“wrong”
answer, and the central tool the time-table telling at which age various
types of problems will normally be solved correctly. Questions concerning
the driving forces behind the process of intellectual development, and even
the very idea of a process, do not occur.
Piaget soon noticed that the “wrong” answers were not only “wrong”
but also systematic. We may illustrate this by an example borrowed from
his later research. A girl of five8, asked whether there are more girls or
more children in her kindergarten may answer “more girls”, “more boys”,
“equally many” [viz, girls and boys] or “I do not know”. The one answer

7
See the introduction to Rotman [1977].
8
A “normal” girl, that is; here and everywhere in this chapter, age indications are
only approximate and subject to large variations. But if your own five-year old
niece gives the “grown-up” answer you can be sure that she would have given
the “five-year” answer at ages three or four. In the language of technical statistics,
ages constitute an “ordinal scale”, and are only numerically true as averages (and
even this only for that geographical and social environment where they were
established – whence the need for re-standardization of British IQ-tests in France).

7
you never get is the “obviously correct” answer “more children”.
I shall return to this experiment below. For the moment we shall only
observe that the consistent deviation from adult thought must correspond
to a different way to conceptualize and think about quantity, not to mere
absence of conceptualization and thought. This was also one of the
conclusions drawn by Piaget, who set out to find the dynamics, the active
forces and processes, of the development of cognition. The other conclusion
was that things like the Kantian categories (object, number, space, time,
causality) cannot be inborn, and thus that cognition has a genesis and
results from a development process.
The titles of some of the books which Piaget published over the
following two decades read like as many empirical tests of Kant’s
categories: Judgement and Reasoning in the Child – The Child’s Conception of
Physical Causality – The Moral Judgment of the Child – The Child’s Conception
of Number. Part of this research was based on the “typical answers” of many
children to “revelatory questions”, e.g., of the type current in IQ testing.
As his own children were born, Piaget took advantage of the possibility
to observe and interact with the same children over several years, thus
interpreting their mind “from within” in a more hermeneutic manner.
During the following decades, Piaget reemerged from his submersion
in empirical child psychology and formulated his general “genetic
epistemology”. i.e., a general theory of human cognition as the outcome
of a process. It is this mature theory (which, it should be noted, remained
a living research programme and was never a fully finished doctrine9)
on which I draw in the following, and which is the starting point for certain
further reflections10.

9
Although, so rumours tell, Piaget was so dominating a personality that the group
of researchers at his Institute in Geneva never developed into a genuine “school”
but remained a circle of “Piaget’s collaborators”. Theoretical innovation seems to
have been the privilege of the master.
10
It should be pointed out that some of the illustrative examples and experiments
referred to in the following are borrowed from Piaget; others are my own
observations, mainly made on my own daughters.

8
Schemes and dialectic
Let us start from an example. A child of (say) 1½ year is familiar with
a variety of balls: White, blue, red, variegated; made from cloth, from
rubber, from leather; with diameters ranging from 3 to 6 cm. It knows that
you may push them, make them roll, throw them from the chair where
you are sitting and thus make your patient big sister pick them up for you.
If a new ball of some colour and of familiar weight and diameter gets into
its hands, the child will demonstrate in practice that it knows how to
behave with balls. We may say that the child possesses a practical concept
or (in Piaget’s language) a scheme for balls. This scheme is not present in
conscious thought but is sensori-motor, i.e., derives adequate movements/
actions from sensual perception.
If now some fully unfamiliar ball is presented to the child – say, a foot-
ball or, even better, a 5 kg leaden ball, certain familiar acts are impossible.
It is too heavy to be thrown (and if you roll in from the table onto your
sisters feet her patience will certainly be gone). But accident or deliberate
experimentation may produce the experience that even this object can be
pushed and rolled along the floor. In this way, the leaden ball is assimilated
to the scheme. The scheme, on the other hand, is changed and made more
flexible by encompassing even this unfamiliar ball: It accommodates to the
larger field of experience.
This illustration introduces a number of key concepts from the Piagetian
theory. Firstly, the pair assimilation/accommodation. Secondly, the concept
of equilibrium, which according to Piaget is the central aim of this twin
mechanism. Equilibrium, in this terminology, is no static condition, but to
be conceived in the likeness (and according to Piaget indeed as a special
case) of the dynamic equilibrium of living organisms. Irrespective of the
surrounding temperature and the precise nature of its food, a dog will
conserve approximately the same body temperature and the same organic
structure; to keep up this equilibrium is the task of its metabolic processes.
Extreme conditions which destroy the equilibrium will, at the same account,
kill the dog. A new-born puppy, on the other hand, can stand fewer
variations of its living conditions. In this sense, the equilibrium of the
mature dog is more stable than that of the puppy. And in the same sense,
the accommodated scheme for balls constitutes a more stable equilibrium

9
than its predecessor, since it is able to grasp without difficulty a wider
range of different balls.
Thirdly, the inescapable duality between assimilation and accommoda-
tion makes Piaget’s theory of knowledge a dialectical theory. It makes sense
of Plato’s claim that you cannot come to know what you do not know
already, and gives substance to Aristotle’s statement that you may, in
different interpretations of the word, know and not know something at
a time: you will only discover that the leaden ball is a ball if you already
know (in the case of the infant, “know” practically) what balls are. But
(and this is the key point where Piaget’s dialectic really goes beyond the
classics): by discovering the leaden ball as a ball you also come to know
more about what balls are. In this way, every act of knowing is at one and
the same time an assimilative interpretation in terms of an existing cognitive
structure and an accommodation of this structure to new domains of reality;
although they are analytically distinct, none of the two processes can take
place in the absence of the other.
Two further observations can be made on the above example. Firstly:
if nothing of what the child usually does when playing with balls fits the
unfamiliar object, it will not be assimilated to the ball scheme; nor will it
of course be assimilated to any other scheme if it does not match in some
way. In most cases, this will result in a practical rejection of the experience
in question – what makes no “sense” is not noticed or quickly forgotten11.
Secondly: assimilation presupposes attentiveness; it is therefore more likely
to result from deliberate experimentation than from accidental events
(behaviourist pedagogical theory notwithstanding, cf. [Høyrup 1993: 191]).
Let us then look at another example. During the first months after birth,
the baby’s world can be described as
an object-less universe, composed of perceptual tableaux which appear and
disappear by a process of resorption, and an object is not sought when it is

11
This is no absolute rule. Human beings, not least children, are curious, and a
puzzling object or class of objects may provoke intense experimentation/investiga-
tion (inquisitive “play” in the case of the child), and in this way provide the basis
for the construction of a new scheme; but if not even the single sensual outcomes
of experiments “make sense” with regard to the existing cognitive organization
of sensory experience, this constructive process is not likely to take place.

10
 hidden by a screen – the baby, for example, withdraws his hand if he is about
to grasp the object and the latter is covered by a handkerchief. At a later stage
the child begins to look for the object, lifting the handkerchief at A where it
has just been covered; and when the object is displaced to a position beneath
B (for example to the right, whereas A was on his left), the child, although he
has seen the object being placed at B, often looks for it at A when it disappears
again; that is, he looks in the place where his action has been successful on
an earlier occasion. He therefore does not take account of successive displace-
ments of the object which he has nevertheless observed and followed atten-
tively. It is not until towards the end of the first year that he looks for the object
unhesitatingly in the place where it has vanished.
([Piaget 1972: 11f]; cf. [1950: 108ff]).
At this age the child will thus remove a blanket under which a coveted
toy has been placed. Still, if a Basque beret and not the toy turns up below
the handkerchief, it abandons the pursuit. Only in a following phase will
it remove the beret and – triumphantly perhaps – find the toy, whose
permanency as an object firmly located in space is no longer subject to
doubt [Piaget 1973: 11ff].
Balls are no necessary ingredients of the universe, and the construction
and stepwise accommodation of a scheme for balls can thus not wonder.
But permanent objects seem to us to be unavoidable, one of the very
fundaments for knowledge of the world. Piaget’s investigations show,
however, that even this as well as all the other Kantian categories without
which no knowledge of the physical world is supposed to be possible is
the product of a development, going through a sequence of accommodative
extensions and equilibria.
In one of his publications, Piaget [1973: 66] defines the scheme of an
action as “the general structure of this action, conserving itself during [...]
repetitions, consolidating itself by exercise, and applying itself to situations
which vary because of modifications of milieu”. On p. 114 of the same
work, a scheme is defined more briefly but in the same vein as “what is
generalizable in a given action”. Both variants of the definition may call
forth the legitimate question, whether the scheme is really part of the mind
of the knowing and acting child or only a construct, made by the observing

11
psychologist for his convenience12. If the scheme is nothing but a psycholo-
gist’s construct, the whole idea of accommodation becomes dubious, and
the question of the status of the schemes must therefore be addressed
before we go on.
Normally questions of this kind are undecidable and hence, one may
claim, meaningless pseudo-problems. In the case of schemes, however, a
decision can be reached – not, it is true, in the precise case of the scheme
for balls, but if we look at grammatical schemes.
The past tense of “hunt” is “hunted”, the past tense of “reach” is
“reached”, etc. These forms, too, are formed according to a scheme, which
can even be seen to apply “itself to situations which vary because of
modifications of milieu” (the “e” is pronounced when coming between
“t” and “d” but not between “ch” and “d”; even in this case, moreover,
we may ask whether the scheme is a grammarian’s shorthand or really
present in the mind of speakers before they have been taught grammar.
The answer is “really present”, and is provided by young children
below the age where they can understand any grammatical explanation.
They may have learned the forms “hunted” and “reached” by hearing them
spoken by grown-ups, and might in principle just store these forms
individually. But they have never heard the form “goed”, since all mature
speakers say “went”. None the less, their first past tense of “go” is “goed” –
and when they discover that this does not agree with adult usage, they
opt for a while for the compromise “wented”13. These forms can only
result from a sub-conscious general scheme which the child has constructed
on the basis of forms like “hunted” and “searched”. This scheme, which
deviates (in extension) from the grammarians’ scheme, must be present
in the child’s mind; it can be no mere observing psychologist’s construct.
If the existence of schemes is established, on the other hand, their accommo-

12
This question is indeed the normal positivist reaction (see below) to any idea of
general structure or universal encompassing and determining particulars – is it THE
HORSE in general which determines the characteristics of individual horses, or
is the universal concept (the species) nothing but a shorthand in which zoologists
sum up their knowledge of similar individuals?
13
“Gåede” and “gikkede”, respectively, in my own observations. But English
linguists confirm the English forms given here.

12
dation (in the case of verbal conjugation in the steps “goed” —> “wented”
—> “went”) is also real.

The periods
“Every act of knowing is at the same time an assimilative interpretation
in terms of an existing cognitive structure and an accommodation of this
structure to new domains of reality”, as stated above. But as a scheme tends
toward equilibrium, the accommodative aspect of the process becomes less
prominent. The different schemes available at a given moment also tend
to form a coherent system and to share a number of basic features and
limitations (to be exemplified below). This is precisely what justifies the
idea of an over-all cognitive structure14. Most of the time, the gradual
accommodation of schemes brings about a maturation of this structure,
increasing its coherence, “stability” and functionality. At certain moments,
however, new mental abilities (one of them being the emergence of
language and conscious thought) destabilize the structure: within a
relatively brief period, new schemes arise, old schemes accommodate
thoroughly, and schemes are integrated in new ways. This lays the
foundation for a new cognitive structure, organized at a higher level, which
is going to assimilate increasingly large ranges of earlier as well as new
experience.
These changes of the over-all cognitive structure demarcate the
developmental periods, of which Piaget counts four (less thorough transforma-
tions of the structure – e.g., the one allowing the child to remove the beret
to find the toy – delimit stages within the period):
1. The sensori-motor period, extending from birth until the age of 1½
to 2 years (with the usual caveat concerning these precise ages).
2. The pre-operatory period, extending from 1½/2 until c. 7 years.
3. The period of concrete operations, from c. 7 to c. 12 years.

14
Piaget, who loved mathematical metaphors and believed them to be more than
metaphors, would often bolster up this explanation by referring to a number of
mathematical and (apparently self-invented) pseudo-mathematical concepts in this
connection. This need not bother us here.

13
 4. The period of formal thought, from c. 12 years onwards15
During the sensorimotor period, the development of intelligence is
characterized by increasingly effective coordination between sensory
perception and motor activity (practical action). Starting off from a number
of separate and unconnected “sensory spaces” (a sucking space, a visual
space, a tactile space, an auditory space), a unitary picture of the surround-
ing world is gradually achieved, where a noise will provoke the child to
look around for the source, and an interesting object coming within the
visual space may make the child move towards it and grasp it16. The
categories of space, time, permanent object, and cause are developed as practical
(but certainly not conscious) categories. What this means in the case of the
permanent object was already elucidated above. Having a practical category
of space means possessing (among other things) the ability to plan a
composite trajectory through a room, picking up a toy at point A and going
on directly (and without new spatial planning) to point B where the toy
is to be used for some purpose. Possessing a practical category of causation
implies, e.g., to get the idea to draw a table cloth upon which a coveted
object is placed outside your range.
Implicit in the formation of the unified sensory space and of the
practical categories is a gradual de-centralization (or “miniature Copernican
revolution”, as Piaget calls it [1967: 79]). The child itself is the only possible
centre of the sucking space, of the tactile space, etc. – these spaces are

15
I remind of the observation made above that the age indications correspond to
averages as established in the (urban European) environment where Piaget
established the sequence. Others have made similar investigations in others
surroundings, finding (with a restriction concerning formal thought in certain
societies, to which I shall return) the same relative sequence but a considerable
time-lag, e.g., in an Iranian rural district where children were “not only without
schools but also without any toys other than pebbles and bits of wood” (Piaget
[1973: 154], citing Mohseni’s research); it is important to note that this delay only
concerned the onset of the operatory stage, whereas tests based on preoperatory
thought showed no difference between these children and children from Teheran.
16
These acts of intelligence should be held strictly apart from certain reflexes which
seem to mimic them. The grasping reflex, e.g., makes the new-born infant grasp
whatever comes within its hand and cling to it; intelligent grasping is an intentional
act which the child may choose or not choose to perform.

14
“egocentric”17. In the integrated space, on the other hand, the child only
occupies one position among many possible and equally valid positions
(which means that it can plan a trajectory passing through a sequence of
positions).
The sensorimotor cognitive structure reaches maturity around the
middle of the second year of life, when integrated space, practical categories
and decentralization are attained. At this moment the child behaves as
adequately as it is possible without the intervention of conscious thought,
given its physical and sensory equipment. At that moment it has reached
the maximum intelligence of chimpanzees. It also stands at the threshold
to a new period, characterized by the first emergence of language and
conscious thought18.
One element of the new cognitive structure is created through the
maturation of sensorimotor thought. Being able to plan a trajectory or
stopping at a problem, reflecting and suddenly knowing how to go on after
an “aha-Erlebnis”19 are indications that the schemes for action have been
“interiorized”, have become an element of thought, which can be antici-
pated before (or without) becoming actual action (cf. [Piaget 1973: 57]).
Another element, and a symptom that thought is emerging, is symbolic

17
If the concept of an egocentric space seems queer one may think of those sensory
spaces which are never really integrated into the unified space: The gustatory
“space”, the olfactory “space”, the pain “space”, the heat and cold “spaces”. All
of these are immediately perceived where you are yourself, and they are often felt
as moods rather than as information about the world around you. Sounds, in
contrast, are not perceived as heard in the ears or anywhere else in the body but
at their presumed source (barring exceptional cases), and light as a rule not at all
as light but as objects seen at a distance.
This does not (or not alone) depend on the nature of the stimulus: dogs, so
it appears, integrate their olfactory space into the unified sensorial space, and
perceive smells as connected to objects, not as mere smells.
18
Chimpanzees seem indeed to stand at the same threshold – so precisely there,
in fact, that the same experiments can be interpreted by some scholars as proof
that they possess the ability for language and by others as proof that they lack this
ability.
19
This term was first coined by Wolfgang Köhler (also known as one of the founding
fathers of gestalt psychology) as a description of the same moment in chimpanzee
behaviour.

15
play. Sensorimotor children “play with” (e.g., with a ball, or with their toes),
whereas older children “play that” (e.g., that they are parents while the
doll is the child). Such symbolic play presupposes the symbolic function,
but it also provides this function with substance through internalization
of the play.
The most conspicuous element, of course, is language, which has been
prepared through extensive sensorimotor play with speech sounds
(“babbling”) but only becomes language and a tool for thought in the
moment where a string of speech sounds functioning as phonemes is used
symbolically for something beyond itself and is interiorized with this
function20.
The (embryonic) emergence of a new cognitive structure does not
abolish the achievements of the preceding period. Children of three are
no less sensorimotor effective than they were a year before. Still, the
immediately interesting point in a discussion of the cognition of children
between two and seven is of course the new level, the characteristics of
thinking.
The thought of “pre-operatory” children (a term to be explained in a
moment) is characterized by Piaget as follows:
— it is egocentric;
— it is centring;
— and it is irreversible.
That thought is egocentric (not the same as egoistic!) means that the
child does not perceive of its thoughts as its own thoughts which others may
not share; in other words, it comprehends its point of view as the only
possible point of view, which makes it difficult to entertain a genuine

20
This may be made more explicit by an example. At one moment in life, you may
use a sound like “mama” in order to make your mother take you in her arms; this
can be interiorized in the same way as any other sensorimotor scheme, and will
allow you to plan how to be taken into your mother’s arms (an instance of practical
causation, similar to the use of the table cloth); but only when the sound is used
as a more general symbol for this all-important permanent object (normally the
first object of whose permanence we are sure) will it allow you to think about your
mother and not only about the specific act.
The linguistic distinction between sounds and phonemes (which is not very
important in the present context) is explained briefly in note 39.

16
dialogue with the child unless you already know what it thinks. This may
be illustrated by means of a conversation with a girl of two who had been
to a zoo with her creche:
Adult: “What did you see in the zoo?”
Girl (with enthusiastic emphasis): “That one!”
Adult: “What is ‘that one’?”
Girl (with increased emphasis): “That one!”
Adult: “Did you see monkeys?”
Girl (happily): “Yes”.
The answers given by young children when confronted with questions
of “why” are characteristic of a generalized egocentric attitude. They are
not answered as concerned with causes but as regarding an anthropo-
morphic aim: “If Lake Geneva does not go as far as Berne, it’s because each
town has its own lake”21.
It has been objected repeatedly to the cognitive interpretation of such
statements that children answer in such unexpected ways because they
understand words differently. This is certainly correct in itself. Children
understand words in a way which corresponds to their answers. But they
do so consistently (and in spite of the way the words are used by their
adult surroundings from which they learned them) because they are unable
to grasp the usage of the grown-ups, which cannot be assimilated to their
own cognitive structure. The “cognitive interpretation”, the interpretation
of children’s sayings as evidence of their basic pattern of thought, is not
only permissible but mandatory.
The “centring” character of thought can be explained through another
example (in fact one of Piaget’s key experiments). A child of five is shown
a ball of soft clay and looks at the experimenter as he rolls it into a sausage.
Asked whether the sausage contains more or less clay than the ball, three
out of four will respond that there is more, since “the snake is longer”;

21
[Piaget 1967: 25]. Twentieth-century adult thought, of course, may still answer
some questions of “why” as concerned with purpose. The motor-road around
Roskilde is there “because cars should be led around the centre of the town” and
not “because workers levelled the surface and spread concrete”. But the Alps are
there because of geological forces and not because they constitute such a nice skiing
resort. At least in the era of secularization, adult thought tends to de-anthropomor-
phize processes not performed by human beings.

17
the rest will claim that there is less, since “it is thinner”. In both cases, the
child concentrates interest on one conspicuous feature of the situation and
does not take other features into account.
The irreversible character of thought can be elucidated by one of my
own observations. A girl who had learned rudimentary counting by means
of the number jingle (and who was thus at the threshold to the next period,
cf. below) was asked as follows: “Seven birds are sitting in a tree. Then
two more birds join them, but soon two birds fly away. How many are
left?”. The answer was simple and characteristic: “You would have to count
them again”. Some months later, in contrast, she was asked the same
question, but dealing with “a hundred thousand million birds” (i.e., a
number which she could not remember but knew was a number), eight
birds joining and eight leaving. She was not encumbered by the impossibly
large number but simply answered “equally many”. At this moment she
could grasp the process “from above”, and administer the mental process
from a meta-level from where the mutual cancellation of the two changes
(in other words, the reversibility of the process) could be perceived.
The absence of a meta-level in the thought of pre-operatory children
is seen in another one of Piaget’s question types – the one exemplified
above by the question whether there were “more girls or more children
in the kindergarten”. The point in the answers given by pre-operatory
children is precisely the absence of a meta-level on which the total category
of children can be comprehended together with the two distinct subcate-
gories. In the absence of this level, separating the girls automatically
transforms the idea of “children” into “the remaining children”, i.e., “boys”.
The mental process is really an ongoing process, a chain where each step
supersedes its predecessor step and makes it inaccessible to renewed
treatment (cf. [Piaget 1950: 113])22.

22
It is sometimes claimed that the young children produce their “absurd” answer
not because of any inability to understand “correctly” but simply because they
assume the psychologist to have something sensible in mind. That this is no
adequate explanation follows from an observation which I made on my other
daughter when, at the age of five, she was close to the end of the pre-operatory
period (having already learned rudimentary counting). When asked whether there
were more girls or more children in her kindergarten she replied “I don’t know”.

18
 The radical egocentricity exemplified by the use of “that one” as a valid
terms for monkeys retreats during the pre-operatory phase, and from the
age of four the child is able to participate in what the adult interlocutor
perceives as sensible dialogue. This demonstrates that mature pre-operatory
thought is a functioning and relatively adequate structure – relatively
adequate functioning is indeed the very condition that a cognitive structure
can be stable and thus mature. It may therefore seem misleading to
characterize this thinking by what it is not (yet), i.e., not based on operations
(to be explained presently). But in spite of relatively adequate functioning
of the thought of the period, its irreversibility and centring character as
well as a less radical egocentricity remain for years, and a meta-level does
not develop. As an alternative to the negative characterization “pre-
operatory”, Piaget therefore uses the term “intuitive thought” ([1950: 129];
[1967: 29]) to describe the typical way of thinking of children between four
and six as what it is: Stating opinions without argument or support from
facts; how should, in fact, a chain of arguments be constructible if you
cannot step outside the chain of your own thinking? And what would be
the use of arguments if you do not see your own thinking as only one
possible way to think which your interlocutor does not necessarily share?
The negative characterization “pre-operatory” is a characterization of
that which typical pre-school thinking is not yet but is on its way to prepare.

I then produced a model, telling that her fingers were the children, those on the
left hand the girls and those on the right hand the boys; when asked, she was able
to identify both girls, boys, and all children correctly and without difficulty.
Next I asked for the number of girls, and she counted until five on her left hand.
When asked about “all the children” she counted until five on her right hand. I
took her hand and continued with “six” on her left – and before I reached “seven”
she burst into violent crying, screaming “one can’t do that” – without my knowing
it, my seemingly innocuous experiment had violated her world order.
As one may imagine I did everything I could in order to comfort her, admitting
that it could not be done according to the way she thought, but that it could in
the way her big sister and I thought, which I then tried to explain. In the end, I
asked whether she understood, which she denied – but at least she had stopped
crying. Two hours later, however, she could understand nothing but the operatory
structure, and for days she went on enthusiastically posing me analogous questions
about mackerel and fish, and other categories as bizarre as possible.
Neither crying nor enthusiasm could certainly have been produced if nothing
but a linguistic misunderstanding had been involved.

19
Around the age of seven, children will know that there are more children
than girls in a mixed school class, and more flowers than primroses in the
garden. Similarly, they will now tell that the sausage and the clay ball
contain the same quantity of clay. In the latter case, the reason normally
given is typical of the acquisition of reversibility: You could roll the sausage
back into its old shape. The idea of “same” is, in fact, quite empty or at
least unspecific; it will take years before the child is able to foretell that
the weight of the sausage is not changed, and still more before it will predict
that the volume (i.e., the raise of the level of water in a glass into which
the clay is immersed) is conserved. Similarly, the child who knows that
flowers outnumber primroses is his garden (and who tells that not all
animals are birds, since “there are also snails, horses ...”, whereas all birds
“certainly” are animals) will not be able to decide whether there are more
birds or more animals outside the window23.
In Piaget’s words [1973: 24], the reason for the latter failure is probably
that “flowers can be gathered in bouquets. This is an easy concrete
operation, whereas to go and make a bouquet of swallows becomes more
complicated; it is not manipulable”. Initially, the intellectual operations
of the next cognitive structure are thus strictly bound up with concrete
imagination. But they are what Piaget calls “operations”, i.e.
“interiorized (or interiorizable), reversible actions (in the sense of being capable
of developing in both directions and, consequently, of including the possibility
of a reverse action which cancels the result of the first), and coordinated in
structures [...] which present laws of composition characterizing the structure
in its totality as a system”.
([Piaget 1973: 76]; cf. [1967: 78])
This concept is beautifully illustrated by the “hundred thousand million
birds” plus and minus eight24. Only at this level can a concept of counting

23
This delay of the unfamiliar as compared to the familiar is obviously contradicted
by the observation reported in note 22. The difference is to be explained by the
blow to which I had unknowingly exposed my daughter’s world order, and
illustrates the importance of affective factors even in the domain of seemingly
neutral cognition.
24
With the qualification that the girl had been subjected to so many Piagetian
problems that she was able to manipulate fairly abstract entities like the fancy
number in question.

20
be constructed (younger children see no problem in having mislearned
the number jingle in such a way that it ends in a loop, e.g., 1-2-3-4-5-6-7-8-
9-10-11-7-8-9-10-11-7-8-...); only at this level is conceptual analysis possible;
etc. In general, operatory thought is felt by adults to be “logical”; actually,
Piaget used the term “pre-logical” as late as 1940 instead of the technical
term “pre-operatory” which he was to coin later (see [Piaget 1967: 30]).
Operatory thought starts in the most concrete domain, as we have seen.
Gradually, larger areas are assimilated – at the typical age of ten, no doubt
will remain that there are more animals than birds outside the window,
nor that the weight of the clay ball does not change when it is transformed
into a sausage; at the age of thirteen, most children will admit that even
the volume cannot have changed. Still, the logic of operatory thought
remains a logic of the concrete. Instead of seeing assimilation as an extension
of operatory structures so as to cover increasingly abstract domains we
should indeed rather see it as an integration of increasingly wider ranges
of experience into the realm of the familiar and thus concretely imagin-
able25. Purely formal operations remain inaccessible. This is illustrated
by the inability of typical children below twelve to answer correctly one
of the old Burt problems, an example to which Piaget returns time and
again (e.g., [1950: 149]): “Edith is fairer than Susan; Edith is darker than
Lily; who is the darkest of the three?”. The child of ten will mostly argue
that both Edith and Susan are fair, while both Edith and Lily are dark.
Edith must thus be in between, Lily must be darkest, and Susan is fairest.
Only at (typically) twelve, thus Piaget, will the child be able to argue
correctly from the purely formal sequence, e.g., by inverting the first
statement into “Susan is darker than Edith”. From now on, the child (now,
normally, an adolescent) will be able to manipulate symbols in agreement
with abstract rules and without consideration of their actual meaning.
The ability to handle such purely formal problems should then
demarcate the emergence of the fourth (and definitive) cognitive structure.

25
This follows clearly from my experience as a teacher of mathematics and physics
to grown-up students of engineering or mathematics. Even many of these would
make typical pre-operative (centring and irreversible) errors as soon as problems
dealt with entities with which they had no familiarity, and which they could not
imagine concretely (e.g., sound energy density).

21
My own investigations and observations of mathematical (i.e., supposedly
purely formal) reasoning as well as the research of the last 30 years
regarding the importance of writing (and, by implication, material
representation) for the mastery of abstract thought26 make me doubt the
validity of the absolute distinction between concrete and formal thinking.
The typical mathematician will immediately transform Burt’s problem into
(actual or imagined) writing: “E>S, E<L”, turn the first statement around
into “S<E”, and combine the two into “S<E<L” (according to my repeated
experience, most of them resort to pen and paper). This is no longer a
sequence of formal operations but a progression of quite manifest mani-
pulations of visually familiar and thus “concrete” entities, facilitated by
an iconic symbol27 – concreteness being, in fact, no immanent characteristic
of the object but a characterization of the attitude of the knowing person
to the object (in themselves, birds are no less concrete than primroses).
Regardless of the absolute or only relative character of the transition
from concrete to supposedly formal thought, the acquisition of the ability
to handle mentally the endless range of problems outside the realm of
direct concrete experience (be it by means of formal thought or through
the construction of pseudo-concrete representations) is of course an
important step; without this step, in particular, scientific knowledge could
never be achieved.

26
Summarized in [Ong 1982]. If, as it seems, supposedly formal thought can only
develop on the basis of writing (or equivalent material representation), there is
nothing strange in the apparent absence of formal thought from cultures without
writing. Writing cultures tend to consider thought based on interiorized writing
as “formal” but to see internalization of other material representations as “concrete”.
The claim that “primitive man is unable to think formally” (which was current in
this or similar forms in the psycho-ethnography of the earlier twentieth century)
is thus nothing but a self-promoting reformulation of the quasi-tautological
observation that “cultures without writing do not base their thought on writing”.
27
Not only is the symbol < smaller in the end pointing to the smaller entity; the
relation between < and > also corresponds to the spatial reversal in the written
line which transforms one relation into the other. Most important perhaps: one
< can be located within the other, as <<, in a way which appeals directly to the
sensorimotor experience of putting a smaller box into a larger one and this into
a still larger one.

22
Supplementary observations
The characterization of the periods of cognitive development is a
skeleton; in order to dress it in flesh and skin one has to make a number
of supplementary observations, of which I shall introduce a few – some
from within the Piagetian perspective and some from the outside.
The first observation to be made extends the remark that “children of
three are no less sensorimotor effective than they were a year before”. Nor
are they at seven, nor at 25. Sensorimotor learning continues, when you
learn to ride a bicycle, when you begin driving a car, etc. Certain skills,
furthermore, are learned at the level of conscious thought and then
absorbed as subconscious sensorimotor skills (e.g., changing gears in a car,
or binding the bow knot of your shoes). To some extent, and increasingly,
the basic or primitive cognitive structures are integrated with the higher
(conscious) structures and made subservient to these (as we shall see in
a moment, this integration is far from complete and no unproblematic
process).
Cognitive structures were totalities integrating many schemes character-
ized by shared basic features and limitations. Even when broadened so
as to encompass bicycle and car riding, sensorimotor cognition is thus
different, and distinct, from operatory cognition. None the less, schemes
may in some sense be transferred from one level or structure to another –
or, perhaps with a better metaphor, serve as models for the construction
of analogous schemes at new levels. Such a suggestion is made occasionally
by Piaget (e.g., [1967: 25f, 48]). It is indeed a characteristic of operatory
thought that it achieves at the level of thought the same decentration,
reversibility and composability which was achieved at the sensorimotor
level during the second year of life. The process could be formulated as
an assimilation of verbal and nonverbal thought to sensorimotor
schemes28. Correspondingly, the characteristic schemes of formal thought

28
The schemes of which Piaget speaks in this connection can be characterized in
a broad sense as “logical schemes”: the fact that there are more flowers than
primroses does not depend on the specific nature of the two categories but only
on the fact that the first category encompasses the second without being itself
exhausted. It might be worthwhile investigating (via the “spontaneous grammar”
of young children, see e.g. [Bickerton 1983]) whether the development of grammati-

23
could be seen as an assimilation of sentences to schemes used to deal with
concrete objects (in as far as it is at all justified to distinguish these two
levels, cf. above).
A different type of observation concerns the status of Piaget’s favourite
discriminative experiment: conservation (the clay sausage and its kin). The
understanding of conservation is more or less taken by Piaget to be the
essential content of operatory thought. Cross-cultural studies makes this
doubtful. Australian aboriginal children are indeed unable to master
conservation at an age where European children do so; use of maps,
however, which also requires operatory thought and only comes later to
European children, comes within their range at the age when European
children grasp conservation29. It seems as if at least operatory thought
is a rather open-ended potentiality, the precise actualization of which
depends very much on that cultural practice which brings one or the other
kind of mental operation to the fore.
A final cluster of observations concerns the mutual relation between
the different co-existing cognitive structures. Firstly there is the phenome-
non of cognitive regression. It was explained above how operatory structures
only assimilate unfamiliar domains gradually, as they become familiar.
Familiarity, however, is only one factor. Engineering students are more
likely to fall back on intuitive thought at exams than during daily teaching.
This recourse to basic cognitive structures is a fairly common consequence
of emotional stress, and can be observed on many occasions – in heated
discussions, when you try to find by systematic trial-and-error the code
of an unknown bicycle lock and get nervous, etc. In the history of science,
repeated examples of elementary blunders committed by eminent minds
approaching the borders of their understanding can be listed.

cal schemes can also be correlated with the preexisting structure of sensorimotor
schemes.
There are good reasons to expect such a correlation. On one hand, the area
of the cerebral cortex that is main responsible for processing syntax is the very area
that processes spatial information (commentary by Ron Wallace in [Burling 1993:
43f]). On the other, in almost all cases where the origin of grammatical cases systems
can be traced, they derive from frozen spatial metaphors (cf. also [Anderson 1971]).
29
See [Cole and Scribner 1974: 152f].

24
 The choice of cognitive structure to apply in a given situation is indeed
no conscious decision. Nor are, in fact, the schemes themselves present
to conscious awareness. This is of course true in the case of the infant
applying the ball scheme or constructing a grammatical form “wented”;
but it is equally true of the daily thinking of the professional philosopher.
He, of course, may notice that his conclusions follow from a particular
syllogistic scheme, like the classical “All men are mortal; Socrates is a man;
thus Socrates is mortal”. But he needs no scheme to know it, and he knows
it immediately before correlating it with the scheme. The schemes inside
which our conscious thinking takes place are themselves unconscious (cf.
[Piaget 1973: 31-48]).
This holds in general, and normally we do not think about it. At times,
however, the complex interplay between that which we are aware of and
that which does not come to awareness may produce paradoxical errors
and failure to grasp correctly what one is able to do correctly in practice.
As an example of this “cognitive repression” I quote Piaget:
A child is given a sling in its simplest form: a ball attached to a string which
is whirled, then aimed at a goal. At first, there is no goal whatever and the
child enjoys whirling the ball at the end of the string and then letting it go,
noting that it flies off from his side (and in general even seeing that it flies off
in the extension of the rotary direction). Next a box is placed thirty to fifty
centimeters away and the child, often as early as five years old, quickly
manages to reach the box by whirling the ball from his side (about nine o’clock,
if we consider as clock dial the rotation surface, the box itself being placed at
noon). Having done so, the child is complimented; he begins again several times
and is asked where he has released the ball.
A strange reaction then occurs. The youngest children claim that they
released the balls exactly in front of them (about six o’clock) and that the ball
left in a straight line, from six o’clock to noon (the diameter of the rotary circle)
into the box. Others (children aged seven to eight) claim that they released the
ball at noon, that is, facing the box. About the ages of nine to ten, there are
often compromises: The ball is released about eleven or ten-thirty, and it is
only about the age of eleven or twelve that the child replies at once that the
ball left at nine o’clock, that is, tangentially and no longer facing the goal. In
other words, the child soon knows how to accomplish a successful action, but
years are needed before he becomes aware of this, as if some factor were
opposed to this knowledge and retained in the unconscious certain movements
or even certain intentional parts of successful behavior.
The factor behind inhibition is easy to discover. The child represents his

25
 own action as divided into two periods: spinning the ball, then throwing it
into the box; whereas without this goal he throws the moving object anywhere.
But, for him, throwing to the goal supposes a perpendicular trajectory to the
box, thus a release facing it. When asked to describe his action, he thus
reconstructs it logically as a function of this preconceived idea and hence does
not wish to see that actually he proceeded differently. Therefore he distorts
and even dismisses an observation contrary to the idea he has and which alone
seems right to him.
[Piaget 1973: 36ff]
At the sensorimotor level, the child thus knows correctly what to do.
But this correct knowledge is not brought to awareness because it disagrees
with a pre-existing conceptual scheme which reconstructs the process
wrongly; one might add that even the correct awareness of the oldest
children is probably the outcome of a reconstruction – albeit a better one –
and not of immediate observation. The path leading from sensory
perception and even intentional action to recorded observation is far from
direct30.
Most often, indeed, awareness only results from challenge or conflict.
If a wrong reconstruction is used as the basis for the planning of further
action one may be forced to recognize that something is wrong (because
the sling ends up in a wrong place), and then be led to better understand-
ing. When the activity in question is unproblematic and everything
functions as expected, awareness is superfluous and need not arise. We
are normally quite unaware of the precise mechanisms of walking, and

30
This does not hold for children alone. Once Piaget’s collaborator A. Papert had
investigated whether children are able to tell afterwards what they did when
walking on all four. The result was that the youngest provided a physically rather
impossible explanation (movement “in Z”, first the arms are moved, then the legs,
...). Somewhat older children would provide a physically possible explanation which
did not agree with that they had actually done (“in N”, both left limbs, both right
limbs, ...; obviously copied from ordinary walking). Only the oldest children tended
to produce a correct description of what they had actually done (movement “in
X”) [Piaget 1976: 1-11].
Before they presented this result to an interdisciplinary symposium, Piaget and
Papert had the participants walk on all four and then asked them to describe in
writing what they had done. According to Piaget’s account, physicists and
psychologists tended to give the correct description – whereas logicians and
mathematicians gave the physically possible but actually wrong description [Piaget
1973: 41].

26
we have no reasons to produce awareness: walking can be integrated
without that in consciously planned movement, e.g. according to a map.
It is only when the terrain is utterly difficult, if one of our feet is severely
hurt, or in similar situations, that we are forced to focus awareness on the
actions involved in walking.
The moral of this observation is of extreme importance for any theory
of knowledge. Since theories are themselves products of awareness and
addressed to conscious awareness, they tend quite naturally to identify
knowledge with conscious knowledge. But the larger parts of our know-
ledge – firstly the basic schemes, but secondly even many of the actions
performed within these schemes – are unconscious, and as much as possible
remains so as long as unawareness gives rise to no problems. Many
activities, moreover, which are learnt at the conscious level are removed
from consciousness by repetition and training (training is indeed intentional
removal from consciousness) – changing gears and shoe-binding were
mentioned above as examples. This is a question of simple economy.
Conscious awareness can keep track of only a few processes at a time (and
it does so relatively slowly); it therefore has to be reserved for these aspects
of our actions which can not be accomplished automatically; it is also easily
distracted and thus more likely to commit errors than automatically
performed routines.
A theory of scientific knowledge which does not take that into account
can be reproached of neglecting the important fact that even scientific
knowledge is human knowledge. It is inadequate, either by being unable
to understand the economy and ease by which even scientific cognition
works through the automatization of sub-procedures; or by copying the
incomplete awareness of the acting individuals, not recognizing the
pertinence (or the very existence) of those procedures which are automated.
In both cases, it also fails to explain the errors to which automatization
gives rise on certain critical occasions.

27
The status of schemes and categories
These are questions to which we shall return (p. 77 onwards). At first,
however, we shall have to close the discussion of individual cognition.
Kant, we remember, held certain categories to be inescapable frame-
works without which we cannot know; Piaget, on his part, investigated
how these categories arise, and demonstrated that at one moment in life
these categories and the schemes in which they are organized were not
yet part of our cognitive equipment; they were thus not only not in-
escapable but in fact inaccessible. At that time we gathered and organized
our experience in other ways. The Kantian equipment is only a preliminary
outcome and not the starting point of our knowing about the world.
But why do we end up with precisely these categories? Might suitably
planned education have produced a cognitive structure not based (among
other things) on the category of permanent objects, on the expectation that
a toy which has been concealed must still be somewhere, and that a rabbit
drawn out of the magician’s top hat must have been hidden in his sleeves?
Might our world have become one possessing no fixed boundaries between
any this and any that, a world ever fluid and elusive?
This other world is difficult to describe, and for good reasons: it is not
our world and not the world to which our concepts and our language are
adapted. Are then the separation of our world into objects, our logical
schemes and categories simply an implicit message we could not avoid
when learning our language and learning to describe our world in
language?
This is certainly not completely false, as illustrated by a cross-linguistic
example. Chinese does not allow the enunciation of counterfactual
statements like “if printing had not been invented, then the industrial
revolution could not have occurred” but only approximate equivalents
translatable into “Printing has not been invented, and therefore the
industrial revolution has not occurred”. In a test where English-speaking
subjects are asked for a conclusion from the first formulation combined
with the statement “The industrial revolution has occurred”, they will have
little difficulty in deciding that printing has been invented. Chinese,
confronted with their equivalent, are likely to protest that “Printing has
been invented” and thus not to accept the game. Bilingual Taiwanese

28
Chinese, moreover, tend to react “in English” when asked the test in
English and “in Chinese” when asked in Chinese [Bloom 1979]. Without
adequate support in language (or other symbolizations), higher logical
schemes are not easily accessible.
But most of the fundamental categories develop as practical categories
before language is acquired, and others (like the girls-children scheme)
may be present in language but systematically misunderstood until the
moment when our cognitive structure is ready to use the linguistic structure
as mature speakers do. Furthermore, basic categories like the permanent
and separate object are common to all languages, which would hardly be
the case if language itself was not constrained in this domain. As far as
the basic categories are concerned, language appears to be only a secondary
regulatory factor, a support which stabilizes the incipient formation of
individual categories and structures by lending them adequate symbolic
expression.
This leaves us with two possibilities. The categories may be determined
by our perceptual and nervous apparatus; that they develop may then
reflect that this apparatus is not fully evolved when we are born but needs
to go through a process of maturation. Or they may really correspond to the
structure of the world in which we live; their development then reflects
a process of discovery.
None of the two explanations taken alone is satisfactory. If nothing but
maturation is involved, why should children in the Iranian countryside
acquire operatory structures at a later age than urban children from the
same country while being no less intelligent according to tests which do
not require operatory thought (see note 15)? And why should Australian
aboriginals develop operatory structures in an order which differs from
that of Swiss children? On the other hand, whatever the structures in which
the world around us is ordered, we would not be able to adopt them into
our cognitive structure if we were not in possession of an adequate nervous
system.
This can be illustrated by another reference to the category of perma-
nent objects (I rely for this on [Jerison 1973] and [id. 1976]). Frogs and other
more primitive vertebrates possess no scheme for the permanent object;
frogs jump at the direct perception of visual signals (e.g., representing flies).

29
A frog whose eyes have been turned around in a surgical operation will
jump in a wrong direction for the rest of its life. Early mammals, however,
who hunted their prey at night and had to rely on the integration of sound
impressions over time in order to construct the path of the prey, developed
a larger brain and a scheme for permanent objects – more precisely: a larger
brain which allowed them to organize their sensual impressions as
representations of permanent objects.
The larger brain with which mammals (and birds) are born has large
biological costs. Birds and mammals are born immature and defenseless,
and would never survive if their parents left them to themselves as
crocodiles do with their offspring (mammals indeed became mammals as
a way to take care of their litter). If the possibility to experience the outer
world as consisting of permanent objects had not implied definite
advantages which could balance this cost, selection pressure would soon
have eliminated the larger mammalian brain. We cannot conclude that the
world consists of permanent objects. In fact, it does not in any absolute
sense: the fox pursued by the hounds exhales and transpires, and matter
which in one moment is fox is not a moment later; physics tells us that
we may analyze the fox into atoms and into still smaller particles. What
we can conclude is that the material world is constituted in a way that
allows an adequate practice if we order our perceptions as representations
of permanent objects. Only in this sense can we say that the world in which
we live “is” itself structured in permanent objects.
Mutatis mutandis, the same will hold for other fundamental cognitive
categories and schemes. Evidently, the evolutionary level at which they
have evolved will be different, and operatory thought appears only to have
arisen during the process of hominization.

30
 III. THE NATURE AND DEMARCATION
OF SCIENTIFIC KNOWLEDGE

Knowledge is always known by somebody, if not actually then at least

before it was stored in books or other receptacles; and stored knowledge
remains knowledge only in so far as it can reemerge as the knowledge of
somebody. Therefore, all knowledge partakes somehow in the character-
istics of individual knowledge.
But part of what we know is only individual knowledge; part of what
we know is only accidentally shared by groups of people who happen, e.g.,
to be witnesses of the same events; and part of our knowledge is produced
so as to be communicated and shared.
Part of what we know, furthermore, concerns particular occurrences –
e.g., the rainy weather in Copenhagen in this moment; another part is of
a more general character: describing, e.g., the Danish climate or the dynamics
of cloud formation and movement. Some of it, finally, consists of isolated
bits, wheras other parts are built up as wholes (“theories” and the like) whose
single constituents only obtain their full meaning as components of these
wholes.
Scientific knowledge is produced so as to be communicated and shared
in stored form; it is generalizing in character; and it consists of wholes.
It is, furthermore, produced by communities with a strong internal
interaction by means (inter alia) of stored knowledge (books, scientific
journals, letters). For all these reasons, scientific knowledge possesses
particular qualities31.

31
From another point of view, it is true, these characteristics of scientific knowledge
highlight general conditions for knowing, or at least for knowing explicitly: all

31
 Some of these concern the relations between producers, users and (their
existence should not be forgotten) victims32 of the knowledge in question,
and the links between its production and its social and technical uses.
Others (and they are the theme of this and the following chapters) concern
more purely epistemological questions. For instance: Is scientific knowledge
true? If it is, in what sense and under which conditions? If not, why do
we then rely upon it, so often with considerable technical success? Etc.
Since scientific knowledge normally ends up as stored knowledge,

expressible knowledge presupposes a language in which it can be expressed

(whence also shared and stored), and language is, if anything, common property
of a community. John Donne’s maxim “No man is an island” holds for human
knowledge no less than for the rest of our existence.
Any piece of knowledge, moreover, however much it may deal with particular
occurrences, it locates these within a framework of general concepts
32
Since these victims are none the less only indirectly visible in what follows, it
may be suitable to repeat the sonnet in which Adalbert Chamisso – poet, Jacobin,
scientist, and believer in scientific progress – commemorated the hundred oxen
which Pythagoras is told to have sacrificed when discovering “his” theorem:
Die Wahrheit, sie besteht in Ewigkeit,
Wenn erst die blöde Welt ihr Licht erkannt;
Der Lehrsatz nach Pythagoras benannt
Gilt heute, wie er galt zu seiner Zeit.
Ein Opfer hat Pythagoras geweiht
Den Göttern, die den Lichtstrahl ihm gesandt;
Es thaten kund, geschlachtet und verbrannt,
Einhundert Ochsen seine Dankbarkeit.
Die Ochsen seit dem Tage, wenn sie wittern,
Daß eine neue Wahrheit sich enthülle,
Erheben ein unmenschliches Gebrülle;
Pythagoras erfüllt sie mit Entsetzen;
Und machtlos sich dem Licht zu widersetzen
Verschließen sie die Augen und erzittern.
(Werke II, 212)
(Truth endures eternally / once the inane world has perceived its light; / the
theorem, named after Pythagoras / is valid today, as in his times. // A sacrifice
did Pythagoras dedicate / to the Gods, who sent him the illuminating ray; / one
hundred oxen, slaughtered and burnt, announced his gratitude. // The oxen, since
that day, when they suspect / that a new truth may reveal itself / bring forth an
insufferable roaring: // Pythagoras fills them with horror / and without power
to resist the light / they close the eyes and quiver.)

32
traditional philosophy of science has approached these epistemological
questions with regard to the stored form of knowledge. In later decennia,
however, certain workers in the field have insisted (against strong
opposition from traditionalists) that genuine understanding of scientific
knowledge (including understanding of the stored final phase) can only
be achieved if we understand its original emergence in process as individual
knowledge generated within a field of specific social interaction.
Chapters IV and V will take up the latter approach. In the present
chapter I shall introduce some fundamental concepts and terms and look
at some of the established approaches and their problems.

A pseudo-historical introduction to some key concepts

An Ancient anecdote33 reported in [Høyrup 1993: 5] may be repeated
with profit together with some of the commentaries made there. It runs
as follows:
One day Plato the philosopher met his fellow philosopher Diogenes, who, as
so often, made a teasing comment on Plato’s philosophy. “Good Friend”, he
said, “Table and cup I see; but your Tablehood and Cuphood, Plato, I can
nowhere see.” “That’s readily accounted for, Dear Diogenes”, replied the other.
“You have that which is needed to see the table and the cup: that’s the eyes.
But you lack what is required to grasp Tablehood and Cuphood: namely the
intellect”.
This story locates the two philosophers with regard to several concepts
and problems which are central to established philosophy of science:
Plato is an idealist. That is, to him things exist primarily as knowledge
or thought, in a mind, as concepts or ideas (in Greek: that which is seen),

33
Told by Diogenes Laërtios in his Lives of Eminent Philosophers VI, Chapter 53;
slightly paraphrased here from [Hicks 1980: II,55].
The characterization of the present section as “pseudo-historical” should be
emphasized. It does not present the points of view of Ancient philosophers in the
context of their total thinking (and still less in the general context of their times);
nor is it faithful to the real complexity of their ideas. It is rather a rash exploitation
of the historical material, intended to procure a pedagogical introduction to a
number of themes which have stayed important in the philosophy of science ever
since Antiquity.
Expositions of the views of the Ancient (and later) philosophers which are
somewhat more faithful to complexities and context can be found in [Losee 1972].

33
whence the term. The single material representatives of the concepts – the
tables as representatives of Tablehood, the tragedy Medea as a represen-
tative of THE TRAGEDY as a genre – are precisely that: representatives
depicting genuine reality34.
Diogenes on the other hand is a materialist: only tables and tragedies
exist; Tablehood and Tragedy – the ideas – are our inventions. He is also
a positivist: we must distinguish that positive knowledge which we get from
observation of the real world from those figments of the mind which we
add ourselves – interpretive frameworks, concepts, generalizations and
metaphysical entities (nobody ever saw or put hands upon the force of
gravitation, an animal which was not a particular individual belonging to
a particular species, justice, or the tragic dimension of human existence).
With regard to another dichotomy, Plato can be characterized as a
realist35: according to him, the universals (general concepts, ideas) possess
the status of something really existing. Plato is indeed an objective idealist.
The ideas of Tablehood etc. do not have their fundamental existence as
images in our individual intellects – these are merely our only access to

34
In order to do Plato justice it should be said that the “Tablehood” and “Cuphood”
of the anecdote are parodic distortions of Plato’s real doctrine, which is concerned
with “ideas” like Courage and the Good, modelled on that Triangle which is the real
object of mathematical proofs even when a particular triangle has to be drawn for
the proof.
Nevertheless, the parody has a point in itself, calling attention to an inner
problem or even inconsistency in the Platonic and similar doctrines, already pointed
out by Aritotle (Topica 143b). Some geometrical proofs concern the Triangle, others
only the Right Triangle, still others all polygons. Does this mean that there exists
a particular idea of the Right Triangle, and which are then its relations to the idea
of the Triangle? Does, in a similar way, the idea of the Artefact (or Material Object)
split up into sub-ideas for tables, cups, ...? When taken to this consequence, the
simple structure of the universe of ideas dissolves into an indefinite and virtually
infinite number of nested and intertwined sub-, super- and interwoven ideas.
35
In later years, it is true, the traditional term has been taken in a very different
sense, grossly corresponding to materialism: The outer world exists, and our
knowledge is about this. To avoid ambiguities we may then speak of Plato’s attitude
as an instance of concept realism, the doctrine according to which concepts
correspond to really existing entities.
The reason for this rather confusing linguistic innovation appears to be political:
Since the term “materialism” has mainly been used by Marxists since the late
nineteenth century, better avoid it!

34
that higher Universal or Divine Mind where they have their real abode.
Diogenes, on the other hand, is a nominalist, and regards the universals
as nothing but names (puffs of the voice, as the nominalists of the Late
Medieval universities used to say) which we invent and use as shorthands
to sum up collections of particular experiences or objects.
As one may perhaps guess from the expression “objective idealism”, another,
“subjective” idealism exists. Much modern positivism is of this breed. According
to subjective idealism, every reference to an external reality is ultimately nonsense:
our mind has access to and registers nothing but our sense impressions. These form
a forever impenetrable screen between our mind and anything which may lay
behind.36
At this point, we have probably exhausted the philosophical contents
of the anecdote. In order to complete the list of essential concepts we will
have to involve a third Ancient philosopher: Plato’s rebellious follower
Aristotle.
Aristotle was an empiricist. If Aristotle is to be believed, all knowledge
derives from experience and thus comes through our senses. The empiricist
attitude has been summed up in the maxim that “nothing is in the intellect
which was not first in the senses”. But Aristotle was no positivist, and no
nominalist. He held that both the particulars (the single tables, the single
tragedies) and the universals (Tablehood and Tragedy) existed – the latter
as essential characteristics objectively shared by the directly observable
tables and tragedies, respectively. The universals constitute the essence
which lays behind and determines the phenomena, that which appears to
and can thus be observed by our senses.
But Aristotle was no idealist of the Platonic ilk. He did not regard the
forms (a term which he came to prefer to Plato’s ideas) as the only really

36
The ultimate consequence of this principle is solipsism (from solus ipse, “oneself
alone”), the tenet that one’s own mind is the only thing of whose existence one
may be sure – everything external, including other persons and their minds, may
be nothing but dreams and imagination (however these are to be understood
without being contrasted with perception of the real world). No philosopher likes
being considered a solipsist, for which reason a variety of protective devices have
been invented by subjective idealists. But if the same strict logic is applied to these
inventions that made their inventors accept the sense impressions as an impene-
trable screen, they evaporate immediately.

35
existing entities, and the phe-
nomena as merely fleeting and
ephemeral, imprecise and ul-
timately non-existing representa-
tives. Tablehood is not something
existing besides and above the
particular tables; Tablehood only
exists as tables – the essence is
only there in phenomena. Aristot-
le is thus a materialist – but he
is no positivist. With regard to
the dichotomy between realism
and nominalism, he has been
labelled a moderate realist. Dio-
genes represents a flat rejection
of Plato’s point of view – an
antithesis; Aristotle, on the other
hand, can be regarded as integra-
ting and going beyond both
positions, producing a genuine If the dialogue between Plato and Dio-
genes ever took place (which in itself is
synthesis. not very likely), then this is the most
According to Aristotelian likely location: In the left of the picture
empiricism, it is the aim of scien- we seen the remains of a portico where
Diogenes spent much of his time. To the
tific investigation to determine
right the road from Athens to Plato’s
the essence of things through Academy.
observation of phenomena. The
method is induction: Examination of many horses allows us to find that
which by necessity will hold for anything that is horse – i.e., to find the
essence of the horse; examination of the constitutional history of many city
states will lead us forward to knowledge of the essence of the state; etc.
When induction has thus provided us with the essential truths
concerning a field, we may deduce through the application of logic alone
what must necessarily hold for particular phenomena without further
empirical observation. If it has been shown to be part of the essence of cows
to have four stomachs, we need only establish that Karoline is a cow in

36
order to know she got four stomachs.
Once again, mathematics, and in particular geometry, supplies the model for
this epistemology, as it is made amply clear by the choice of illustrative examples
in Aristotle’s main work on the philosophy of science, the Posterior Analytics. From
thorough investigation of the properties of geometrical figures we establish a
number of fundamental truths (“axioms”) concerning ideal points, lines and figures
in the geometrical plane; from these we may deduct, e.g., that the sum of the angles
of a triangle equals two right angles (i.e., 180°). When that is done, we only need
to establish that a given figure is a triangle, i.e., contained by three straight lines,
in order to know the sum of its angles; empirical measurement is no longer
necessary.
In Antiquity already it was objected to Aristotle’s methodology and
epistemology that induction is a logical impossibility. Even if we have
dissected 10 000 cows, we can never be absolutely sure that No 10 001
possesses the same number of stomachs as the others. Couldn’t Karoline
belong to that fraction of a percent which has only three stomachs? Or
perhaps be the exception? To accept four stomachs as an aspect of the
essence of the cow can be nothing but an inspired guess, the validity of
which is never absolutely guaranteed37.
This point of view is classified as scepticism. According to the sceptical
point of view, science can never establish necessary truths concerning the
essence of phenomena, only plausible truths38. It will be seen that

37
Alternatively, we may of course take four stomachs as part of the definition of
a cow. But then we shall only know whether Karoline is a cow when cutting her
up. If we want to know about empirical reality, definitions do not solve our
problem.
This is the point of another Plato-Diogenes anecdote: When Plato had defined
Man as a biped and featherless animal, Diogenes plucked a fowl and brought it
into the lecture room with the words, “Here is Plato’s man”. In consequence of
which it was added, “having broad nails” (Diogenes Laërtios, Lives ... VI.53, ed.,
trans. [Hicks 1980: II, 55]).
38
One may also speak of “probable truths”, as often done. But if so, then only in
a loose sense: logically seen, observation is just as unable to establish that it is “more
than 90% sure that all cows have four stomachs” as it is to establish a 100% necessary
truth. Strictly speaking, talking of quantified probabilities in such cases is pure
nonsense, since it tells that “in nine out of ten worlds, all cows have 4 stomachs” –
there is only one world, as far as we can know.
This observation may seem trivial. None the less, it is often overlooked in

37
scepticism, nominalism and positivism are related philosophies: if we
cannot know whether we have penetrated to the essence of things
(scepticism), then it is close at hand to claim that our general concepts
about the structure of the world are mere abbreviations and names which
we have invented for convenience (nominalism) – and it is tempting to
regard all discussion of a not directly observable essence behind observable
phenomena as metaphysical rubbish contributing nothing to the real process
of knowing (positivism).
According to the Ancients, scientific knowledge consists (roughly speaking)
in finding the objects which exist within this world, and in listing their properties
and characteristics (in technical language: it consists in establishing an ontology).
To them, the essences of things could thus be listed as objects and properties. Modern
sciences, on the other hand, also look for relations, structures, and dynamical
interactions. In the perspective of modern sciences, essences should thus involve
these categories (even though they rarely use the term essence, which smacks too
much of pre-Modern thinking)39. None the less, and irrespective of terminology
and the exact character of general/universal features, the old question remains
whether universals have any status within reality. Mutatis mutandis, the concepts
of objective and subjective idealism, materialism, empiricism, positivism, nominalism
and scepticism apply as much in the modern as in the Ancient world.

practice. The conclusion of, e.g., a medical double blind experiment is frequently
explained to be that “it is 95% sure that this new drug works”; but what should
be said is that there is only a 5% chance that anything as suggestive as our actual
outcome would occur accidentally if the drug has no effect.
39
Evidently, modern sciences also encounter the question of existing objects or
entities – linguistics, e.g., what kind of existence to ascribe to the phoneme /t/.
The status of sounds may be left to physicists, and so may also the specific t-sounds
in ten and steam (the first of which is aspirated and the second not – cf. [R. H.
Robins 1971: 122]). Linguists, however, who notice that the first of these sounds
[th] cannot occur in English in places where the second is possible, and vice versa,
speak of the two sounds as representing the same phoneme /t/. No physical
analysis (whether a frequency analysis of the sound or a description of the sound
production process) can reduce the two to one, since phonemic identity is language-
specific – the Danish r-sounds in arbejde and rede (one of which is vowelized and
the other not) represent the same phoneme; in Arabic, the former would be heard
as representing the initial phoneme / / of Alı̄, and the second the phoneme /ġ/
used in Baġdād. No other science can decide for linguistics whether, and in what
sense, phonemes exist – in the speakers’ minds or expectations, in some objective
structure of language, or as linguists’ shorthand or abstractions.

38
Empiricism and falsificationism
These pages are not written with the intention of summing up however
superficially the history of philosophy. The preceding section – let it be
stressed once more – was only meant to serve as a pedagogical framework
for the presentation of certain important concepts and views. The present
section will therefore leave pseudo-history (and step a bit outside the
confines of professional philosophy) and look at certain attitudes to the
production of scientific knowledge which are widespread today even
among those who take part in this production.
One of these attitudes is nothing but Good Old Empiricism, coined as
a rule of conduct rather than as a stringent philosophy. Formulated in
maxims it runs as follows:
Scientific explanations are only allowed to make use of concepts and to
postulate relations and structures which can be rooted in experience,
observation or experiment. Mythological explanations referring to entities with
no such empirical underpinning are inadmissible: they only obstruct genuine
scientific insight.
This programme contains some essential points – first of all that science
should deal with reality and not be spun out of pure imagination. It also
corresponds to the immediate impression which working scientists get from
most of their work: we simply describe (so we feel) what we find, including
the relations and structures which turn up in our material. But it does not
avoid fundamental problems.
Traditionally, the whole complex of logic and mathematics is regarded
as falling outside the domain regulated by the empiricist rule. The
statement
if I know that rain makes the street humid, and that it is raining, then I can
conclude that the street will be humid
appears to be unproblematically and unconditionally true – expressed in
the appropriate language it will have to be accepted by a Touareg in Sahara
knowing neither streets nor rain, and it stays true if “humid” is replaced
by “dry” or “spherical”. It is a tautology, a statement which is true because
of its logical structure.

39
 That difficulty is normally solved by means of a distinction (going back
to Kant) between two kinds of scientific statements: Synthetic statements
which deal with reality, and for which the empiricist claim must be upheld
if they are to be meaningful; and analytic statements (the theorems of logic
and mathematics), the truth of which follows from their own structure,
but which on the other hand tell us nothing about the structure and
properties of reality40.
According to this distinction, the truth of analytic statements is given
a priori, i.e., in advance (viz, in advance of experience). The truth of synthetic
statements only follows a posteriori, i.e., after experience.
Kantian philosophy, it should be noted, accepts the analytic and a priori
character of logical tautologies, the truth of which follows from definitions –
“all husbands are male”, if we define a “husband” as “the male part of
a married couple”. Likewise, it accepts the synthetic and a posteriori
character of normal descriptive statements. But “2+2=4” is, according to
Kant, a synthetic a priori: It cannot be false, whence it is a priori; but none
the less it tells us something about reality (e.g., that two married couples
are exactly what you need for playing bridge). The same holds for all those
categories which are a priori necessary for making experience: Space, time,
causality, etc. Precisely because we cannot know reality without making
use of these frameworks, knowledge about the frameworks by itself tells
us something about our reality, – namely, the only way in which we can deal
with reality.
As it will be remembered from the previous chapter, Piaget started out
to find the roots of the Kantian a priori categories, demonstrating that they
are in fact the outcome of a genetic process, and hence not really a priori.
Since they result from practical experience in the world (experience which

40
Einstein once summed up this distinction and its consequences in a nice aphorism:
“Inasfar as the theorems of mathematics refer to reality, they are not certain, and
inasfar as they are certain, they do not refer to reality” [1921: 1].
A remark on terminology may be useful: the English language allows a
distinction between “analytic” and “analytical”. The former adjective characterizes
statements whose truth follows from their structure; the second means “based on/
using analysis”. “Analytic” may also be used in the second sense, but the present
essay takes advantage of the possibility to distinguish, and observes a similar
distinction between “synthetic” and “synthetical”.

40
is made in interaction between our biologically determined equipment and
the outer world), they also seem to be synthetic though dealing with the
most general properties of reality, properties which reveal themselves in
any environment in which children are brought up. Piaget’s results thus
appear to imply a reduction of the classification of statements into a new
dichotomy: Analytic a priori, which only encompass tautologies by
definition; and synthetic a posteriori, which embrace not only normal
descriptive sentences but even the theorems of logic and mathematics.
Certain empiricist philosophers (e.g., Willard Quine [1963/1951]) make
the point that words can never be fully reduced to simple definitions (what,
e.g., happens to the concept of “husbands” in recent Danish matrimonial
legislation?). This would abolish the category of analytic statements
definitively. Ultimately all statements would have to be judged in the light
of the “empiricist imperative” formulated above; the apparent exceptions
will have resulted from a too naïve understanding of logic and mathemat-
ics.
Closer investigation of the empiricist program reveals more severe
difficulties than those which can be circumvented by segregating logic and
mathematics or by arguments that after all they constitute only relative
exceptions.
Firstly, experience is never completely pure, i.e., made before and
independently of every theory; if this was not clear before the invention
of the telescope and the microscope, it should at least be glaringly obvious
to anybody looking at the technical equipment of a modern scientific
laboratory. The cell biologist presupposes the functioning of contrast media
and microscopes and thus theories belonging within chemistry and optics;
the literary scholar and theatre critic watching Medea knows beforehand
that what goes on is a play (in any other case he would be morally obliged
to save Medea’s children); etc. Experience and theories are elements of a
network, in which certain elements can be regarded as more elementary
than others and less subject to doubt; they can therefore be regarded as
legitimate underpinning and background. But it is impossible to make a
sharp cut between empirical knowledge (constituting a theory-free back-
ground) and theories derived from this background knowledge, as required
by the empiricist ideal.

41
 Secondly, the scepticist argument remains valid. You can never,
however great the amount of your experience and the number of your
experiments and observations, derive even the simplest theoretical
generalizations with absolute logical necessity. All your acquaintance with
and examination of winged feathered beings will not force you to bring
forth the concept of birds; reading all the books written to date and listening
to all the monologues and dialogues of history will not by necessity make
you discover a deep structure in language. Even if we forget for a moment
about the impossibility to obtain immaculate empirical knowledge, we are
forced to conclude that “science” created in agreement with the empiricist
letter will be restricted to listings and descriptions of singular cases, and
perhaps to tables and statistics – “natural history” instead of biology. Theory
will never emerge.
Ringleader in the statement of twentieth century empiricism was the
so-called “Vienna circle” and a number of associates, who in the 1920s
formulated the programme of “logical empiricism” (by others often labelled
neo-positivism)41. Ringleader in the destruction of the programme was
the same group, astonishing as that may seem: Over decennia this school
tried off strategies for a complete empirical underpinning of scientific
statements, searching for a verification criterion by means of which precisely
those statements may be singled out that can be founded upon and proved
from experience, i.e., verified, from all those which cannot be verified and
which are therefore scientifically meaningless. Their work never produced
a verification criterion which could demarcate science from non-science
precisely, but instead a triple negative conclusion. Firstly, existing science
cannot be reconstructed in this way, i.e., reduced to a system of single
statements which directly or indirectly have their complete reference to
experience; secondly, no science embracing theoretical statements can be

41
“Logical” because of the way in which it tried to get behind some of the
vagueness of classical empiricism, including the “empiricist imperative” formulated
in these notes: science does not build directly on experience – perceiving something
and enunciating a theory belong at different levels of existence. But experience has
to be formulated in simple sentences which state what has been experienced
(“protocol statements”, since these are to be entered into the experimental protocol
of the scientist). In agreement with the rules of logic, these statements are then to
be combined into higher level statements (generalizations, “laws”, “theories”, etc.).

42
constructed in agreement with the prescriptions; thirdly, no verification
criterion can be found which distinguishes sharply between empirically
meaningful and empirically empty statements42.
As an alternative to the untenable demarcation by verification, the
Austro-English philosopher Karl Popper has proposed his own recipe. In
“naïve” formulation, and thus corresponding to the above “empiricist
imperative”, it can be summed up as follows:
We are allowed to use in our explanations whatever self-invented concepts
and hypotheses we like; but we should be aware that our hypotheses are indeed
nothing but hypotheses, preliminary explanatory models, and not the truth.
We should therefore constantly check our hypotheses as thoroughly as we can,
and we must reject them as useless as soon as they enter into conflict with
our observations of reality – i.e., as soon as they are “falsified”.
Popper did not invent this canon. In the main, it coincides with the
idea of the “hypothetical-deductive method”, which has been practiced
for centuries43: Instead of working our way inductively from particular
observations in order to find the inner regularities and laws behind
phenomena (their “essence” in the classical language), we guess at a set
of laws or relations and try to deduce predictions about how things will
behave if our guess is correct. If the prediction turns out to be correct, the
guess is regarded as strengthened or “corroborated”; if not, we try another
guess. Nor is Popper’s stress on the forever preliminary character even
of corroborated hypotheses original; already the American philosopher
Ch. S. Peirce emphasized that new counter-evidence may always turn up,
and labelled this idea fallibilism44.

42
The whole argument, together with the arguments against the existence of
genuinely analytic statements, is given in Quine [1963/1951].
43
E.g., by Newton in his analysis of the planetary system, although he tried to
present his results in empiricist garb.
44
Even Peirce of course has his forerunners. In the Malleus maleficarum, a handbook
in witch-hunting from 1484, it is taught that the inquisitor should never pronounce
any accused innocent however much she might seem so, only declare that no
evidence had so far proved her guilt (III.20, ed., trans. [Summers 1971: 241]). Even
then, new evidence might always turn up. Nobody should be acquitted, cases

43
 Popper’s fundamental idea is thus less original than his own writings
try to tell. Nevertheless he must be credited with spreading the gospel,
and indeed with making it a gospel, to such an extent indeed that Peirce’s
original term is at times used as a synonym for Popperianism45. Writing
in the wake of the Vienna circle (to which he was close in the late 1920s
without being a genuine member), whose use of verification as the criterion
of demarcation between meaningful and empty statements was presented
above, he was also the first to use falsification as a criterion of demarcation:
Statements and theories which are compatible with every imaginable
situation (and which can therefore never be falsified) have no place within
science46.

should be postponed for want of proof.

45
But not always! In his discussion of Popper’s methodology and aims, Lakatos
[1974a: 93-132, in particular 112 and 114] makes Peirce’s term cover the sceptical
position that any knowledge, including knowledge of presumed facts that are
supposed to falsify a theory, may equally well be mistaken. This “sceptical
fallibilism” is almost as far from Popper’s philosophical inclinations as can be.
46
As examples of such not genuinely scientific theories Popper [1972: 34] refers
to “Marx’s theory of history, Freud’s psycho-analysis, and Alfred Adler’s so-called
‘individual psychology’”. He illustrates this (p. 35)
by two very different examples of human behaviour: that of a man who pushes
a child into the water with the intention of drowning it; and that of a man who
sacrifices his life in an attempt to save the child. Each of these two cases can
be explained with equal ease in Freudian and Adlerian terms. According to
Freud the first man suffered from repression (say, of some component of his
Oedipus complex), while the second man had achieved sublimation. According
to Adler the first man suffered from feelings of inferiority (producing perhaps
the need to prove to himself that he dared to commit some crime), and so did
the second man (whose need was to prove to himself that he dared to rescue
the child).
In contrast, Einstein’s General Theory of Relativity would have been “simply
refuted” (p. 36) if starlight had not been seen to be bent around the solar disk
during the solar eclipse of 1919. According to Popper, this “risk” taken by the theory
is what makes physics scientific, in contrast to psycho-analysis and historical
materialism.
Rhetorical zeal makes Popper forget that his drowning episode is not of the
same kind as the eclipse observation. When it comes to describing single aspects
of events taken out of context, physics is no different from Popper’s aversions.
Physics too may explain that a piece of lead flies upwards (it has just been shot
out from a gun) and that it falls downward (it was shot upwards 50 seconds ago

44
Instrumentalism and truth
If our explanations are built on arbitrary constructions and remain
forever preliminary and subject to rejection at failure, they cannot be “true”
in the classical naïve sense – things which may be false tomorrow (because
they may then have been falsified) cannot be true today. If theories cannot
be claimed to be true, however, the best explanation of their role seems
to be that they are tools for practical action. There is thus a close connection
between Popper’s ideas and instrumentalism: Scientific theories have no
truth value, are neither true nor false. Since they cannot be judged on the
scale of true and false, we should not try to do so. Scientific theories should
be evaluated in the way you evaluate instruments, according to effective-
ness, e.g. for prediction.
According to instrumentalist tenets, Copernicus’ theory, which claims
the Earth and all planets to circle around the Sun, is thus neither more nor
less true that the Ancient Ptolemaic notion of a fixed Earth around which
Sun, Moon, planets and stars move. Both are applicable as models, and our
only reason to prefer Copernicus’ model is that it is simpler and therefore
gives rise to less complex calculations if both models are built up with
orbits corresponding to empirical observations47. Being no more true than
the alternative, Copernicus’ model is to be preferred for a reason that will
convince any artisan: It feels better in the hand.
A fundamental objection against the instrumentalist interpretation of
scientific statements is this: Instruments can be used for precisely that for
which they have been designed; they can be used for other purposes only
if their constitution reflects, or corresponds to, features which are shared

and is now falling back), or that water evaporates (the kettle stands on an electric
boiler) or freezes to ice (somebody cut out the current, the window is open and
the weather frosty). This oversight is rather typical of Popper’s ways, and may
provoke the question why so sloppy a thinker is worth mentioning. The reason
is three-fold. Firstly, Popper is not always rhetorical and therefore not always
sloppy; secondly, precisely his sloppy thinking has become extremely popular;
thirdly and finally, coming to grips with Popper is a useful step in the present
argument.
47
Strictly speaking, the Copernican model is only decisively simpler than the
Ptolemaic system if we refer to Kepler’s revision of the Copernican theory.

45
between the
intended use
and the other
possible uses. A
screwdriver can
be used for
many different
screws, but only
because they all
carry a notch;
and it can only
be used at all
because its edge
fits the notches Perspective drawing of the principles of perspective draw-
and its rotation ing. From Dürer, Unterweysung der Messung
symmetry corre-
sponds to the
rotation by which its target is screwed in. Similarly with theories. We may
claim that we judge them according to instrumental value; but we cannot
invent that value freely, it is revealed (or denied them) when they are
applied – “the proof of the pudding is the eating”. The applicability of a
tool is thus a consequence of its correspondence with certain features of that
reality to which it is applied – features that themselves are not brought about
by the tool. Similarly for theories regarded as tools: their truth value can
be explained precisely as a structural agreement or correspondence with features
of reality. Further on, this conception will be spoken of as a materialist
notion of truth. Evidently, as also demonstrated by the example of the screw-
driver, correspondence is something quite different from similarity, not to
speak of identity: a screwdriver provided with a notch instead of an edge
would be worth nothing. In general, theories as well as screwdrivers belong
to other categories than the reality described by the theories and the screw
to be put in by the screwdrivers. Reality consists, among other things, of
atoms, birds, emotional states, and poems (according to physics, biology,
psychology, and literary scholarship, respectively). Theories, on the other
hand, consist of words and other symbols. Only from an idealist point of

46
view do reality and theory belong to the same realm, both being in some
way idea or concept; but then Plato’s idealism postulates an absolute
categorical rift between real reality, i.e. the realm of ideas, and apparent,
material everyday reality48.
Instrumentalism is thus right in seeing family likeness between a
screwdriver and a theory, and has a good point in its subversion of the
metaphysical concept of truth which ultimately presupposes an idealist
stance; but it is mistaken in believing that the screwdriver and the theory
are alike because they are equally arbitrary with regard to reality. Provoca-
tively speaking we may say that the reason for the usability of the
screwdriver is that it possesses a structure which corresponds to certain
essential features of the structure and function of screws – it embodies,

48
That theories “consist of words and other symbols” points to another characteristic
which they must possess beyond “structural agreement or correspondence with
features of reality” if they are to be considered “true”: logical consistency (or, put
differently: words put together without consistency can correspond to nothing
beyond themselves, they are meaningless). Much work has been done in twentieth-
century formal logic to render precision to this requirement, which practical
scientific workers tend to treat no less commonsensically than the idea of “corre-
spondence with facts”. The Polish-American logician Alfred Tarski, in particular,
is known for having formulated a “theory of truth” determining the conditions
which must be fulfilled by a formal sentence system if it is to possess this logical
consistency; he is also known for having shown that attempts to determine the
truth or falseness of the sentences of such a system from within the system itself
lead to self-referential paradoxes of the type “this statement is false”. Truth has
to be ascribed from without, by a metalanguage.
Sciences are not written in formal but in technical languages which ultimately
derive from common daily language. None the less, Tarski’s latter observation is
important for understanding the difficulty with which we are presented when we
try to understand the nature of the “correspondence” between sentences (or
theories) and reality. If correspondence is revealed through interaction with reality
(“praxis” in a Marxist sense) functioning in the role of the metalanguage, then it
can not be discussed within the quasi-formal discourse of logical theory but only
in a (genuine) metalanguage which steps outside: a metaphorical language which
evokes for us our experience of such interaction – e.g. the above screw-driver. In
another metaphor: Perspective drawing is a way to render three-dimensional reality
in a two-dimensional plane. How you make a perspective drawing can be shown
in a perspective drawing (see the Figure) – but it is only because we know about
three-dimensional reality and move around in it that we give the right interpretation
to the Dürer’s woodcut, and that we see the present drawing as a cube and not
as a plane jigsaw:

47
in materialized form, part of the truth about screws and about our way
to deal with them49.
One decisive difference remains between the screwdriver and scientific
theory. The instrumental validity of the screwdriver is static and limited;
science, on the contrary, is in continual development, constantly searching
for new correlations and ever extending so as to grasp new phenomena.
Kepler’s Copernican cosmology is more true than Ptolemy’s planetary
system because it allows a unified treatment of celestial and terrestrial
physics (until Kepler, the heavenly bodies were supposed to move by
necessity according to other laws than those which governed movement
below the sphere of the Moon)50. By saying that the reason for the usability

49
We may give this a somewhat more precise formulation. The edge of the
screwdriver, of course, agrees with a feature of the screw, and with what we want
to do with screws. The rotational symmetry, however, beyond these, also fits our
hand and the way we move it when driving in a screw. It may hence be claimed
to correspond to the Kantian categories.
50
This extendibility is crucial if we want to formulate a truth theory which is
relevant for the humanities. Claiming that your interpretation of a Platonic text
is true because it coincides with Plato’s own understanding makes no empirical
sense – how do we know that it does? But interpretation of a Platonic text (or any
other past text) makes use of techniques which are also used in the present – some
of them in everyday dialogues with people with whom we share a material practice,
some of them in the court-room, where textual evidence is combined with material
evidence. If the interpretive techniques which we use on Plato do not function in
the communication and together with the material practice and evidence of our
own age, we will have to reject them as general tools.
Extendibility and coherence is also what allows us to distinguish between
historical truth (be it partial or preliminary) and useful myth. According to crude
instrumentalism, the Blut-und-Boden-“theory” and the legend that the Danish colours
fell from Heaven during the Danish crusade in Estonia (cf. [Høyrup 1993: 170
n.134])are as good historical truths as any, since they are socially and culturally
useful (or, we might say with hindsight, Blut-und-Boden is false because and only
because its implementation as social practice resulted in all German Boden being
occupied by the Allied armies, not to speak of German and Allied blood). As a
consequence of instrumentalism, the choice between such “truths”, when they enter
in conflict, comes to depend on power: as does, in the present case, the decision
whether Estonia is legitimately to be considered a Danish or a German satellite
(or Russian or no satellite at all, according to still other national myths). Extendibili-
ty and coherence, on the other hand, that is, the claim that the same techniques
should be applicable to all sources and that the different sources at our disposal
should yield a consistent picture, allows historians to dismiss the myths as, exactly,

48
of theories is that they reflect features of reality, we also claim that reality
carries objective features which can be reflected by theory – “objective” in
the sense that they are contained in the object, in that reality which is
studied. The assertion that theories are better (“more true”, as just said
concerning Kepler) if their range can be extended implies that the objective
features carried by reality are of general validity, that reality is coherent,
i.e., potentially one51.
The affirmation of instrumentalism, that there is no truth, and that we
should choose our theories as it fits our aim, ends up by being inverted:
No, our choice is not arbitrary and not subjective: The aim of science must
be to capture (as far as it is possible in the given moment) as many of the
objective features52 in as general a form as can be done, and thus to be –
in this sense – as true as possible. Only then will our knowledge be
instrumental.
This is the real crux of the empiricist imperative as formulated above;
this is Aristotle’s old programme, to find the essence behind phenomena,

myths.
It may be relevant to remember in this connections that the textual criticism
of Renaissance humanists, from Petrarch to Casaubon, consisted precisely in the
application of techniques used to expose forged juridical documents.
51
This is another way to approach a question dealt with in the end of Chapter II.
Here it was concluded that “the material world is constituted in a way that allows
an adequate practice if we order our perceptions as representations of permanent
objects”, and it was suggested that other fundamental cognitive categories and
schemes had a similar foundation. The same kind of argument applies in the case
of scientific knowledge: if the requirement of logical consistency and extensibility
works (as it normally does), then this must tell something about the reality that
our theories deal with.
But we might continue: the permanency of the fox was not absolute, although
we might discuss its shortcomings in terms of the same principle (exhalation of
air, etc.). If other fundamental schemes (including the requirement of logical
consistency) are also “biologically a posteriori”, we have no guarantee of their
absolute validity; we only know them to be a priori and hence inescapable in our
actual life.
52
As we shall discuss later on, however, our questions to reality determine the kind
of features that will be reflected. Only our aim of driving screws into the wall makes
the edge and the symmetry of the screwdriver relevant – if we wish to use our
screws as weight units, a pair of scales would be the relevant instrument.

49
but stripped of the belief that any definitive essence can be found once
and for all; and it is, for that very reason, the core of the dialectical-
materialist understanding of the aims and possibilities of scientific insight.
But still it tells nothing about the ways to attain this scientific insight;
let us therefore return from this excursion into the theory of truth to
Popper’s recommendations in the matter.

Instruments or models?
Occasionally, Popper formulates himself as if he were an instrumental-
53
ist . But his fundamental attitude is certainly different. This is made clear
by his judgment of people who do not reject a theory when (in Popper’s
opinion) it is falsified. His rhetoric is that of a preacher denouncing sin,
not of the carpenter censuring a bungler who reaches out for the screw-
driver when he is to knock in a nail. Even if truth is only preliminary,
maximal truth is set forth as a moral obligation.
Nor is Popper’s real point of view, however, identical with the italicized
imperative formulated above; or rather, this imperative he only uses for
polemical purposes – the point of view he is willing to defend in a serious
discussion is more sophisticated.
In order to see why we may look at the difficulties presented by the
“naïve-dogmatic Popperian imperative”.
Two objections were already raised against empiricism, where they
were equally relevant. Firstly, observation and theory belong on different
categorical levels. Therefore, facts cannot contradict theories; only statements
(e.g., about facts) can contradict other statements (e.g., predictions made
by theories). This was the problem which logical empiricism tried to
overcome by concentrating upon the connection between “protocol
statements” and theories, leaving to practicing scientists the translation
of observations into protocol statements. The same could of course be done
in the Popperian perspective. But this leads to the second objection. No

53
“The tentative solutions which animals and plants incorporate into their anatomy
and their behaviour are biological analogues of theories; and vice versa: theories
correspond (as do [...] honeycombs [...] and spiders’ webs) to endosomatic organs
and their way of functioning. Just like theories, organs and their functions are
tentative adaptations to the world we live in” [Popper 1973: 145].

50
observation is pure, every observation presupposes a number of general
cognitive structures or theories – increasingly so in contemporary experi-
mental science. But what precisely are we then to do when (a statement
about) an observational fact contradicts our predictions? If, e.g., a telescope
observation of the planet Mars finds the planet in another place than
predicted by the Theory of Relativity? Should we regard the Theory of
Relativity as falsified and reject it? The theory of the telescope? Both? Or
none?54
A tentative solution might be gained from the observation that the
functioning of the telescope has been confirmed through many other uses,
including observations of terrestrial phenomena, and that it is thus more
likely than the Theory of Relativity to hold water. But “confirmation”
belongs with empiricism, being in fact nothing but that “verification” which
falsificationism tries to replace. The solution thus ends up with the same
conclusion as that which came out of the analysis of the logical empiricists:
science cannot be analyzed into single statements which are confirmed or
rejected one for one: to some extent, scientific truth depends upon the
totality of the scientific explanation.
As already told, these objections hit naïve falsificationism on a par with
empiricism. A final objection, on the other hand, turns one of the objections
against empiricism upside down. Empiricism could not explain the origin
of theoretical concepts since they could not be derived directly from
experience. It is precisely the aim of falsificationism to make space for these
unsubstantiated yet indispensable entities. But the cost is as large as the
gain: falsificationism makes possible the existence of theoretical concepts

54
This is a somewhat simplified version of a real dilemma which once presented
itself to Newton and the Royal astronomer Flamsteed. Flamsteed did not find the
Moon where Newton had predicted it to be; Newton, however, was able to convince
Flamsteed that he had made wrong corrections for the refraction of light in the
atmosphere, and that the real position of the Moon was where (Newton’s) theory
would have it to be. In coming years he went on to correct Flamsteed’s supposed
“facts” time and again. See [Lakatos 1974a: 130 n.5].
The example may also be correlated with the discussion between Galileo and
his opponents when he published his new telescopic observations of hitherto
unknown celestial phenomena (the Lunar mountains, the satellites of Jupiter).
Among the objections to Galileo was the question, how he could be so sure that
the effects were not artificial creations of the telescope.

51
by disconnecting them completely from experience. In this way, theories
end up by being nothing but computational models, which bear no more
similarity to the reality they describe than the gears of a watch bear to the
movements of the Solar system – the only connection being that the
pointers of the watch can be used to predict the position of the Sun in the
firmament. If the precision of the watch is insufficient, we scrap it and
replace it by a different model: a digital watch containing no gears but only
a quartz crystal and a printed circuit.
This is not the way real theories change. When one theory is replaced
by another one dealing with the same field, the new theory will contain
some new concepts and certain new relationships. But the concepts are
rarely quite new, nor are the relationships. As classical mechanics was
replaced by the Theory of Relativity, e.g., a number of fundamental entities
like time, space, and mass (the first two being Kantian a priori categories,
we observe) had to be understood in new, more sophisticated ways than
believed till then; they also turned out to be mutually connected in ways
which Newton had not imagined. But they were not abolished. The pattern
of which they were elements was restructured, which changed their
meaning – much in the way cognitive schemes accommodate when they
are integrated into new cognitive structures during individual development.
As phonemes (cf. note 39) replaced letters as elements of linguistic analysis,
this was more than a mere change of names. Yet even though there is no
one-to-one correspondence between letters and phonemes, the agreement
is large enough to permit us to name most phonemes after letters. It was
indeed the introduction of the phoneme concept that allowed linguistics
to change its focus from written to spoken language with much less abrupt
changes than an unmediated reference to speech sounds would have
required55.
Theories are thus not mere computational models, and predecessor

55
We may also note that the decision to spell “ten” and “steam” with the same
letter t shows the generations who introduced and adapted the writing of English
(long before the emergence of any science of language) to have had an understand-
ing of sounds as elements of language not too different from that of modern
phonemic linguistics – a striking case of continuity of theoretical concepts in spite
of theory change (theory emergence, as a matter of fact)..

52
theory and successor theory are more intimately related than the two
watches. This falls outside the comprehension of falsificationism, which
is at best able to explain the continuity between classical mechanics and
the Theory of Relativity as a consequence of Einstein’s lack of imagination.
Curiously enough, this problem is solved by empiricism with the same
brutal elegance as falsificationism solves the problem which empiricism
creates concerning the justification of general concepts: if theoretical
concepts are after all founded in experience, then there is no reason to
wonder why they undergo only relative change instead of disappearing.
The two approaches to the problem of knowledge solve each other’s
difficulties – but in mutually unacceptable ways. They stand as thesis and
anti-thesis, in a way which in the first instance is as barren as a marriage
between Plato and Diogenes.

IV. A NEW APPROACH: THEORIES ABOUT THE

SCIENTIFIC PROCESS

It was already told that Popper is only a “naïve falsificationist” for

polemical purposes. But it is the naïve Popper who is generally known;
the naïve Popper is the real counterpart of empiricism; and the “philosophi-
cal” Popper is, after all, an attempt to keep together with string and tape
the naïve Popper where he falls into too obvious pieces. For all these
reasons, the naïve Popper is the more interesting of the two. The philo-
sophical Popper is (in the present context) mainly of interest as a step
toward that “realistic” Popper which his one-time follower Lakatos has
created through a reinterpretation of the key concepts of his theories, and
toward the understanding of scientific knowledge as resulting from a scientific
process.

53
Popper and Lakatos: theories or research programmes?
The philosophical Popper (whom I shall call Popper1 in the following,
while Popper0 is the naïve Popper and Popper2 is Lakatos’s construction)
differs from Popper0 on three essential points56.
Firstly, Popper1 does not take theories to be falsified by conflicts with
experience, i.e., by “facts”. As stated above, facts and theories belong to
different domains. Theories consist of statements expressed in words or
other symbols, and therefore they can only enter into logical conflict with
other statements. Theories are therefore not falsified by facts but, according
to Popper1, by statements dealing with facts – basic statements, in Popper’s
idiom, evidently a concept which is closely related to the “protocol
statements” of logical empiricism57. The “theory” “all swans are white”
cannot be in logical conflict with a bird in flesh and blood; what falsifies
it is the basic statement “here is a black swan”.
Superficially seen this is only a shift of the problem which does not
solve it – “basic statements” and facts still belong to different categories –
and a specification – how should a theory be in conflict with empirical facts
if it was not contradicted by the enunciation of these facts in statements?
But in connection with “improvement” no 3 the shift will turn out to have
important consequences within all ideologically sensitive scientific domains
(cf. below).
The next innovation replaces precipitate dogmatism with philosophical
and historical common sense. The infant mortality of theories would be
exceedingly high if every theory in conflict with (statements of) facts were
to be rejected. Grossly speaking, every theory is contradicted by lots of facts
at birth. But are we to reject Galileo’s law of falling bodies because it is
not obeyed by a falling withered leaf? Or a theory of prices referring to
costs of production because it does not fit rare stamps?

56
The exposition of these differences owes much to [Lakatos 1974a]. This article
is also the source for the labels Popper0, Popper1 and Popper2.
57
“What I call a ‘basic statement’ or a ‘basic proposition’ is a statement which can
serve as a premise in an empirical falsification; in brief, a statement of a singular
fact” [Popper 1972a: 43]. Further on in the same book Popper tries to construct a
fence between his own concept and that of the logical empiricists, which he finds
too “psychologistic”.

54
 Such rash rejections are not usual. In both cases you will have a definite
feeling that the deviations from theory are due to specific circumstances,
even though you may not yet be able to specify and quantify them. But
you would evidently be dismayed if the speed of heavy leaden balls and
the price of eggs went equally astray.
Popper1 attempts to formalize this consideration by restricting the range
of inconsistencies with regard to reality that will count as falsification.
Evidently such a restriction cannot be specified in general. But when
working on a theory you should point out (yourself and beforehand) the
specific domains where the theory should in any case hold good58; if it
does not, you should reject it without mercy (Popper’s rhetorical style) –
if you don’t, you are dishonest (ditto). According to Popper, Galileo should
thus “at his peril”59 determine beforehand that “if my law of falling bodies
does not fit within 98% the speed of a leaden ball of 20 kg falling 50 m,
I will burn my manuscripts and turn to literary criticism”. Theories should
“stick out their neck”: they should be unable to escape the hangman of
contradictory experience, should he happen to pass by.
The third difference between Popper1 and Popper0 relates to the problem
that scientific “facts” are obtained by means of methods themselves
presupposing theories, as in the case of the telescope observations of the
Moon60. Even this problem Popper gets around by making it a moral
obligation to choose in advance. Before testing your theories you should
also stick out your neck by deciding beforehand which theoretical
fundament you accept as unproblematic and hence above criticism. Woe
to the scientist who post festum, when his theory has got into trouble, starts

58
“... criteria of refutation have to be laid down beforehand; it must be agreed which
observable situations, if actually observed, mean that the theory is refuted” [Popper
1972: 38 n.3].
59
[Popper 1972: 42 n.8].
60
Popper himself will rather point to human weakness, but the conclusion is the
same: “... we may point out that every statement involves interpretation in the light
of theories, and that it is therefore uncertain. This does not affect the fundamental
asymmetry [between possibly falsifying observation and falsifiable theory – JH],
but it is important: most dissectors of the heart before Harvey [who discovered
the blood circulation – JH] observed the wrong things – those, which they expected
to see” [Popper 1972: 41].

55
doubting his telescope. He is, according to Popper, dishonest61.
Lakatos has composed an ultra-short-story demonstrating the divergence
between Popper’s methodological requirements and the real behaviour
of the scientific community62:
A physicist of the pre-Einsteinian era takes Newton’s mechanics and his law
of gravitation, (N), the accepted initial conditions, I, and calculates, with their
help, the path of a newly discovered small planet, p. But the planet deviates
from the calculated path. Does our Newtonian physicist consider that the
deviation was forbidden by Newton’s theory and therefore that, once estab-
lished, it refutes the theory N? No. He suggests that there must be a hitherto
unknown planet p´ which perturbs the path of p. He calculates the mass, orbit
etc., of this hypothetical planet and then asks an experimental astronomer to
test his hypothesis. The planet p’ is so small that even the biggest available
telescopes cannot possibly observe it: the experimental astronomer applies for
a research grant to build yet a bigger one.63 In three years’ time the new
telescope is ready. Were the unknown planet p’ to be discovered, it would be
hailed as a new victory of Newtonian science. But it is not. Does our scientist
abandon Newton’s theory and his idea of the perturbing planet? No. He
suggests that a cloud of cosmic dust hides the planet from us. He calculates
the location and properties of this cloud and asks for a research grant to send
up a satellite to test his calculations. Were the satellites instruments (possibly
new ones, based on little-tested theory) to record the existence of the conjectural
cloud, the result would be hailed as an outstanding victory for Newtonian
science. But the cloud is not found. Does our scientist abandon Newton’s theory,

61
One is tempted to ask how to characterize Popper’s own attitude. At the age of
17 he engendered his marvellous falsificationist epistemology, though in the naïve
“Popper0” version. In coming years, when he discovered the shortcomings of this
programme, he did not give it up. Instead he repaired it by means the distinction
between facts and “basic statements” and all the other subtleties belonging to
“Popper1”.
But since, as we shall argue below, the prohibition of a posteriori criticism of
the theoretical foundations of observations is ill founded, there is no serious reason
to censure Popper for his failure to submit to his own rule.
62
[Lakatos 1974a: 100f]. As in the case of my above Mars/telescope example,
Lakatos’s story refers to somewhat more complex but similar real-life events.
63
If the tiny conjectural planet were out of reach even of the biggest possible optical
telescopes, he might try some quite novel instrument (like a radiotelescope) in order
to enable him to ‘observe’ it, that is, to ask Nature about it, even if only indirectly.
(The new ‘observational’ theory may itself not be properly articulated, let alone
severely tested, but he would care no more than Galileo did). [Lakatos’s footnote]

56
 together with the idea of the perturbing planet and the idea of the cloud which
hides it? No. He suggests that there is some magnetic field in that region of
the universe which disturbed the instruments of the satellite. A new satellite
is sent up. Were the magnetic field to be found, Newtonians would celebrate
the sensational victory. But it is not. Is this regarded as a refutation of
Newtonian science? No. Either yet another ingenious auxiliary hypothesis is
proposed or ... the whole story is buried in the dusty volumes of periodicals
and the story never mentioned again.64
This story agrees well with what goes on within even the most exact
sciences. Within the realms of social sciences and the humanities, where
precise predictions are rare, and where the distinctions between facts,
theoretical notions and ideological conventional wisdom are not easily
established – there, as one might guess, the rules are even more rarely
observed.
A follower of Popper might reply that he knows: Popper does not
describe what scientists actually do – he proposes a programme which
would make science advance more rapidly and with fewer wasted efforts
than it actually does precisely because Lakatos’s story is correct65. Popper’s
rules would not ensure that no mistakes were made; but they would reduce
the number of mistakes and, especially, the time that is wasted on mistakes.
Our follower of Popper would be wrong. Scientific work according to
Popper’s rules would, like most “work according to the rules”, be a
synonym for a strike in disguise.
There are several reasons for that. Firstly one may ask what happens

64
At least until a new research programme supersedes Newton’s programme which
happens to explain this previously recalcitrant phenomenon. In this case, the
phenomenon will be unearthed and enthroned as a ‘crucial experiment’. [Lakatos’s
footnote; in the next note he refers to Popper’s polemics against the Freudian and
Adlerian psychologies which can be made agree with any state of the actual world,
pointing out, as done in the present pages on a simpler example, that the same
holds for Newtonian physics].
65
This answer is not always given by Popper himself, in particular not when he
lapses into Popper0. The point in his discussion of psycho-analysis and Marxism
versus the testing of the General Theory of Relativity in his [1972: 34ff] is precisely
that the latter, representing science, behaves differently from the former, representing
pseudo-science. Elsewhere, however, Popper is more clear about presenting a set
of prescriptions and no description.

57
when a theory has been falsified and therefore rejected. When planet p’
does not show up in the telescope, should we then reject Newton’s
understanding of the planetary system and of mechanical physics in
general? Should we stop calculating the dimensions of steel beams for
bridge buildings and make them at random instead? Having rejected
Newton’s mechanics we have no theory at all, since all predecessor theories
have also been falsified. The absurdity of the claim is blatant, and shows
the Popperian notion of “merciless rejection” to be empty jargon.
One may also – which is Lakatos’s main point – observe that Popper’s
understanding of the nature of a theory is much too static and formal. A
theory which is to fit Popper’s prescriptions is complete and fully finished –
a Tarskian formal language (cf. note 48); it consists of a set of formulae
(verbal or symbolic) which definitively state the mutual relations between
the concepts of the theory, and a set of rules which allow the translation
between theory and observation, i.e., allow observation to function as a
metalanguage telling which statements are true and which false.
Few theories, if any, have been born in that form. It is debatable how
many attained it before they died. Theories are born instead, as we shall
see in more detail below, as open structures. Only through the collective
work of the scientific community with the theory does one fully discover
its implications and possibilities and its relations to other theories66.
Already for this reason it is impossible to indicate when a theory is
conceived at which points decisive testing should be performed.
In social and human sciences, Popper’s methodology would give rise

66
This was in fact pointed out at one moment by Popper himself (in an article from
1940): “The dogmatic attitude of sticking to a theory as long as possible is of
considerable significance. Without it we would never find out what is in a theory –
we should give the theory up before we had a real opportunity of finding out its
strength; and in consequence no theory would ever be able to play its role of
bringing order into the world, of preparing us for future events, of drawing our
attention to events we should otherwise never observe” (reprint [Popper 1972: 312]).
And similarly: “... this dogmatism allows us to approach a good theory in stages,
by way of approximations: if we accept defeat too easily, we may prevent ourselves
from finding that we were very nearly right” [1972: 49]. Yet, as observed by Lakatos,
this glimpse of insight does not influence his general thinking and his magisterial
preaching.

58
to yet another problem, which has to do with the conventionalism of Popper1.
In general, conventionalism belongs to the same family as
instrumentalism. Like instrumentalism it holds that one theory is no more
true than another, competing theory – ascribing “truth values” to theories
is as nonsensical as ascribing colours. Whether we use one or the other
theory for our description of reality is decided by convention, no more
compulsory than the convention which makes us speak of “cigarettes” and
“pamphlets” and not of “cigamphlets” and “parettes”; empirical evidence
may at worst force us to change the way we interpret our theory, the “rules
of translation” between observation and theoretical prediction67. Scientific
objectivity is thus nothing but agreement about a shared convention – and
if you disagree with the majority of your scientific colleagues, you are
automatically the sinner against objectivity.
This breed of conventionalism is treated with as intensive scorn by
Popper as are Marxism and psychoanalysis. Conventionalism, indeed,
denies the falsifiability of theories, eschewing it through a reinterpretation
of the rules of translation between theory and observation. Popper sees
clearly and correctly that conventionalism can function as a cloak for
scientific opportunism and for facile thinking in grooves. But his own
philosophy contains obvious conventionalist elements: convention and
nothing but convention points out which kinds of conflict should be
regarded as falsification; and convention decides which theories should
be ranked as unassailable and which should be submitted to continuous
attempts at falsification68.
In ideologically sensitive areas, i.e., in particular within the social and
the human sciences, even this brand of conventionalism will easily entail

67
All this may be more clear from an example. If you sit in a train, you will
normally state that it starts moving after the doors have been closed; but you might
equally well state (and at times get the momentary expression) that you and the
wagon stay at rest, and the rest of the universe starts moving. In the first case, the
observation that empty bottles start rolling along the floor is explained as a
consequence of the law of inertia; in the second by a changing gravitational field.
The standard example is in larger scale but built on precisely the same principle:
the question whether the Earth or the Sun is at rest.
68
This is not kept secret by Popper, who speaks explicitly of his methodology as
“methodological conventionalism”.

59
stagnation in ideological opportunism. What is more easily agreed upon
by the majority than the set of already familiar, stereotype ideas? Once
more the objective scientist will be the one who accepts the received
opinions of respectable people, and the dishonest worker the one who
challenges conventional wisdom and sticks to his own scientific convic-
tions69.

As stated above, Popper’s one-time follower Lakatos has formulated

that more accurate epistemology which in his opinion might grow out of
Popper1, and has baptized it Popper270. The central point in Lakatos’s
epistemology is reflected in this labelling: “Popper”, in fact, does not refer
to the person but to what Lakatos calls a research programme, evidently
inspired by the person; Popper0, Popper1 and Popper2 are nothing but single
stages within the development of this programme.
Precisely this example may provide the most easy explanation of the
concept. A research programme is not, as a theory according to Popper,
a static and solid entity; it is a progression through a number of static
theories superseding each other. A research programme thus roughly
coincides with the more loose parlance of a “theory” as something which
is in continuous development.
The feature which welds a sequence of theories into a research
programme is the existence of a shared hard core, a set of notions about
the entities which exist within the field dealt with by the research pro-
gramme. In the Popper programme thus theories which cannot be derived
from empirical observation; falsifications which kill theories; and some kind
of facts or representatives of facts taking care of the killing. In the Newton
programme, material particles and forces. These entities are the tools which
the theories belonging to the programme apply in order to describe/explain

69
Popper’s requirement thus stands in curious contrast to Merton’s norm of
“organized scepticism” (see p. 138), which Merton illustrates by the German dictum
“ein Professor ist ein Mensch, der anderer Meinung ist”, i.e., one who does not
automatically submit to received opinions.
70
Rumour has that Popper2 is already in manuscripts written by the real Popper
but kept unpublished – maybe because their publication might reduce the famous
Popper0/1 to ashes.

60
empirical reality. In addition to this ontology, the hard core prescribes the
kinds of explanation which should be aimed at, the methods which are
to be preferred, etc.
The hard core is “hard” in the sense that it does not bow to conflict
with observational reality – it is, in another word, irrefutable. If we use the
Popper programme as an example, we see that empirical observations
similar to Lakatos’s short story may demonstrate that falsification à la
Popper0 and Popper1 does not describe the real process of scientific
development; but they cannot prove that the rejection of unsatisfactory
theories may not in some way or other be understood as a “falsification”.
Experiments might show that the gravitational force does not depend on
distance in the way Newton believed; but they could hardly prove that
forces are in general untenable explanations.
The protection of the concepts (etc.) belonging to the core at “any price”
is called by Lakatos a negative heuristic – a guide as to what one should
avoid finding. The core also contains a positive heuristic – a guidance
prescribing how increasingly extended ranges of reality are to be explained,
and in which order “anomalies” are to be solved. The existence and
efficiency of this positive heuristic is of course the reason for the cohesion
and continuation of the research programme – if Popper0 and Popper1 had
produced no interesting points there would have been no reason to stick
to the programme and to a fundamental idea like falsification.
The theories which make up a research programme are not only
gathered into a common heap because they are characterized by certain
shared features constituting a shared hard core. As already intimated, and
in the likeness of Popper0, Popper1 and Popper2, they are ordered in a
progressing sequence, T1, T2, T3, ... TN, ... .
In order to introduce more fully the relations between the members
of such a progression of theories it might prove useful to look at a more
substantial example than the Popper sequence. Such an example is provided
by the progression of economic theories built upon the labour theory of
value.
The first step, the one which originated the hard core and the pro-
gramme, was Adam Smith’s Wealth of Nations from 1776. In this work,
Smith formulated a theory of the price of a commodity as proportional

61
to the working time required to produce it71. This doctrine was no loose
postulate but argued from the competition between workers as well as
between manufacturers and thus connected with the concept of three social
classes72. This was a radical innovation with regard to the preceding
physiocratic conception, according to which only work in agriculture was
productive, while all kinds of industrial transformation (e.g., grain into
flour and bread) were unproductive. ( Cf. [Høyrup 1993: 129-135, passim]).
In his more systematic Principles of Political Economy and Taxation from
1817, David Ricardo took over the labour theory of value, and used it
among other things to explain the mutual advantage of foreign exchange
in a market system. By using the concepts of competition and scarcity he
also managed to explain from the theory how the mere possession of
agricultural land would allow landlords to earn money (the rent)73.
Ricardo, however, wrote in a situation where some industries were
significantly more “capital intensive” than others; this difference had been
less conspicuous 40 years before, when Smith wrote his book. Ricardo knew
that the difference had to influence prices, without being able to integrate
this knowledge into the theory of value, and therefore restricted himself

71
Evidently, my exposition of Smith’s and other economic doctrines is cut to the
bare bones, and simplified to that extreme where only the features which are
essential for the epistemological argument stand back.
72
“Those who live by profits”, i.e., capitalists; “those who live by wages”, i.e.,
workers; and “those who live by rent”, i.e., landlords. In Smith’s England, it should
be remembered, landowners would normally lease their land to farmers. The latter
would thus be counted as capitalists, whereas the landowners (who owned only
land but no means of production) were a separate class.
73
In brief: If a country needs N tons of grain per year, this requires that the best
A acres are cultivated. Some of these acres yield more than others at the same
expense of labour, but the price of the grain will of course be the same. Competition
will fix the price at the level corresponding to the labour costs of the poorest land,
which is the highest level: if the price were lower, nobody would care to cultivate
this land, which would result in shortage, famine and raising prices; if it were
higher, even poorer land could be cultivated with profit, which would lead to
overabundance and falling prices. The landlords possessing the best land will thus
get more from their grain than the labour costs – or they will lease to capitalist
farmers who will be willing to pay the difference between their selling price and
their labour costs as rent.
Keine Hexerei, nur Behändigkeit!

62
to the statement that the working time required to produce a given
commodity would determine its price until at least 93% – one has spoken
of Ricardo’s theory as a “93% labour theory of value” [Barber 1967: 95,
92 n.9]. If we speak of Smith’s theory as T1, Ricardo’s will be T2.
Both T3 and T4 are due to Marx. T3 solves the difficulty which arises
if the labour theory of value is applied to the price of labour itself, i.e., to
the wages. It would seem that a labour cost determination of wages should
lead to a payment for 8 hours of work which would be able to buy
precisely the product of 8 working hours. This would leave no space for
profits – in flagrant contrast with the normal state of affairs.
In Marx’s writings from the late 1850s onwards (T3) this problem is
solved. Prices, according to this theory, are still determined by the working
time needed to produce commodities; but the wage is not the payment
for the working time but for the working power. The working time needed
to produce 8 hours of working power is the time normally required to
produce that working power, i.e., to produce the commodities which the
worker consumes in order to keep going for another day74. If the working
class of a country only consumes half of its social product, we see that the
time used to produce what an average worker needs to go on for another
average day is produced in 4 hours. The price of 8 hours working power
equals 4 hours working time. This leaves space for profit.
Another problem is still unsettled – viz Ricardo’s problem of varying
capital intensities. This difficulty is only resolved in volume III of Das
Kapital, which was published posthumously by Engels in 1894 (T4). The
breakthrough consists in a separation of the concepts of “value” and
“price”. The value of a commodity is now defined as the working time
normally required for its production under given technological, societal
and historical conditions. If prices were equal to values (after a suitable
conversion of time into monetary units), capital would be invested in those
sectors where a maximum of work was done (and hence a maximal profit

74
Of course averaged over life, so that the costs of procreating and feeding children
is included. In the present simple version, a two-class model for society (capitalist
and working classes and nothing else) is presupposed.

63
earned) per unit of invested capital75. These sectors would soon be
overproducing compared to actual demand, while those depending on
larger investments per working hand would be underproducing. The
imbalance between demand and offer would make prices fall in the
overproducing sectors and make them rise in the underproducing ones.
This would continue as long as profit rates differed; in the end, prices
would be stabilized precisely so far from values that the profit rates of all
sectors were equal76. Ricardo’s problem is solved – indeed by means of
theoretical considerations borrowed from his own theory of rent. At the
same occasion another difficulty dissolves: How to explain within the labour
theory of value the incomes of banks, wholesale merchants and rentiers.
After the death of Marx, only Marxist economists continued work
within the framework of the labour theory of value (grossly speaking). The
reason was obviously the political consequences of the doctrine, as they
had been uncovered by Marx77. Further development was branched, as
was the labour movement itself. One further development (T5’, T6’, etc.)
consists of inconsistent crisis theories (Rosa Luxemburg, Ernest Mandel
and others). Another branch contains a better theory of the dynamics of
economic crises (T5, Kalecki) and the solution of subsisting problems
concerning the relation between value and price (T6, Sraffa).
What follows from this whole story? First of all that Lakatos’s research
programmes are no description of real history but “rational reconstruc-
tions”, in Lakatos’s own words. Marxism is certainly more than a further
elaboration of Adam Smith’s research programme, and one gets no genuine
understanding of Marx’s thought from isolated exposition of his economic
analyses. Worst of all, Marxist analyses of the increasing monopolization

75
Equality of prices and values would mean that the profits per working hour would
be same in all sectors; the more capital you need in order to employ one worker,
the less will be your profits per invested £ Sterling.
76
In the actual world they will of course never be completely stabilized – continuous
technological development is one of several factors which cause the point of
equilibrium itself to be moving, as emphasized by Marx.
77
Gustafsson [1968: 14-16] lists a variety of sources which document this explicit
concern. The main problem was the separation of working time and working power
(T3), which automatically entailed a concept of exploitation.

64
of capitalist economies after 1870 are not easily fitted into a rational
reconstruction relating everything to the Smithean starting point78.
At the same time, however, the story shows that Lakatos is far from
being totally wrong – the rational reconstruction reflects central aspects
of the historical development, and can thus be claimed to be a true theory
for the structure of scientific development in the sense explained above.
Finally, the process exhibits some of the characteristics which according
to Lakatos distinguish the development of research programmes.
Firstly it shows that research programmes may easily live with
“anomalies” – points where they disagree with observation without being
able to explain precisely why. Ricardo’s 93% theory is in fact nothing but
an attempt to talk away an acknowledged anomaly, the influence of capital
intensity on prices. The labour theory of value was not dismissed by
Ricardo because he got stuck in a problem which could not be solved for
the moment. Nor did the difficulty paralyze Marx, though only two
theoretical breakthroughs allowed him to solve it.
Still, a research programme cannot live with all its difficulties without
doing something about some of them. If Ricardo had only been able to
introduce his 93%-restriction and had not increased the explanatory power
of the programme on other points (of which foreign exchange and rent
were mentioned), the programme would have degenerated (Lakatos’s term),
and it would have been abandoned by active research as soon as an
alternative approach had been found.
Changes which are not degenerative are called progressive problemshifts
by Lakatos. A progressive problemshift occurs when a new theory is both
empirically and theoretically more powerful – if it predicts more than the
predecessor (greater theoretical content), and if it predicts better (greater
empirical content). If we forget about the 93%-restriction, the whole
sequence T2-T3-T4-T5-T6 consists of progressive problemshifts. T5’, however,
which aimed at increasing the theoretical content of the theory, was no
progressive shift: on one hand, it did not increase the empirical content
of the theory; on the other it was ridden by inner inconsistency. The same

78
In justice it should be said that Lakatos did not propose the application of the
research programme idea to the development of the labour theory of value; but
similar features would turn up in many other instances.

65
holds for T6’, Mandel’s attempt to show how the spread of fully automatic
industry would entail the collapse of capitalism.

Theories falsified by theories

An important feature of Lakatos’s conception is his notion of falsifica-

tion. A theory, as we have seen, is not falsified by an anomaly, however
serious. According to Lakatos, a theory is falsified by another theory – by
a theory with greater theoretical and empirical contents, by a theory which
so to speak explains why its predecessor failed to explain specific anomalies.
This is in itself a progressive problemshift, solving or dissolving no
less than four of the central dilemmas presented by Popper1:
— Firstly, the question what to do in the interlude between the falsification
and resulting merciless rejection of one theory and the devising of a
replacement. There is no such interlude, since falsification only follows
from the development of a new and better theory.
— Secondly, a difficulty which, though not formulated explicitly above,
follows from the lack of continuity of theoretical concepts through the
cycle of falsification and ensuing free invention of a new theory. At
best, Popper’s methodology might bring forth a sequence of increasingly
precise models of reality; but even in the best of cases the falsification
cycle will never procure us with increasingly deep theoretical insight: every
time a theory is falsified and thus rejected we replace it by something
which in principle is totally new (like the digital watch replacing the
mechanical watch). We cannot raise the question what was wrong in
the rejected theory: It is the model as a totality that is wrong and thus
rejected. Within the framework of the materialist notion of truth (see
p. 46) we may say that if the key concepts of the hard core reflect
essential aspects of reality, then the research programme allows an
increasingly exhaustive investigation of these features, and thus an
increasingly objective reflection.
— Thirdly, falsification à la Lakatos does not invite to ideological oppor-
tunism as does Popper’s methodological conventionalism. On the
contrary: If a theory is only regarded as falsified by another theory
which offers deeper and more precise explanations, then disagreement

66
 with superficial ideology and conventional thinking will be a less
threatening argument against it79. Theories get greater opportunity
to confront reality directly, bypassing the censorship of received
opinions.
— Fourthly and finally, the paradox evaporates that every observation
is polluted by theoretical or proto-theoretical presuppositions. If theories
are falsified by theories this is no longer a source of logical trouble but
only another expression of the tangled character of scientific (and other)
knowledge: a totality which cannot be fully analyzed into mutually
distinct elements, be it into the verified and thus meaningful statements
of empiricism or into Popperian basic statements and static theories.
At one point, Lakatos’s conception (Popper2, as we remember) can be
claimed to constitute a degeneration with respect to Popper1 (the sophisti-
cated real Popper). Popper’s aim is to formulate a logic of scientific
research80, a formalized system which can be set before the scientific
community as a set of rules which it ought to obey. Lakatos’s rational
reconstructions preserve this aim to some extent. But Lakatos has given
up the conviction that the falsification processes of the rational recon-
struction (not to speak of that real history which it rationalizes) can be
formulated in a way which complies in full with the requirement of
formalization81: who is able to balance the degenerative versus the

79
Once again we may correlate with Merton’s “organized scepticism”: whereas
Popper’s prescriptions would tend to undermine this norm, Lakatos explains it
and makes it a methodological necessity.
80
This is the best English translation of the original German title of [Popper 1972],
and describes Popper’s intentions adequately. The actual English title (The Logic
of Scientific Discovery) may be better for advertisement purposes but misses the point
completely: The only part of the research process which according to Popper should
be completely free and subjectively creative and not subject to any logic is precisely
the phase of discovery, the invention of a new theory. Formalization and strict rules
belong with the control process, the compulsory stubborn attempts at the life of the
assumed discovery.
81
Conversely, of course, Lakatos also gives up the belief in the completely unfettered
process of invention: as long as innovation takes place within the same research
programme, it is guided by the positive heuristic and restricted by the negative
heuristic, both of which are constituents of the hard core of the programme. But
even if guided and restricted and thus no act of pure subjectivity, innovation has

67
progressive elements in the shift from Smith to Ricardo? Everybody, of
course – but very precisely to balance, not to state the definitive and
indubitable decision. In many instances it is also only through extensive
and meticulous research that one is able to decide whether a theory
possesses greater empirical content than a competitor; it may even be that
agreement with observation improves on some points and decreases on
others. The decision can no longer be reached by an impersonal and
objective judge, it is attained instead in an informal arena, viz the scientific
community “voting with its feet”. The choice between theories is the sum
of individual choices made by individual workers deciding which theory
they are going to make the foundation for their own further work.
Paraphrasing the jibe against Ricardo, Lakatos’s theory can be characterized
as a “93% logic”.
However, rather than speaking against Lakatos’s epistemology this
observation tells something about the concept of degeneration: Degeneration
need not be a defect, even though this is the obvious moralistic implication
of the term. It may just as well be a rejection of empirically degenerative
aberrations contained in earlier theories. In this vein we may look at
Popper0 and Popper1 as aberrations within a research programme starting
informally from a hypothetical-deductive understanding of scientific
method, and at Lakatos/Popper2 as an alternative and more fruitful
development from the same roots. In any case, theoretical degeneration
need not be a development for the worse – as well known by military
planners, a tactical retreat from unwisely occupied positions may be the
only way to avoid imminent defeat and the best preparation for further
offensives.
Two final points should be addressed before we leave the presentation
of Lakatos’s ideas. Both may, like the discussion of degeneration, be
approached through the example offered by the theory itself.
According to Lakatos, the normal situation within a scientific discipline
is the existence of several competing research programmes. Evidently, the
philosophy of science offers a striking example of this. During a protracted
period, e.g., logical empiricism and Popperianism were both pursued (along

not become formalized.

68
with other programmes). Similar examples can be found in many other
disciplines; still others, on the other hand, seem to be dominated by one
programme at a specific moment.
Here it is important to remember that falsification takes place within
research programmes. One research programme cannot (if we follow
Lakatos) be falsified by another programme, because its hard core is
irrefutable. Research programmes are not falsified, they are given up when
they degenerate and when alternative choices are at hand82.
To some extent, the process by which one research programme displaces
another thus constitutes a parallel to the falsification within a research
programme. Yet the parallel is imperfect: a new research programme does
not necessarily get the upper hand because its empirical content is larger
than that of the programme which it supersedes; it may be preferred
because it explains a specific anomaly which has resisted the predecessor
so stubbornly that the whole validity of this programme has come to be
doubted. This is what happened within chemistry in the late eighteenth
century, when a programme explaining combustion as the absorption of
a new chemical element oxygen replaced a predecessor theory which
explained combustion as the liberation of an igneous substance called
phlogiston. Among other things, the phlogiston theory had explained the
colours of many chemical substances. Within the framework of the oxygen
theory, these colours became inexplicable. None the less, the oxygen theory
was preferred, because the areas where it possessed increased empirical
content were considered more important83.

82
In rare cases, programmes have even been considered degenerating beyond hope
and have been given up by practitioners notwithstanding the absence of alternative
programmes. In such cases the whole discipline has been abandoned by the
scientific world and considered a pseudo-science, and an earlier belief in the results
of the programme is declared superstitious or at least illusive. One example of this
process is phrenology, a nineteenth century science about the alleged relation between
people’s character and the form of their skull (cf. [Høyrup 1993: 155f]). The rejection
of astrology by astronomers in the seventeenth century may be looked at in the
same perspective.
83
One of these areas was the specific weights of chemical substances – if weight
was to be conserved in chemical processes, phlogiston had to have changing and
sometimes negative specific weight.

69
 From Lakatos’s point of view, this is a decisive difference. All things
considered, however, it seems to amount to no more than a difference of
degree. True enough, the abandonment of one programme for another
cannot be described as a formalized process – it results from a process of
balancing and “voting with the feet”. But precisely the same was, though
to a lesser extent, the case when we considered the falsification process
within a programme.
The last feature of Lakatos’s epistemology is that it is reflexive, i.e., that
it is able to describe itself. Although it is nowhere said, exactly this must
be the coquettish point in Lakatos’s use of the term Popper2 as a label for
his own approach. In so far as it claims to be a description of the actual
process of knowing and of its conditions, i.e., to be itself knowledge about
that process, reflexivity must of course be required from any epistemo-
logy84. Yet far from all epistemologies are in fact reflexive. As already
hinted at, Popper’s own rules would have forced him to give up his ideas
as exhaustively falsified if he had followed them. Empiricist philosophy
is no better off – logical empiricism, in particular, would probably be forced
to see its own statements as ultimately meaningless if it applied its own
verification standard. Similar auto-destructive conclusions will be reached
in the cases of instrumentalism and conventionalism.
Still, Lakatos’s theory is not the only reflexive offer on the epistemology
market. The epistemology of dialectical materialism has the same character-
istic85. Moreover, full reflexivity is only achieved by Lakatos if he is given

84
Evidently, reflexivity is in itself no proof of the adequacy of an epistemology.
As pointed out in note 48, the truth of a system cannot be proved by self-reference.
But as a minimum it must be required that the self-references contained in or
implied by a system which pretends to be true are of the type “this statement is
true” and not variants of the so-called liar’s paradox, “the present statement is false”.
85
But not the kind of vulgar Marxism that claims consciousness to be nothing but
reflection of the socio-economic circumstances under which it is produced, and
rejects the relevance of any discussion of it in terms of truth value. If thinking in
general should be understood on a par with Pavlov’s conditioned reflexes, why
should then the status of vulgar Marxism itself be different? Similar conclusions
will be reached in the case of other unrestricted sociologisms, as also when we look
at Skinner’s behaviourist epistemology or other deterministic theories – indeed for
all epistemologies which deprive knowledge of the possibility of being a true

70
a materialist interpretation, through which a truth value can be ascribed
to a research programme and its appurtenant theories in spite of the
metaphysical and irrefutable character of its hard core.

The limits of formalization

Lakatos’s epistemology is able to grasp essential features of the
development of scientific knowledge. But it is not able to grasp all essential
features – no theory is. And it has not solved all the problems to which
it directs attention.
Like Popper and the logical empiricists, Lakatos still regards science
as a formalized system: A theory consists of unambiguous statements
dealing with concepts and their mutual relations, and of “rules of trans-
lation” telling how the predictions of theory and empirical observation may
be compared; – and science, on its part, consists of theories.
This conception Lakatos shares with Popper, and for that matter with
logical empiricism. To be sure, their formal understanding of the system
of scientific knowledge does not imply that Popper and the logical
empiricists (nor, a fortiori, Lakatos) have not discovered the importance
of ideas without formalizable foundation for the development of know-
ledge. Quite the contrary, the logical empiricists distinguished sharply
between the “context of discovery” and the “context of justification”: they
were fully aware of the possibility that the context of discovery of ideas
may be far removed from empirical verification, involving intuition,
religious and metaphysical ideas, etc. What they asked for was that an idea,
once proposed, in the context of justification could be “verified” empirically.
Popper, on his part, made a cardinal virtue of what logical empiricism had
felt forced to accept: new theories should be freely devised, any attempt
to make rules for this process would be inimical to science. Only in the
moment when the theory has been formulated does the merciless effort
to falsify set in. The whole model looks as if inspired by traditional

description/reflection of the real world (cf. [Høyrup 1993: 190]).

In general, any epistemology claiming possession of a complete and exhaustive
explanation of the nature of knowledge can only be reflexive if it is able to explain
the existence of complete and exhaustive knowledge. No middle road appears to
exist between Platonism and epistemologies which are as open as that of Lakatos.

71
liberalistic ideology: anybody should be allowed to settle down as a
shoemaker or as a railway baron, free of state regulation and control; but
in the moment when he has started his business, he should be subjected
to the objective verdict of the market, which kills unsound undertakings
without mercy. Like attempts to keep a falsified theory alive, efforts to keep
unsound businesses afloat through public intervention will only do damage
to the common good.
Lakatos is less of a liberalist than Popper in his epistemology. The
concept of a “research programme” and of progression within the
programme, and particularly the idea of a positive heuristic, describe the
process of invention as less than fully arbitrary and as somewhat open to
theoretical comprehension – as taking place within a certain pattern. But
the origin of research programmes and of their hard cores is still left
outside the area considered by the philosophy of science as inaccessible
to formal analysis. This does not invalidate the rest of his analysis, but it
remains a serious restriction that the theory leaves out of consideration
the life-giving moment of the development process as unexplainable; what
would we think of a theory of ecological metabolism which explains that
animals live from each other and ultimately from plants but disregards
the photosynthetic process through which plants secure the energy supply
for the total ecosystem?
In other respects too, Lakatos’s search for formalized structure (and
his desistence from describing what cannot be sufficiently formalized)
creates more problems than it solves. The distinction between falsifiable
theories and the irrefutable hard core of a research programme is surely
meaningful. Yet the two levels can not be regarded as absolutely distinct;
in the case of the labour theory of value, for instance, we must probably
see the separation of price and value (T4), and perhaps even more the
separation of working time and working power (T3) as so radical reinterpre-
tations of the foundations of the programme that its hard core is changed.
If we look at the shift from Popper1 to Popper2 we must also acknowledge
that the new concept of falsification is so far removed from its predecessor
that even here the hard core is affected – no wonder that Popper rejects
Popper2.
Finally, the absolute separateness of research programmes is dubious.

72
It is true that different research programmes build on different sets of
cardinal concepts, and no complete translation from one programme into
the other is possible – cf. the relation between empiricism and naïve
falsificationism. But they are still connected via their reflection of features
of the same reality (if we presuppose a materialist view), and often of the
same features (even though the problem for empiricism and naïve
falsificationism seemed to be that each of them was formulated with regard
to features which were inaccessible to the competing programme). Thereby
the possibility emerges that the concepts of one programme may be
explained at least with some approximation in terms of the core and
theories of the other; perhaps one programme may even develop to the
point where it is able to explain the accomplishments and the failures of
the other – which was the criterion for falsification within a programme.
Apart from being built upon the basic premiss of a materialist view (viz that
knowledge reflects features of a reality existing independently of the knowledge
in question), this conclusion corresponds to experience borrowed from the history
of a variety of sciences.
If we look at the confrontation between the phlogiston- and oxygen-theories,
the former was mere nonsense as seen from the stance of early oxygen theory, and
its triumphs nothing but lucky accidents. In some instances, in fact, the “substance”
phlogiston “turned out” (thus oxygen theory) to be identical with carbon or
hydrogen; in others it represented the absence of oxygen. But the development
of the concept of “degrees of oxidation” in the later nineteenth century provided
an explanation of what was common for carbon and the absence of oxygen. It thus
became clear which features of reality (as seen by the mature oxygen theory) were
reflected in the doctrine of phlogiston. Phlogiston theory, which had originally been
abandoned, could now be seen as falsified in Lakatos’s sense.
Corresponding examples can be found everywhere. Most obvious is perhaps
the relation between Ptolemaic and Copernican planetary astronomy. If we accept
the Copernican system (or one of its later variants) it is easy to calculate how
planetary orbits behave as seen from the Earth, and hence to see how the Ptolemaic
model manages to account with relative precision for the position of planets on
the celestial vault. But even the more ambiguous field of social sciences offers some
instances – as many, indeed, as can be expected in a domain where woolly
conceptualizations and cross-fertilizations often make it difficult to speak of distinct
research programmes.
A striking example is provided by the relation between the labour theory
programme and that “neo-classical” or “marginalist” programme which replaced
it within academic economics after c. 1870, when the former programme had become

73
politically unacceptable (cf. above, n. 77). The neo-classicists started out from
concepts and problems which were explicitly incompatible with the labour theory
of value. Asking for a theory which was equally valid for rare stamps and for eggs,
it had to take its starting point in consumers’ preferences and not in the costs of
the producer86. But gradually the neo-classicists were forced to approach the
questions that had occupied Smith and Ricardo, viz the global economic process
of society. At that moment they had to develop a concept of the price of production
which determined the long-term price level of products which (like eggs) could
be produced in any quantity87. This price of production turns out to be explained
by arguments that follow the fundamental structure of the discussion in Das Kapital
III of the relation between value and price (only published some years later). Marx
is hence able to explain Marshall, just as Copernicus/Newton is able to explain
Ptolemy88.
Lakatos, we may say, sees the development of a scientific discipline
as consisting of a number of parallel lines (each representing a research
programme) competing for the favour of the scholars of the field and
terminated at the moment when favours fail. A more realistic view, on the
other hand, would have to look at the lines as partly interconnected.
Lakatos’s idealization is correct in so far as the connections between
research programmes are weaker than the connections between theories
belonging within the same programme; but an understanding which aims
at getting beyond Lakatos’s formalization should start by recognizing the
existence of interconnections89.
We may add that even the lines themselves (the single programmes)

86
Jevons, The Theory of Political Economy, published 1871.
87
Marshall, Principles of Economics, 1890.
88
In both cases, on the other hand, the reverse explanation (Ptolemy of Newton,
Marshall of Marx) turns out to be impossible, for the simple reason that Newton’s
and Marx’s theories include dynamic explanations which fall outside the scope
of their competitors.
Marx, on the other hand, is not able to explain Keynes’s theory of the economic
cycle; this is only done by Kalecki (T5) – see [Robinson & Eatwell 1973: 48ff].
89
This is no point of pure philosophy but carries an important message for practical
scientific work: You should never dismiss the reflections and theoretical results
achieved by another school with the argument that they belong within another
research programme and that they are therefore irrelevant for you. Dialogue is
possible and often the crucial condition that you may progress along your own
road – not least within the human and social sciences.

74
possess an inner structure. Branchings are common, and it is not always
possible (as it was in the case of the labour theory of value) to distinguish
between a sound trunk and vain aberrations. The solution of single
problems (concerned, e.g., with specific anomalies) may be the occasion
for the emergence of specific sub-programmes within the same global
research programme. At times such sub-programmes may be absorbed into
the main programme when a satisfactory solution to their specific problems
has been found; at times they may provide the starting point for a new
discipline or sub-discipline.

Kuhn: Paradigms and finger exercises

Chronologically and historically, Lakatos’s concept of “research
programmes” is an attempt to describe from a Popperian point of view
an approach to the problem of scientific knowledge which in many respects
constituted a radical break with established ideas. Making Kuhn – the
originator of the new approach – appear as a commentary to and an
elaboration of Lakatos’s ideas is thus a pedagogical trick and no reflection
of historical connections. None the less, the trick may be useful.
Thomas Kuhn is a former physicist turned historian of science and no
philosopher. This is clearly reflected in his book The Structure of Scientific
Revolutions ([1970]; 1st ed. 1962), in which his ideas were first presented
in print. It does not, like Popper’s presumed “logic” of research, attempt
to prescribe rules which are supposed to guarantee more steady scientific
progress; instead, Kuhn’s first aspiration is to find structure and coherence
in the baffling imbroglio of the history of the sciences; his second aim
(which need not be secondary) is to understand why this structure is able
to produce knowledge, and to show how it may indeed be adequate and
perhaps even necessary for the production of scientific knowledge,
regardless of its conflict with time-honoured ideas about the nature of good
science90.

90
This second question was formulated in the title of Kuhn’s contribution to a
symposium on “Scientific Change” held in Oxford in 1961 (published as [Kuhn
1963]): “The Function of Dogma in Scientific Research”. Some years later, Kuhn
[1974: 237] formulated his double approach as follows: “The structure of my
argument is simple and, I think, unexceptionable: scientists behave in the following

75
 The central concepts in Kuhn’s understanding of scientific development
are the paradigm; normal science; and the scientific revolution. Normal science
is science whose development is governed by a paradigm, and a scientific
revolution is the replacement of one paradigm by another. The paradigm
itself is thus an adequate starting point.
The term is borrowed from traditional language teaching, and is another
name for the exemplar.
An exemplar or paradigm is a word which is used to train a conjugation
scheme – as in Latin amo, amas, amat, amamus, amatis, amant, or in German
ich liebe, du liebst, er liebt, ... . Other words belonging to the same category
(a-stem verbs and weakly conjugated verbs, respectively) will then have
to be conjugated “in the same way”, in a way which is understood quite
as much through subconscious training as from explicit rules. The point
of using the paradigm instead of the abstract sequence of endings -o, -as,
-at, -amus, -atis, -ant is precisely this subconscious way of functioning. If
you had only learned the latter system you would have to switch from
speaking to analytical thinking each time you were to use a verbal form.
The paradigm, on the other hand, functions much in the same way as the
subconscious sensorimotor schemes described by Piaget91 – or it may serve
at least as the starting point for the construction of a subconscious
scheme92.
The key point in Kuhn’s approach to scientific activity is that it is a

ways; those modes of behaviour have (here theory enters) the following essential
functions; in the absence of an alternate mode that would serve similar functions,
scientists should behave essentially as they do if their concern is to improve
scientific knowledge”.
91
Cf. also what is said on p. 12 regarding grammatical schemes.
92
It should be observed that in structuralist linguistics the term paradigm is used
in a way which differs fundamentally from Kuhn’s : In the rudimentary sentence
structure “subject – verb”, the phrases “a dog”, “the bird”, and “Susan” are part
of one paradigm, the set of words or all phrases which may serve as subject, and
from which precisely one element is to be chosen; the phrases “runs”, “dies”, and
“is born” belong to another paradigm.
This use of the term is derived from its meaning in traditional grammar, too;
even from the sequence ich liebe, du liebst, ..., one element is to be chosen when a
sequence is to be constructed.

76
creative and active practice in the same way as the use of language. You
learn to use German verbs through reading and speaking German and
through the training of paradigms, not from the mere reading of grammati-
cal rules; you learn to ride a bicycle by trying and not merely through
explanations of the role of the handlebars for maintaining balance (I still
remember the explanations I got at the age of five; they would have sent
me headlong into the pavement, had I ever tried to follow them); you learn
to play the piano through finger exercises, transposition exercises and
training, not from mere explication of the musical notation of the cor-
respondence between single notes and keys, and of the major/minor
system. In a similar way you learn to work inside, e.g., the Newtonian
research programme by using its theories and by observing their use, not
from a mere abstract exposition of “Newton’s three laws” and the law of
gravitation or of the “hard core” of the programme with its appurtenant
negative and positive heuristic; you learn to perform a structuralist analysis
of a literary work by doing and by following analytical work, not from
mere exegesis of the principles of structuralism.
– not merely from theoretical and abstract exposition, though evidently
also in this way. Scientific work does not stop at being skill and knack, it
is also a conscious activity. Researchers are not sensorimotor infants but
analytically minded adults that integrate the schemes of their cognitive
unconscious as tools for conscious operatory thought. The essential point –
and a point which is neglected by both Popper and Lakatos and indeed
by almost all philosophers of science – is that scientific activity also contains
an essential element of skill.
From where, then, do scientific workers get their skill? In former times,
before the systematic training of future research workers in universities,
by reading THE BOOK – that decisive book which had moulded the basic
understanding of their discipline. Astronomers read Newton’s Principia;
before this seminal work was published they read Kepler or Copernicus93;

93
With one historically important exception: After the Galileo trial, Jesuit astrono-
mers read Tycho Brahe, who allowed the Earth to remain quiet. Even after the
publication of the Principia they were supposed to do so. However, since they
undertook to translate and publish explanations of the Newtonian system, we may
presume that they used it just as much as other astronomers at least for training

77
and before Copernicus was accepted they had read Ptolemy’s Almagest.
The Principia, Kepler’s Astronomia nova, Copernicus’ De revolutionibus and
the Almagest functioned, each in their time, as those exemplars through
which astronomers were trained to see the planetary movements, conscious-
ly and subconsciously, as astronomers could be expected to see them, and
to analyze the problems of their field as currently done.
These books thus functioned in a way analogous to that of the
paradigms of language training, which explains the origin of Kuhn’s central
term. As it often happens, however, the actual meaning of the term came
to differ from the etymological origin – in fact already in Kuhn’s own book.
The paradigm concept, if it had referred to the role of such books alone,
would have described an earlier stage in the development of the sciences
and nothing more. In modern times, natural scientists are trained by means
of textbooks and prepared exercises, not by following immediately in the
footsteps of the founding fathers of their field; they will only be confronted
with original research papers at a relatively late stage of their education,
and rarely at all with the classics94. In the humanities, early confrontation
with research literature is customary; but one will seldom find (neither
at present nor in the past) a field to be defined by one book to the same
extent as physics was once defined by Newton’s Principia and economics
by The Wealth of Nations.
Even though Kuhn does use the term “paradigm” to denote the pivotal
books which once defined their respective fields, he therefore mostly uses
the term in as somewhat different sense, viz about that collective attitude,
that collective intuition, those shared techniques and that “tacit knowledge”95

purposes, irrespective of conceivable metaphysical reserves.

94
Quite a few biologists, of course, will read (passages from) Darwin’s Origin of
Species, some physicists may take a look at Galileo’s Discorsi, and many economists
may study some chapters from Smith’s Wealth of Nations. But these classics are so
removed from what goes on now in the respective fields that they can have no
genuine training function; having them on your bookshelves and having looked
into them is rather a way to affirm your professional identity.
95
The concept of “tacit knowledge”, which has been amply used in explanations
of the Kuhnian view, was created by the philosopher-chemist Michael Polanyi. The
insights which (even then with a considerable delay) gained wide currency with
Kuhn were thus not totally unprecedented – Polanyi is not the only precursor.

78
which natural scientists once got by working their way meticulously
through THE BOOK, and which is now acquired in other ways. In a
postscript to the second edition of his book, Kuhn even proposes to reserve
the term paradigm for the shared “constellation of beliefs, values, techniques
and so on” [1970: 175; cf. 191], and to label the “exemplary past achieve-
ments” instead exemplars for the sake of clarity.
The paradigm in the sense of a “constellation of beliefs ...” is a totality,
and those constituents which can be brought to light through analysis will
only direct and govern scientific work because they are parts of an
integrated whole. Recognizing this restricted value of analysis, however,
should not prevent us from having a look at the constitution of the whole.
One element of the paradigm may be familiarity with an exemplar,
a fundamental work or group of central works. In certain cases this
exemplar need not belong to the discipline itself – thus, the works of the
anthropologist Claude Lévi-Strauss (La Pensée sauvage etc.) and the linguist
Ferdinand de Saussure (Cours de linguistique générale) have played the role
of exemplars for structuralist currents within many disciplines.
More important than the exemplar itself, however, is what you learn
from it. The contribution of the exemplar to the paradigm may be found
on several different levels. From Newton’s Principia, e.g., you may learn
about the actual movement of physical bodies influenced by forces. More
generally, you learn that the forces acting upon one body originate with
another body, and that the acceleration of the body multiplied by its mass
equals the total force acting upon it; you learn mathematical techniques,
and you learn that these techniques are the means by which you compute
the movement of bodies. You learn a precise, “Euclidean” deductive
construction of your line of argument, and thus that physics may be (and,
implicitly, should be) constructed as a rigorous deductive progression of
propositions and calculations. You learn that physical theory should relate
to and explain phenomena, and you learn how to relate theory to pheno-
mena.
From Adam Smith’s Wealth of Nations (which functioned as an exemplar
in classical British political economy) you also learn on several levels at
a time. You learn to divide the population of a country into social classes
according to the source (not the size) of their income. You learn that the

79
relevant sources are wages derived from work; profits derived from the
possession of means of production (capital); and rent, derived from the
possession of land; for which reason the classes are working class, capitalist
class, and landed proprietors (cf. note 72). You learn about competition
and its effects, and about the formation of monopolies and about their
consequences. You learn about quantity of work as the factor which
determines prices within a market economy. You learn that economic
analysis presupposes social statistics and historical considerations. You
learn a specific way to analyse and to argue.
What you learn from an exemplar may thus be summed up as follows:
— You learn about the kinds of entities which constitute the world of the
discipline: Physical bodies, forces, ... / kinds of income, social classes, ...
(in philosophical jargon: an ontology).
— You learn which types of explanations belong legitimately within the
discipline – which explanations should be used by a physicist and an
economist, respectively. Implicitly, you also learn which kinds of
explanation should be avoided (the moving intelligences of celestial bodies
and just prices, respectively, to mention kinds of explanations used
before Newton and Smith).
— You learn about a number of techniques which can be used to attack
the problems occurring within the discipline, and you learn how to
use them.
— and you are provided with a total idea of what the world (of the
discipline) looks like, a global perspective on (the pertinent) reality.96
The paradigm is thus related to the “hard core” of a Lakatosian research
programme (no wonder, since the hard core is just Lakatos’s explanation
of the paradigm concept from a Popperian perspective). But there are
important differences. The Kuhnian paradigm is not as precisely delimited
and formalizable as Lakatos presumes his hard core to be. A “total idea”
and a “global perspective” cannot be summed up exhaustively in well-
defined propositions. Learning “how to use” the techniques of a discipline

96
In the postscript to the second edition of The Structure ... [1970: 187ff], Kuhn
introduces some more precisely defined constituents of the paradigm; but since
these are geared specifically to the paradigms of physical sciences they need not
concern us here.

80
is the acquisition of a skill; skills one may speak about, but a skill in itself
is not something which can be enunciated (as can a theory or an ontological
presupposition). Presupposing his hard core to be clearly expressible,
Lakatos can imagine that a scientist may reject one research programme
and start working upon another by a fully conscious choice. The idea of
the paradigm as containing an prominent factor of training and skill, on
the contrary, implies this shift to involve more of a new learning process
and less of a free instantaneous choice (the choice actually involved is the
choice to start learning anew, to assimilate a new perspective which is not
yet yours and thus not fully understood in the moment you choose).
Ultimately, the paradigm shift is not an individual affair but rather a
process affecting the whole disciplinary community – “a new scientific truth
does not triumph by convincing its opponents and making them see the
light, but rather because its opponents eventually die, and a new generation
grows up that is familiar with it”, as Kuhn [1970: 151] quotes Max Planck
(the physicist who took the first step in the development of quantum
mechanics in 1900). The paradigm involves elements of collective intuition,
and intuition, as we all know, cannot be changed by deliberate choice or
majority vote.
Becoming familiar with an exemplar is not the only way you learn to
work within a paradigm. It is even questionable whether you learn it in
full in that way. The kind of knowledge which is contained in the exemplar
may contribute to the collective intuition; yet it is mainly through working
as a physicist or an economist while using the exemplar as a navigation
mark that you make the exemplar paradigmatically productive. This is why
the contemporary training of natural scientists (and, to a large extent,
economists, sociologists, linguists, etc.) can be successfully effected without
exposition to exemplars but by means of textbooks and appurtenant
exercises, the gist of which is that the exercise is to be performed as
presupposed within the paradigm; and this is why many fields of human
science can transmit their paradigms through exposing students to select
pieces of current research literature combined with independent work.

81
The structure of scientific development
Kuhn’s primary aim was never to describe the socialization of future
research workers. It was to understand how scientific fields develop. But
once the insight was gained that the dynamics of scientific development
could not be understood unless the moment of production of this knowledge
was taken into account (cf. the initial passage of Chapter III, p. 31), the
socialization of workers turned out to be pivotal: The distinctive character
of scientific knowledge must then depend, among other things, on the
particular way workers within a field see and deal with this field, and
hence on the process that makes them see it thus.
When a new field becomes the object of systematic (“scientific”)
investigation for the first time, there is as yet no such particular way to
see it and deal with it. Those who approach it do so from common sense
understanding of its character and common sense definition of its con-
tents97, and from a general intention to understand it “scientifically”. As
examples of such “pre-paradigmatic” (proto-)sciences one may take
“women’s studies” from around 1970 and the study of electrical phenomena
between 1600 and 1750.
In a pre-paradigmatic science the approaches are multiple and
uncoordinated. The results obtained by one worker will normally build
on presuppositions and refer to concepts which are not shared by others,
and others will therefore have difficulties in assimilating them. Instead they
will tend to be neglected and eventually forgotten, maybe to be rediscover-
ed 20 or 40 years later98. Borrowing the Piagetian idiom we may say that

97
With the reserve that they will often have been trained as scholars within other
fields. Their “common sense” is thus the common sense of the general scholarly
community as tainted by their specific training within particular fields. As a friend
of mine once asserted about a former physicist who had gone into peace research
and from there into sociology, where she had met him as a teacher: “A is a
physicist; he will never be anything but a physicist”.
98
This is, for instance, what happened to Vladimir Propp’s analysis of the invariable
morphology of (Russian) fairy tales from 1928: It only became influential in the
1950s, when Lévi-Strauss and others had established the structuralist paradigm –
within which, by the way, the implications of Propp’s findings were interpreted
in a way that differed decisively from Propp’s original “diffusionist” understanding
of the matter. Cf. the prefaces and introductions to [Propp 1968] and [1984].

82
the workers in the field possess no common cognitive structure which is
fit to integrate unexpected results and to keep them available for further
use and elaboration by the community as a whole. Pre-paradigmatic
sciences are not cumulative – at best, single schools with a certain inner
coherence (as found, e.g., in women’s studies from the mid-seventies
onward) exhibit cumulative characteristics.
It may happen, however, that a particular contribution or a specific
school obtains so convincing results that other workers of the field accept
its approach as an exemplar, trying to emulate it (the precise nature of the
contribution is irrelevant, as long as it only convinces and is able to
transmit some relatively coherent approach to the field). This breakthrough
may start the development of a genuine paradigm, and as long as this
paradigm serves, the field is in a phase of normal science.
During such a phase, work is directed at expanding the paradigm: to
understand more and more of reality through the paradigm. One may
speak of applying the theory to new areas, or of developing new theory
for new areas on the basis of the paradigm; the latter formulation may be
preferable, since the expansion to new areas may require addition of new
concepts and presuppositions, and an articulation of the paradigm with
regard to its original content by which it is made more precise, explicit
and conceptually clear.
This articulation is the other aspect of what goes on during the normal
science phase. Clarification of concepts and increasing adaptation of the
paradigm to that reality which is studied may result as secondary effects
of the expansion of the paradigm – if you apply the outlook generated
within women’s studies to the situation of sexual minorities or suppressed
racial groups, then you get a new perspective even on the original core
of your field, and you get new skill in dealing with it (cf. below, p. 164).
Similarly if you apply Newton’s laws to the flow of water through pipes,
or the principles of structural phonology to analysis of kinship structures.
But articulation may also follow from conscious efforts to get better insight
into the foundation of earlier results.
Much work within normal science is concerned with the solution of
“puzzles” (Kuhn’s term). The metaphor refers to such everyday pheno-
mena as riddles, crossword puzzles and chess problems. In all of these

83
we know that a solution exists; it is only up to our ingenuity to find it.
The same thing characterizes normal science: since the paradigm “knows”
what the world (of the discipline) consists of and which types of relations
hold good between its constituents, all problems within this world must
(if we are to believe the implicit claim of our paradigm) be solvable; the
question which remains open is whether I am smart enough to find it. If
a problem resists my efforts, at least my colleagues in the field will
conclude that I was not – after all, they know from proper experience that
the paradigm is fully efficient for all relevant purposes. Only if others fail
like I did will the problem cease to be a mere puzzle and become an
anomaly which challenges the paradigm.
The appearance of an anomaly may lead to focusing of work on
precisely this stubborn problem (cf. Lakatos story as told above), and then
perhaps to a solution; or it may remain unsolved, and if the paradigm
remains effective in other respects it may then be encapsulated while work
goes on under the paradigm without being disturbed99.
The puzzles which are taken up during a phase of normal science are
not selected at random. In combination, the global view which the
paradigm gives of the constitution of its object and the array of results
obtained until a given moment will suggest which problems are now most
urgent and most likely to be solved100. This explains the recurrent pheno-
menon of simultaneous discovery: extremely often, the same essential
discovery is made by workers who have no direct connection and indeed
nothing in common beyond a shared paradigm; this, however, is enough
to make them take up the same problem, and provides them with sufficient

99
This happened to the discovery that the perihelion of the planet Mercury (the
point where it comes closest to the Sun) rotates (“precesses”) in a way which cannot
be explained by Newtonian mechanics. The anomalous precession was discovered
in the early nineteenth century, and everything suggested that it should be
explainable. It was not, and was thus shelved – and was ultimately solved by the
General Theory of Relativity a full century later.
100
This process is seen en miniature each time a professor allots thesis topics to
doctoral students. The teacher is expected to know in advance which questions
are now solvable: it would be irresponsible to make students run on a track leading
nowhere, but equally irresponsible to make them repeat what has already been
done. The teacher is, so to speak, supposed to be the paradigm in person.

84
knowledge about what should be expected to make them see an actual
outcome as epoch-making.
Pre-paradigmatic science was not cumulative. Normal science is.
Stubbornly, it sweeps up everything which the paradigm is able to
interpret, shelving anomalies encountered in one direction if it is still
successful in others. Eventually, however, anomalies accumulate and tend
to turn up at all essential points. The efficiency of the paradigm for puzzle
solution shrinks, and during the effort to explain one or another anomaly,
the paradigm is articulated in increasingly divergent fashions. Eventually,
without having been given up the paradigm may exist in almost as many
versions as there are active workers within the field; ultimately, this will
of course undermine its credibility (and obliterate its character of shared
beliefs etc.). The field will end up in a state of crisis, where doubt about
the efficiency of the paradigm grows into general distrust of its previous
accomplishments: If phlogiston theory runs into paradoxes when trying
to account for specific weights, are we then really entitled to believe in
its explanations of colours? Are these not likely to be spurious and only
accidentally in agreement with observations? The willingness to engage
in quite new approaches spreads, varied proposals come up, for a while
different schools may exist alongside each other. In many ways, the
situation is similar to that of the pre-paradigmatic phase. Only when one
approach has proved its ability to solve precisely those problems which
had become central during the crisis period (and only if competitors are
unable to do it as satisfactorily) will this approach come to serve as the
starting point for a new paradigm, inaugurating a new phase of cumulative
normal science.
The shift from one paradigm to the next constitutes a scientific revolution,
which is characterized by sharp rupture. Taking the Copernican Revolution
as an example, Kuhn [1970: 149f] suggests that we consider
the men who called Copernicus mad because he proclaimed that the earth
moved. They were not either just wrong or quite wrong. Part of what they
meant by “earth” was fixed position. Their earth, at least, could not be moved.
Correspondingly, Copernicus’ innovation was not simply to move the earth.
Rather, it was a whole new way of regarding the problems of physics and
astronomy, one that necessarily changed the meaning of both “earth” and

85
 “motion”. Without those changes the concept of a moving earth was mad.
The content of a concept is only partially to be derived from empirical
observations (cf. what was said above in the discussion of the problems
of empiricism); in part it depends on the total theoretical structure in which
it partakes and the practice inside which it serves101. Mutatis mutandis,
the observations on the changing meaning of terms must therefore hold
for all paradigm shifts. The discourses before and after a change of
paradigm (or across a paradigmatic border) are “incommensurable”. A
conference may “bring people to talk to each other who would never read
each other’s papers”102; but it was my definite impression on the occasion
where this was formulated that they did not understand each other too
well.
Many of Kuhn’s early critics (and quite a few superficial followers in
later years) have taken the claim for incommensurability to imply that no
communication and no rational argumentation is possible across the
paradigmatic border. This is evidently a wrong conclusion, built among
other things on an absolutistic concept of rationality, and it was never
intended by Kuhn103. Breakdown of communication is partial. This suffices

101
In the sciences, this practice is in part constituted by the research process, in part
by teaching and applications. For the astronomers of the later sixteenth century,
astronomy teaching in universities (which had to be traditional) and the computation
of planetary positions to be used in court astrology (which by necessity asked for
these positions as seen from the Earth) were no less weighty than astronomical
research (see [Westman 1980]).
102
Mogens Trolle Larsen, formulated at the dinner table the last evening of the
symposium “Relations between the Near East, the Mediterranean World and
Europe – 3rd to 1st millennium BC”, Århus 1980. The participants were mostly
archaeologists falling in two groups: those oriented toward social anthropology
and the use of statistical analysis of the distribution of finds – and those for whom
“the only facts are artefacts”, i.e., for whom archaeology should make no theorizing
about societies and their structure and interaction but simply dig and describe the
finds and their stratification meticulously.
During the symposium, an exasperated member of the former group commented
upon the attitude of the latter with the phrase “Oh yes, the only acts are artefacts!”;
the immediate answer was a simple “Yes, of course”.
103
In his postscript to the second edition of his Structure ..., Kuhn [1974: 198ff] takes
up in some detail the problem of incommensurability and the misunderstandings
to which his original statements had led.

86
to exclude unambiguous proofs that one part is right and the other is wrong;
but it does not prevent critical, rational discussion, where appeal can be
made to those cognitive structures which are shared across the border104.
The situation bears some similarity to the description of the same situation
in two different languages possessing non-isomorphic conceptual struc-
tures105.

104
The only partial breakdown of communication distinguishes Kuhn’s analysis of
the paradigm shift from two apparently related lines of thought: Wittgenstein’s
notion of “language games”, and Foucault’s “archaeology of knowledge”.
Wittgenstein’s analysis [1968: §11 onwards] comes close to Kuhn’s in pointing out
that a language game is connected to and rooted in a particular practice; but it leaves
no space for description of the process by which one “paradigmatic language game”
develops into another (for good reasons, since this is only a characteristic of certain
“games”, like the paradigms of scientific disciplines and – with some modification –
artistic “schools”), and it leaves aside how the practice underlying the language
game is itself to some extent a result of the game.
Foucault’s analysis produces a much coarser grid than Wittgenstein’s multiple
coexisting language games. He speaks [1966: 13f] about two prominent discon-
tinuities in the Western episteme, one around the mid-seventeenth century (which
we may connect to the reception of Descartes and Galilei) and one in the early
nineteenth century (the epoch, e.g., of Comtean positivism). But he is even more
explicit than Wittgenstein in his statement that seeming continuities within single
sciences over one of these watersheds (e.g., between the “general grammar” of the
mid-seventeenth century and modern linguistics) are nothing but surface effects;
in Foucault’s view, Linné’s biology has much more in common with seventeenth-
century Grammaire générale than with Cuvier’s comparative anatomy or Darwin’s
theory of evolution. This may be quite true, but only in a perspective which
concentrates on other aspects of the disciplines in question than their relation to
their object, and which eo ipso (as also stated by Foucault) excludes any idea of
cognitive progress through the shift, and indeed any critical communication.
Maliciously one might maintain that Foucault is only right (in this respect) under
the perspective where it does not matter whether what he says (in this and any
other respect) is right, only that the way in which he says it reflects a particular
French intellectual style and the make-up of the French book market; one need
not be a follower of Foucault to find this to be a distorting and reductive perspec-
tive.
105
Cf. the relation between the conceptual clusters “knowledge/cognition” and “Wis-
sen/Erkenntnis/Erkenntnisvermögen”. “Cognition” encompasses only little of what
is covered by “Erkenntnis” and most (all?) of what is meant by “Erkenntnisver-
mögen”, and “knowledge” correspondingly more than “Wissen”. This is one among
several linguistic reasons (non-linguistic reasons can be found) that epistemology
looks differently in English and German; still, translations can be made that convey

87
 The analogies between Kuhn’s and Lakatos’s formulations are evident.
As already stated, the paradigm corresponds to the research programme,
normal science to work within a research programme. Still, differences are no
less conspicuous106. One of them turns up if we look for the analogue
of Lakatos’s “hard core”. A paradigm possesses no hard core, no sharp
distinction between the absolutely inviolable and that which can be freely
reinterpreted and changed in order to obtain better agreement with
observations. All levels of a paradigm may be affected by articulation.
In spite of articulation, however, the main task which normal science
sets itself is the solution of puzzles, where the paradigm not only “ensures”
(i.e., assures) that solution is possible but also mostly tells what the
approximate outcome should be (in experimental or other empirical
research), or how an explanation will have to look (in theoretical investiga-
tions). If things turn out in a totally unexpected way, and if they cannot
be explained even with hindsight to agree with what could be expected
(that is, if they constitute an anomaly), the results will often be neglected
(as told above). Rebellious thought is rare, “dogmatism” prevails.
This agrees badly with common sense ideals concerning the character
of science and the behaviour of scientists (not to speak of Popper’s rhetoric).
It is also at variance with the way scientists experience their own work:
the efforts to grasp things in new ways, the struggle to get around
apparently impossible obstacles and the eventual success by means of a
sudden deep insight – these are predominant features. How comes?
The latter problem may be postponed for a while. But the first, “how
science can make progress if it is so rigid and dogmatic”, should be
approached now.
For one thing, the “dogmatism” of normal science does not imply that
the exemplar (or the textbook) is regarded as sacred scripture which cannot

most of a German message to an English-speaking public.

106
Analogies and differences taken together illustrate the partial yet only partial
breakdown of communication between incommensurable paradigms. Where Lakatos
will see a research programme and look for its hard core, its positive and negative
heuristic, etc., all of which can be put on paper, the Kuhnian will look for
interpretations and collective understandings and intuitions, which can only be
described with approximation, and which together constitute a seamless whole.

88
be wrong. Firstly, the very principle of cumulative science is to use
preceding knowledge (including the exemplar) in order to succeed where
predecessors (including again the exemplar) have failed. The attitude is
nicely summed up in the statement that we are like “dwarfs perched on
the shoulders of giants [seeing] more and farther than our predecessors,
not because we have keener vision or greater height, but because we are
lifted up and borne aloft on their gigantic stature”107. Secondly, what is
learned from the exemplar is not specific sacrosanct results but a general
and open-ended way of thinking (in Aristotelian jargon: not the content
but the form of the exemplar is important); nothing prevents workers from
using this thinking to correct concrete mistakes committed in the exemplar.
But normal science is not only functional because it allows the errors
of an original accomplishment to be corrected; it ensures that the carrying
capacity of the paradigm is tried out to the full, and that it is not rejected
at the encounter of the first apparent anomaly or the first change of fashion.
It regulates and structures systematic examination of the field which it
covers (as opposed to what takes place in the pre-paradigmatic phase,
where work is unsystematic, unstructured, and largely ineffective); finally,
and for the same reasons, the paradigm is a most efficient instrument for
bringing forth and establishing the anomalies that eventually make it break
down.
The latter point is contained in an aphorism which Engels formulated
long before Kuhn: “In chemistry, only a century’s work according to the
phlogistic theory supplied the material which allowed Lavoisier to discover
in the oxygen that Priestley had produced the real counterpart of the
imaginary phlogiston substance, and thus to throw over the whole
phlogiston theory”108. Approbation of the “dogmatism” of normal science
is thus no conservative attitude, and no endorsement of static thinking;
it is connected to a view according to which scientific progress comes from

107
Bernard of Chartres (c. 1120), quoted by John of Salisbury in Metalogicon III, 4
(transl. [McGarry 1971: 167]).
108
Dialektik der Natur, MEW 20, 335f. Another “Kuhnian” point is also contained
in the passage: Priestley produced oxygen, but understood it within the framework
of the phlogiston theory as “phlogiston-free air”; only Lavoisier “invented” oxygen,
thus engendering what developed into a new paradigm.

89
that “essential tension” of which Kuhn speaks in the title of another book,
and not from gratuitous rhetoric à la Popper (cf. the quotations in note
90 and on p. 120).

Collective and individual knowledge

The question may be shifted somewhat. A theory of scientific develop-
ment is a theory about the production of collective knowledge. There is
thus nothing strange if scientific development exhibits features which are
similar to the characteristics of other types of cognition. And even here,
stability appears as a prerequisite for development and change.
Let us first have a brief look at art – postponing to a later chapter the
intricate question which (if any) kind of knowing is involved in art. A
comparison between scientists working within a paradigm and artists
belonging to the same “school” or tendency (“impressionists”, “serial
composers”, “absurd drama”) is close at hand. Even here it is obvious that
working out the possibilities of one school is one of the factors which make
innovation and even rupture possible109.
Another parallel may be followed further and more precisely. If we
replace Kuhn’s “paradigm” with “scheme”, his “expansion of the para-
digm” with “assimilation to the scheme”, and the “articulation of the
paradigm” with “accommodation”, we shall get the gross structure of
Piaget’s epistemology. Even here, as we know, the child is only able to
search for and gain knowledge because it possesses a cognitive structure
organizing the search and the transformation of sense impressions into
comprehensible experience. Only because this structure exists and is
relatively stable can it create the conditions for its own replacement by
a higher structure.
In individual cognitive development, the cognitive structure is evidently
individual, even though, e.g., the over-all character of one individual’s pre-
operatory cognitive structure is very similar to that of another individual.

109
Looking with historical hindsight at Cézanne’s or Marie Krøyer’s paintings, one
will easily see cubism working itself toward the surface; mindful listening to Richard
Strauß’s early operas or even to Schönberg’s still Late Romantic Gurre-Lieder may
make one understand why Schönberg came to feel the need for his dodecaphonic
technique.

90
But in that process of knowing collectively which is the essence of the
scientific endeavour, the cognitive structure must by necessity be shared.
At the same time, its development is not regulated and monitored by a
pre-existing language and set of concepts, a pre-established stock of relevant
everyday experience, and an already unfolded life-world (although,
evidently, all of these are there and contribute to the formation of the
scientist’s mind). The collective cognitive structure must be brought forth
by the scientific community itself – and this is precisely what is done
through the establishment of the paradigm, through the common reading
of and work from exemplars, educational “finger exercises”, etc.
It is an important feature of Piaget’s epistemology that the cognitive
structure is not constituted by conscious and explicit knowledge; it belongs
at the level of the cognitive unconscious. That this is so is an empirical fact
as good as any that can be established by psychology. If we oppose Kuhn’s
and Lakatos’s approaches, we will remember one of the important contrasts
to be that Kuhn supposes the paradigm to consist much less of formulated
theory and explicit statements than Lakatos assumes regarding his “hard
core”. According to all we know about general human cognition, Kuhn’s
view is thus more empirically plausible than the alternative; similarly, the
parallel suggests that those structures which both see as mandatory for
scientific development result to a considerable extent from socialization
and training, and not exclusively from conscious choice. On the other hand,
scientific knowledge in stored form presents itself in explicit and relatively
unambiguous statements110. Important elements of the paradigm/the hard
core must therefore consist of clearly expressible statements. Part – but
only part – of the paradigm consists of “tacit knowledge”.

110
Some ambiguity remains. Reading research papers from an unfamiliar discipline
is difficult, not only because you have not read the textbooks but also because you
have not been brought up in a way which makes you understand the hints,
connotations and implicit arguments contained in the texts. Reversely, writing out
from your own discipline may be difficult not only because you have to present
simplified versions of that textbook knowledge which your readers do not possess
but also because you have to bring your own implicit knowledge to awareness;
if you do not, you will neither be able to explain it explicitly or nor to communicate
with the readers’ implicit knowledge by means of hints, connotations and
metaphors.

91
 From the discussion of the relation between sensorimotor schemes and
operatory thought (Chapter II) it will be remembered that whole structures
(e.g., “Riding a bicycle”) may be made subservient to conscious (e.g.,
operatory) thought (“in order to get to Roskilde I shall have to get at the
train at Østerport; to Østerport I may ride on bicycle”) without being
themselves brought to awareness, i.e., without requiring conscious reflection
regarding details (e.g., the problem of balance, coordination of the feet,
...). The way in which scientific work integrates sub-functions assimilated
during professional upbringing (laboratory technique; the way a literary
scholar reads a novel) appears to be more than a vague analogue of this
aspect of general human cognition and problem-solving.
Above, we touched at the conflict between Kuhn’s description of normal
science and the participants’ own experience of the situation: even within
normal science the worker will often feel his activity to be a continuous
struggle with the material – a struggle that is only brought to success by
means of new ideas. The apparent paradox may be elucidated through
this parallel between scientific and general cognition and the displacement
of routines from the focus of awareness. According to Piaget’s epistemo-
logy, we remember, every act of knowing is at the same time assimilative
and accommodative. Every new item of knowledge which is assimilated
to a scheme alters this scheme, at least by extending its scope – cf. Aha-
Erlebnisse of the kind “My God! This thing that has puzzled me for so long
is really nothing but ...” and “Oh, that is how it is to be understood”.
Scientific processes of knowing carry the same Janus face. The historian
of sociology who looks at the first applications of functionalist explanations
within the sociology of science will tend to see the assimilative aspect of
the event: “here, the sociology of science simply expands the familiar
anthropological paradigm so as to cover a new range of phenomena –
typical normal science”. As a practising sociologist you will see things
differently. The application of the paradigm, of everything which is self-
evident, of all your tacit knowledge, will not be in focus. You may be aware
of this aspect of the matter, precisely as you know that you use your
computer and your typing skill when putting your results in writing. But
why bother about such peripheral and trivial matters if you are to tell the
crux of your endeavour. The crux is clearly the thing which was difficult,

92
that which did not go by itself, the new ideas and the reinterpretations of
familiar concepts which were necessary before a theory explaining the
functioning of religious rituals and kinship structures in tribal societies
could be used to explain patterns of scientific communication111.
Such extensions of our scientific knowledge which are only assimilative,
which require no cognitive initiative whatsoever but only trite routine, will
normally be regarded as mere applications of science (to be distinguished
from “applied science”, cf. below, n. 169) and not as genuine scientific
activity. One reason that many ideologues of science (not least Popperians)
reacted so strongly against the concept of normal science will probably
have been that Kuhn, through his emphasis on assimilation as an important
aspect of the process, appeared to equate most scientific activity with what
they see as a rather boring and unimaginative routine112.

Two kinds of “logic”

The parallel between Kuhn’s specific and Piaget’s general epistemology
may carry universal implications. The similarity between the development
of collective scientific knowledge and individual knowledge suggests the
double (assimilative/accommodative) constitution of both to correspond
to a necessary characteristic of human knowledge. Seen in this light, Kuhn’s
theory turns out to be a “logic” for the development of scientific knowledge
as a social, productive process113. What Lakatos has formulated is (in

111
Another aphorism may highlight the matter: “In normal science, 95% of
everything is routine; during a scientific revolution, only 90% is routine” (Donald
T. Campbell, at the “Symposium on Evolutionary Epistemology”, Ghent 1984).
112
Thus John Watkins, in an insipid panegyric of Popper’s genius [1974: 32]: “The
careful drawing up of a horoscope, or of an astrological calendar, fits Kuhn’s idea
of Normal Research rather nicely”. Or Popper, in an otherwise much more
interesting essay [1974: 53]: “The ‘normal’ scientist, as described by Kuhn, has been
badly taught. He has been taught in a dogmatic spirit: he is a victim of indoctrina-
tion. He [...] has become what may be called an applied scientist, in contradistinction
to what I should call a pure scientist” (Popper’s emphasis).
113
And still, in fact, a “rational reconstruction” and no faithful rendition of actual
historical processes. Some years ago, much fun was made of the fact that the word
“paradigm” did not turn up at all in Kuhn’s book [1978] about Max Planck’s first
steps toward quantum theory. Without reason, I would say: it is quite legitimate

93
agreement with his Popperian starting point) rather a logic for the
development of (stored) knowledge viewed as an abstract process, in which
the productive mediating role of the working scientific community between
one stage of stored knowledge and the next is regarded as immaterial.
At the same time (and in the same moment), Lakatos understands
“knowledge” in a way which excludes creativity and fantasy from the
domain which epistemological theory can legitimately investigate. Kuhn,
on the other hand, who regards the production of knowledge as carried
out by actual human beings, opens up the possibility that the creative and
the systematic aspects of the process of knowing may be regarded
together – even though he makes no remarkable suggestions himself in
this direction, and does not want to do so.

Objections and further meditations

The primary purpose of the present pages is neither to present a survey
of the opinions of Piaget, Popper, Kuhn and others, nor to investigate
systematically what may be the insufficiencies of their theories. It is to
present a general (though neither complete nor encyclopedic) view of the
characteristics of scientific knowledge and of the social process in which
it is established, and presentation as well as critical discussion of the
theories are subservient to this aim.
Even for the purpose of establishing a general view, however, it is
worthwhile to consider some of the problems left open by Kuhn or even
called into existence by his work.
A problem of the latter type presents itself when Kuhn’s arguments
about the efficiency of normal science are used for science policy purposes
(in a way Kuhn would never do himself, it must be stressed). It it be
mentioned first because it justifies some of the dismay called forth by
Kuhn’s Structure ....
The scientific understanding of a problem area (say, the failures of

to derive an overall structural “logic” from the shimmering of real historical

processes – but this logic should not necessarily be used to redraw in black lines
the contours of the single historical process, thus eliminating the shimmering from
view.

94
education) does not really progress as long as it stays in the pre-
paradigmatic phase where many approaches compete – thus certain policy-
makers’ (so far sensible) reading of Kuhn. If, however, this area has become
socially important or politically hot, then we’d better know something
scientifically about it in order to implement a sensible policy – thus
managerial rationality since 1945 (more about this in Chapter VII). Alas,
research about the area (educational studies) is so obviously pre- or at least
non-paradigmatic that we cannot expect genuine progress to take place
within a reasonable time horizon; then let us do something about it, and
declare one of the approaches to be the paradigm, and channel all research
monies accordingly – thus the conclusion.
This line of argument may seem attractive to bureaucrats, whether
professional officials or academic members of advisory bodies. It should
be obvious from the above, however, that the reasoning is highly fallacious:
paradigms indeed acquire their status by deserving it, by convincing workers
in the field of their efficiency; and they loose it (and should loose it,
irrespective of the preferences of grant-giving authorities) when no longer
convincing. Better perhaps: the underlying epistemology is conventionalism
and not Kuhnian dialectic – if all theoretical approaches are equally valid,
then bureaucrats can be allowed without risk to choose, and there can be
no serious objection to their choosing the one which seems most immediate-
ly promising or which agrees best with their preconceived ideas114.
Updated versions of conventionalism in Kuhnian disguise are not too
rare, but not of urgent interest in the present context115. We shall thus

114
This was already the underlying idea when conventionalism was formulated
by the Catholic philosopher of science Pierre Duhem in the early years of the
twentieth century. According to conventionalism, indeed, Cardinal Bellarmin had
been quite rigth when allowing Galileo to discuss the Copernican cosmology as
a hypothesis but not to set it forth as indubitable truth – and Galileo had proved
himself to be a mediocre philosopher when he did not understand the wisdom
in this directive.
115
To mention only one example, Barry Barnes [1979] presents Kuhn straightforward-
ly and with great sympathy as a “conservative thinker”, without noticing – if we
use the same political metaphor – that Kuhn would praise the conservatism of the
Ancien Régime for being able to produce such radical novelty as the French
Revolution, and point out that British Liberalism – the analogue of empiricism,

95
leave further discussion of this matter aside, and turn to issues deriving
from what Kuhn actually says.
Firstly to a question which has been emphasized much more in the
discussion than it really deserves, but which has the merit to suggest
further reflection: during phases of Kuhnian normal science, disciplines
are as a rule dominated by a single paradigm; Lakatosian disciplines, on
their part, are normally split between discordant research programmes.
The view appears to depend heavily on the eyes?
In any case, the view depends critically on the direction in which you
look, and the way you describe it on the sense you give to your words –
in casu of the ambiguous term “discipline”. The hard core of a research
programme, we remember, encompasses among other things a distinctive
view of reality and hence a specific demarcation of the discipline. Com-
peting research programmes within (what university administrations and
Lakatos see as) the same discipline may therefore define themselves so
divergently that it makes better sense to speak about competing disciplines
dominated each by its own paradigm116. Much depends on the question

both indeed having received their most famous philosophical formulation from
John Locke – never gave rise to such real innovations.
Genuine conservative thinking as rather the (unacknowledged) consequence
of certain recent currents: postmodernism and deconstructivism. On the surface
it may not look so: that everybody is entitled to think what he pleases since no
point of view is better than any other was never a prominent conservative stance.
But the core of these programmes is the denial of the critical potentialities of human
reason. As in the case of conventionalism (with which the postmodernist persua-
sions have much in common), the flat rejection of reason (and not just of particular
limited rationalities) leaves power and inertia as the only generally effective criteria
for decision, and lends them that legitimacy which the enlightenment project had
rejected. This is where “conservatism” creeps in.
116
Cf. the two kinds of archaeologists described in note 102. People who “would
never read each other’s papers” are not really members of the same discipline, even
though they may have their positions in the same university department.
That two disciplines compete about understanding the same section of real-life
reality does not imply that they have to be understood as two variants of the same
discipline – cf. the unending discussion between psychiatrists and psychologists
about who has the better understanding and who dispenses the correct treatment
of the same patients. As discussed below (Chapter VII), disciplines only form
through specific approaches.

96
whether disciplines are to be defined institutionally (e.g., from appur-
tenance to specific university departments) or cognitively.
But much also depend on the choice of prototype disciplines. Kuhn
tends to choose his from the physical sciences: Astronomy, physics,
chemistry. Lakatos, on his part, looks more often to softer sciences (cf. my
illustrations through economics or philosophy of science) – or to phases
in the development of physical sciences where Kuhn would see a crisis
or a pre-paradigmatic field of research.
All in all, the paradox dissolves into disagreement about concepts (a
typical case of “incommensurability”) and about the delimitation of the
“typical”. It points, however, to a much more fundamental question: How
much, and how many, fall under a specific paradigm?
As an example I shall take my own situation in 1969, at the moment
where I finished my master thesis in high energy physics. I was associated
with a subdiscipline comprising at most around 100 publishing participants
(50 may be a better estimate). Everybody within this circle followed
closely117 everything that was done under a paradigm which had been
born in 1967 and was still under articulation (it never matured before it
was superseded), and we had our own quite distinct methods and our own
argot. Evidently, other methods and techniques we shared with other high
energy physicists, under what can be described as an open-ended paradigm
ten to fifteen years old. A common trunk we shared with physicists in
general, e.g. quantum mechanics (brought to maturity between 1926 and
1935) and the theory of relativity (1905-1912).
Kuhn speaks about the paradigm. If so, how much belonged to the
paradigm under which I worked, and how many were my companions?
One may observe (Kuhn does so) that quantum mechanics when seen
as a paradigm (i.e. as something to which you are socialized, not as a body
of formulae or theory) is not the same thing in chemistry, solid state physics
and high energy physics. The way you have learned to work with quantum
mechanics differs from one field to the other. Quantum mechanics is a

117
At one moment the head of my working group, the late Koba Ziro, presented
a publication as “several weeks old”.

97
constituent of all three paradigms, but not the same quantum mechan-
ics118.
The consequence of this point of view is that the paradigm characterizes
the small unit with its 100–200 publishing participants. Normal science
phases hence become relatively short, and revolutions rather frequent –
but also rather moderate. A revolution in the small unit to which you
belong does not imply that you have to learn everything anew: those who
replaced “my” paradigm with the first version of string theory could
continue to use quantum mechanics much as they had done before.
Since even much of that knowledge and many of those professional
skills which are conserved through a revolution of the “local” paradigm
share the characteristics of a paradigm (resulting as they do from socializa-
tion, training and practice), it may be reasonable to modify the idea that
the scientists is submitted to a single paradigm, and to look at him as a
member of several circles: some broader and others narrower, some
intersecting. Each circle shares a cognitive pattern of paradigmatic character,
and single workers may well experience a revolution in their local circle
without being for that reason forced to rethink and relearn everything in
their professional practice. Such things only happen when the larger circles
are struck by a revolution: no scientifically living branch of physics (if we
stick to that example) was practiced in the same way after the maturation
of quantum physics and relativity theory as before 1900; even the structure
of subdisciplines was thoroughly revised, many specialties disappeared
and many more emerged.
Whether one wants to reserve the term “paradigm” for that which is
shared by the small unit or one accepts the notion of paradigmatic circles
or levels may be a matter of taste. Whatever our choice we should
remember, however, that the complete cognitive structure of the individual
scientist is, precisely, a structure and not exhaustively described as a
paradigm cast in a single piece (cf. the diagram on the following page,
which may seem complex but is actually utterly simplified and only meant

118
This I can confirm from personal experience: the quantum mechanics I taught
to students of chemical engineering dealt mainly with concepts and techniques
I had never heard about when studying physics. To be able to teach it adequately
I had to work hard.

98
 The embedding of a high energy physicist’s paradigm. The shaded area
corresponds to the paradigm of his small unit.

to be suggestive). We have to acknowledge that some of its constituents

are shared with everybody else who has attained fully mature operatory
thought; some–not least an analytical attitude–are shared by scientists from
other disciplines; some may be shared by members of certain disciplines
but not by the members of others (we remember how physicists and
psychologists fell into one group in a Piagetian experiment and mathemati-
cians and logicians into another, see note 30). Some elements of a physicist’s
paradigm he will share with other physicists since Galileo: I recall my own
awe when reading Galileo’s Discorsi as a young student – here spoke an
eminent physicist, however much his theories have been buried under
repeated paradigm replacement. Many, of course, he will only have in
common with contemporaries or with other members of his sub-discipline
(the Kuhnian paradigm stricto sensu).
Beyond these elements, the diagram refers to “general” and “scientific
norms and attitudes”119. As will argued below (Chapter VI), it is impos-

119
These, we may note in passing, encompass (but are not exhausted by) what makes
it possible for Foucault to see the multitude of sciences of a certain epoch as

99
sible to separate completely normative or moral attitudes from the cognitive
structure – as a matter of fact, much of the paradigm or the hard core is
of normative/quasi-moral type, prescribing what should be done and looked
for.
The composite nature of the single scientist’s cognitive structure
provides us with a scheme inside which creativity can be accounted for.
If you encounter a problem within your discipline – be it one requiring
the “5% creativity” of normal science or the “10% creativity” of a scientific
revolution – this discipline does not constitute your sole cognitive resource.
You will indeed have been trained in many practices beyond your scientific
specialty and have acquired a wide array of skills and patterns of thought
through these processes; in many cases, these provide you with models
which you may transfer to the solution of your scientific problem and use
in combination with what you know from your paradigm.
Scientists rarely tell the sources for their ideas except when these sources seem
“honourable”, i.e., scientifically relevant. As a consequence, I shall have to illustrate
this by having recourse once again to an example from my own experience:
In the early 1980s I began working on the corpus of mathematical cuneiform
“algebra” texts from the early second millennium BC. I was soon led to a complete
reinterpretation of almost every term and technique. If more than a dozen of people
around the world had been active in the field, this could have been characterized
as a (local) revolution; in the wider contexts of the history of mathematics or
assyriology it looks more like the assimilation of the field to an anti-anachronistic
paradigm which looks at Greek geometry or Babylonian laws not as incomplete
forerunners of our thinking but as expressions of the culture within which they
were created.
But neither assyriology nor the anti-anachronistic ideal provided me with the
tools which allowed me to understand the Babylonian texts. One aspect of my
method was instead structural semantics; that, however, I did not know at the time,
and my improvised method was actually inspired from my supervision of student
projects analyzing literature structurally (even this I did not think about in the
process, but it could be seen with hindsight). Another aspect was hermeneutic
reading – yet not taken over from what I knew about hermeneutics but an
application of the way I had once used to read wrong answers to mathematical
exercises closely in order to discover the underlying reasoning and thus be able
to make their errors pedagogically fruitful. At least one visualization of a procedure

representatives of a single episteme – cf. note 104.

100
I might have taken from many places, but I happened to borrow it from my half-
forgotten particle physics via a different use I had made of it in an analysis of
ancient Egyptian mathematics.
However frivolous these inspirations may seem, they can all be traced
unambiguously to various kinds of professional experience of paradigmatic
character. The anti-anachronistic drive, on its part, did not come from any allegiance
to a historicist programme, but rather (as far as I can see – but at such points
introspection becomes suspect, as any psychoanalyst can tell) from deep-rooted
personal norms and attitudes also reflected in the way I once read my students’
mathematics exercises.
Innovation and creativity cannot be reduced to mere heaping of such accidental
extra-disciplinary inspirations, and extra-disciplinary inspirations change when
they are brought together and applied in a new context (my hermeneutic readings
of Babylonian texts were certainly more analytical than those I made intuitively
as a mathematics teacher). But the anecdote will hopefully show that the fine
structure of the paradigm provides us with a framework within which the creative
process can be discussed and thus, to some extent, understood..

V. TRUTH, CAUSALITY AND OBJECTIVITY

From the making of scientific knowledge we shall now return to three

classical issues: the problem of truth or correspondence between theoretical
statements and reality (where we shall have to connect a number of
discussions and reflections from the previous chapters); the nature of
causality; and the question of objectivity versus subjectivity.

101
Truth
As all empiricists, the logical empiricists worked from the implicit
premiss that the correspondence between the single observed fact and the
statement of that fact was straightforward120. For them, as for the whole
philosophical current descending from them, the “theory of truth” is
concerned with how truth values can be ascribed consistently to the
sentences of a formal language (cf. note 48), and not with the question of
“agreement with facts”.
The position of Popper0 is similar: the difference between the two
approaches hinges not on any disagreement concerning the “naïve”
correspondence theory concerning elementary facts but on the explanation
of the way from single statements of facts to theory. Popper1, as well as
the later phases of logical empiricism, came to admit that “naïve” corre-
spondence does not hold water, but none of them succeeded in creating
a credible “critical” substitute. The consequence drawn by the logical
empiricists instead verged toward scepticism, whereas Popper’s later
writings lean toward an unacknowledged instrumentalism (cf. note 53).
In classical philosophical terms, the consequences drawn by Kuhn from
his view of scientific development would also have to be characterized as
sceptical. He argues strongly that scientific progress is real, but in the sense
that “later scientific theories are better than earlier ones for solving puzzles
in the often quite different environments to which they are applied” [1970:
206]. But scientific theories do not come closer and closer to truth through
successive revolutions, as common sense would have it. Scientific progress,

120
That any philosopher would do so after Kant may seem astonishing: whatever
one thinks of Kant’s solution, he should recognize the existence of a problem. If
we can only know by imposing our categories on the world, then simple correspon-
dence appears to be excluded.
Yet the explanation need not be neglect of Kant’s insight (even though the
logical empiricists wanted to render earlier philosophy superfluous rather than
continuing it). It could also be that they took Kant fully to the letter: categories
which we cannot help applying in an invariable form which we cannot influence
may be counted as a part of that reality which it is the aim of science to describe –
“observed facts” are precisely observed facts. Such a solution had already been
proposed by d’Alembert in the preface to the Encyclopédie, decades before Kant
formulated the problem (see [Høyrup 1993: 213]).

102
like Darwinian evolution, is instead to be understood as a steady movement
“from primitive beginnings but toward no goal” [1970: 172].
At closer inspection, the truth which Kuhn cannot discover as the goal
toward which scientific development moves turns out to be an ontology,
a “match, that is, between the entities with which the theory populates
nature and what is ‘really there’” [1970: 206]. So far Kuhn is indubitably
right. Newton introduced forces into nature, and Einstein’s General Theory
of Relativity abolished them again; if Newton’s move was one toward
greater ontological truth, and forces are “really there”, then Einstein’s move
was in the wrong direction. If Lavoisier’s explanation of combustion as
absorption of oxygen constituted ontological progress, then the acceptance
of the phlogiston paradigm had been a mere error, irrespective of its actual
successes (including its role in Priestley’s production of oxygen).
In the case of competing paradigms approaching the same subject-
matter in different ways we are no better off. Word classes, e.g., may be
defined from morphology and meaning or from syntactical function. If we
choose to interpret, e.g., the form gone in “Peter is gone”, as a conjugated
form of the verb go, and if we follow this principle throughout, we get one
type of insight into the structure of language; if we choose to interpret it
as an adjective because it is parallel to clever in “Peter is clever”, we get
other insights121. If one of the two approaches is ontologically correct,
then the insights gained from the alternative approach are spurious and
not to be relied upon.
Certainly, situations exist where one approach turns out to be mistaken
and the other not (or not clearly) so. In other cases, however, later
developments show that none of the two was quite mistaken. Above (p. 73),
the interpretation of the phlogiston theory in terms of “degrees of
oxidation” was mentioned. In linguistics, we may refer to the relation
between the neogrammarian theory of language development (which
referred to laws of phonetic change) and Saussure’s early structuralist
description of language as it looks at one particular moment. In Saussure’s

121
To be sure, the example is simplified into the extreme – but not significantly
more than the references to planetary systems and combustion theories. See
[Diderichsen 1971: 20ff] for an exposition of the problem in the context of a
descriptive grammar.

103
view [1972: 193-195], the two approaches are incompatible; when discussing
development he works within the neogrammarian paradigm (when needed
correcting specific laws – thus [1972: 201]). A couple of decades later,
however, structural linguists of a new generation (in particular Roman
Jakobson) reinterpreted the sound shift laws as resulting from structural
constraints. In both (and many similar) cases we may conclude that the
later integration of the two approaches into a more mature theory has
shown both of them to be in some way true – which requires that we
formulate a concept of truth where this can be said meaningfully.
As a matter of fact this concept was already formulated on p. 46, where
the truth value of a theory was interpreted as a “structural agreement or
correspondence with features of reality”; on the same occasion it was
observed that “correspondence is something quite different from similarity,
not to speak of identity”.
This materialist notion of truth is not ontological – or at least not
necessarily ontological. Certainly, if we suppose that a certain entity does
exist (be it with qualifications, like the fox on p. 30), then the best theories
will be those which contain it. In this case we may speak of “ontological
existence”.
We may surmise the element oxygen to be such an entity. During the
two centuries that have passed since its discovery, we have come to know
increasingly more about oxygen: that it consists of atoms composed in a
specific way; that three different isotopes of oxygen exist; that it has a
particular spectrum; etc. – and even how oxygen can change into other
elements through nuclear processes. We have come to know answers about
oxygen to which neither Priestley nor Lavoisier suspected the questions;
we have not, however, succeeded in getting rid of oxygen as a particular
way in which matter can be organized.
Phlogiston had no similar ontological existence, however real the
regularities of chemical processes which were explained through the
assumed existence of this substance. When taken to be a substance it turned
out to resemble the elf maiden: one sees her, one hears her enticing
proposals; but when he tries to grasp her, her back is hollow, and his arm
catches nothing but thin air. Newtonian forces are similar, they only
manifest themselves through the regularities which they formalize, and

104
one can get into no other contact with them. Word classes may fall
somewhere in between: a delimitation made on the basis of declination
turns out to have semantic and syntactical implications (if only approxi-
mate). Social classes defined from income sources as done by Adam Smith
may also be located at an intermediate position, since individuals who
belong to the same class are likely also to share much of their general
experience and their material and spiritual culture. The labour value of
goods (or the customer preferences of the competing marginalist economic
theories) or the Gothic style in architecture, on the other hand, are probably
no be better off than the Newtonian forces; few of the basic entities to
which social and human sciences refer seem to be. If we expect truth to
be ontological truth we do not need Kuhn to tell us that few of the theories
about human social and cultural life are true.
But why should truth be ontological? No sensible scholar would claim
that the most adequate (or “true”) description of Shakespeare’s works is
a dictionary listing his vocabulary, and no biologist would be satisfied by
taxonomy alone. The “truth” of the screwdriver did not consist in the
isolated correspondence between its edge and the notch of the screw, but
in that combination of edge and rotational symmetry which corresponded
to the entire make-up of the screw and allowed us practical interaction
with it (namely to put it into the wall). Taken in isolation, the “ontological”
truth of the edge is even meaningless, since only the practical interaction
turns the edge into something which corresponds to a notch – screws have
no edges, they are constituted (and used) in a way that makes the
application of an edge appropriate. The truth of a theory, we may repeat,
consists in its “structural agreement or correspondence with features of reality”
as revealed in practical interaction with the object (be it in interpretation,
cf. note 50)122.

122
Experimental science, it is true, may aim at controlling whether a certain
theoretical entity can be interacted with in new ways, so to speak testing whether
the elf maid whom you see and hear and whose hair you smell can also impress
the sense of touch. It was precisely because they could be contacted through several
channels (semantics, declination, syntax, or economy, living conditions and culture)
that word classes and social classes were held above to be more ontologically real
than phlogiston. But since physical science provides the scale on which degrees
of real existence is conventionally measured it is worth remembering that quantum

105
 Then, however, the basic categories which form the framework of our
cognition become true – cf. the discussion of the practice allowed by the
category of the permanent object on p. 30. The progress within a paradigm
also becomes a progress toward greater truth, not a mere accumulation
of solved puzzles: expansion of the paradigm means that more features
of reality are accounted for coherently, whereas articulation implies greater
precision in the structural agreement. Even the replacement of one
paradigm by another is a progress toward greater truth, at least if the
replacement follows the Kuhnian pattern expansion —> accumulation of
anomalies —> crisis —> convincing new interpretation achieving paradigmatic
status123 – cf. the discussion (in terms of “research programmes”) on p. 73.
In formal as well as Aristotelian logic, a meaningful statement is either true
or false; the everyday idiom of “not quite true” and “almost true” has no place here.
Of two conflicting statements, furthermore, at most one can be true. Neither seems
to apply any longer if we allow (for instance) both the phlogiston and the oxygen
theory to be materially true, however much one is held to be “more true” than
the other. As long as we move within the same conceptual framework (within which
the problem of non-commensurability does not present itself), it is no more difficult
to speak of the sentence which is less true than the other as false than it is to say
in traditional logic that the statement “Peter is a boy, John is a boy, and Joan is
a girl” is true while the statement “Peter is a boy, John is a boy, and Joan is a boy”
is false and not just “67% true”. If the two statements belong within incommen-
surable conceptual frameworks things are less simple – even if we regard the early
oxygen theory as more true than the phlogiston theory, the statement “carbon and
hydrogen have nothing in common beyond being elements” cannot be declared
“so true” that the phlogiston theory identification of the two is completely false;

mechanics (though ridden with paradoxes as it still is when used to describe more
than isolated experiments) dissolves the ontological existence of physical matter:
if an electron has to be something specific, it is a particle in some experimental
settings and a wave in others – maiden to the eyes and thin air to the sense of
touch. A consistent description (relative as this consistency is) can only be reached
at the cost of giving up its ontological separateness – and thus ultimately by
relativizing the validity of the category of the (conserved, separate) object more
radically than we were forced to do by the breathing and perspiring fox.
123
The situation is different if an ideologically inconvenient paradigm is replaced
by something more convenient – as was the case in the neo-classical revolution
in economics, cf. note 77 and p. 74. In such cases, progress toward greater truth
is evidently not assured. Nor is, however, the Kuhnian progress “from primitive
beginnings” – Jevons is certainly more primitive than Ricardo.

106
only when a common framework (in this case the developed oxygen theory) ripens
is it possible to decide – in the present case that the phlogiston theory had a point.
In the end, the Lakatosian-Kuhnian criticism (or even rejection) of the
naïve correspondence theory of truth thus unfolds (in materialist interpreta-
tion) as a genuine critique. While sceptical postures in the manner of Pilate
(“What is truth” – John 18:38) are often meant as a mere way to wash one’s
hand (cf. Matt. 27:24), the critique tells us that truth, though never final
nor absolute, is not arbitrary nor to be decided from fancy124. Not every
point of view is as good as any other.

Causality
The concept of causality goes back to two types of immediate or daily-
life experience. One is acquired already in the sensori-motor period, in the
form of that “practical category of causation” which makes you draw a
table cloth in order to get hold of a flower vase which your mother has
tried to put outside your range (cf. p. 14). When your cognitive structure
develops, the category enters awareness. In this mature form it is the
foundation of any planned action aiming toward a specified end, and thus
of every technology – physical, chemical, medical, psychological, or social.
Action according to this scheme falls into separate phases: first you
conceive a strategy; then you start drawing the table cloth; after a short
while, the vase falls over the edge of the table onto the floor, and you get
hold of it (maybe in some unforeseen state, but this concrete problem is
common to all strategic planning and not our present concern).
The other kind of proto-causal immediate experience only enters our
life with pre-operatory thought: it is the question “why?”. To the question
“Why is the floor wet and covered with broken glass?” you may answer

124
Francis Bacon, in the opening passage of his essay “On Truth”, also drew on
St. John for a comment on those who claim truth to be a mere whim:
What is Truth; said jesting Pilate; And would not stay for an Answer. Certainly
there be, that delight in Giddinesse; And count it a Bondage, to fix a Beleefe;
Affecting Free-will in Thinking, as well as in Acting. And though the sects of
Philosophers of that Kinde be gone, yet there remaine certain discoursing Wits,
which are of the same veines, though there be not so much Bloud in them,
as was in those of the Ancients.
[Bacon 1937: 5].

107
“because I wanted to get hold of the flowers”; alternatively you may tell
that the vase fell over the edge of the table (and your brother may sneak
on you and tell who was responsible for pulling it). The latter explanations
are what we are used to call causal, dealing with various aspects of the
process behind the broken glass; the former is teleological, an explanation
of the purpose which you had in mind when pulling.
The sciences also ask and answer why’s; in general we feel that
theoretical sciences are characterized precisely by posing and trying to settle
such questions for their own sake, whereas applied sciences translate
practical aims into questions and translate the answers back into strategies.
In the humanities (“theoretical” as well as “applied”), the following
question types will be familiar:
— Why will so many people use time and money on reading Bild
Zeitung?
— Why will so few people read Thomas Mann?
— Why did jazz develop among the Black in New Orleans?
— Why is the normal form of the Oedipus complex absent from the
Melanesians of the Trobriand Islands?
— What made the Roman Empire collapse?
— What were the reasons that made research a central activity for the
nineteenth century university?
All six questions may be understood causally; the first two (which ask
about the actions of people) may also be given teleological answers (e.g.,
“because the majority sees no point in playing high brow”, or “because
Bild Zeitung arouses one’s feeling of being alive, while Thomas Mann’s
prose is so complex that you fall asleep”).
In sciences which do not deal with the conscious decisions of human
actors, only causal answers to the question “why” are normally accepted
nowadays – biologists do not believe that the giraffe got its long neck “in
order to” be able to eat leaves from trees; instead, what eighteenth century
theologians would see as examples of God’s design is explained as the
adaptation to a specific ecological niche through Darwinian selection
pressure.
The different meanings of the “why” has been used by some philoso-
phers to delimit the humanities – in particular by the Dilthey school. Causal

108
answers to the above questions may be quite legitimate, according to this
view; but the sciences which provide them (the sociology of literature and
art, psychoanalytically oriented anthropology, economic history, etc.) belong
outside the humanities (more precisely: the Geisteswissenschaften). Explana-
tion is causal; understanding, the purpose of the humanities, may well
investigate how reading Bild Zeitung or Thomas Mann affects one’s cultural
status. But this knowledge is only relevant inasmuch as it is also known
by the potential readers – what we know is only relevant for understanding
the actions and opinions of people who also know it.
Others have claimed that the only way the humanities can pretend to
the status of sciences is if they allow causal explanations. Explaining reading
habits through the delusive motives people give for their actions (or, still
worse, by inventing motives and claiming that these are the motives of
the actors) is no better than referring to the keen desire of the giraffe for
green leaves.
Often this latter stance is coupled to a professed positivist view and
to the claim that causation has to be understood according to Hume’s
definition and not through the multiple causation proposed by Aristotle.
Since the positivist understanding of scientific knowledge is not without
problems, and in view of the importance of the why’s for every scientific
practice, an investigation of the characteristics of scientific knowledge will
have to probe this claim.
Hume’s view is set forth in his Enquiries Concerning Human Understand-
ing. Causation is no necessary connection between one event (“the cause”)
and another (“the effect”) (section VII,ii,59, ed. [Selby-Bigge 1975: 74f]).
It is nothing but an expectation on the part of the observer produced by habit.
When we have seen innumerable times that a billiard ball starts rolling
when hit by another, then we expect that billiard ball A, when hit another
time by billiard ball B, will start rolling as usual. This, and nothing more,
is meant when we say that being hit by B causes A to start roll. Moreover,
thus Hume, the concept of causation requires that the effect comes after
the cause.
To this a modern Aristotelian will object that there are many answers
to the question why A moves as it does. Being hit is evidently one; but
if the balls had consisted of soft clay the outcome would have been

109
different; so it would if A and B had not been spherical, or if A had been
located at the very edge of a table not provided with a cushion. A complete
answer to the question why will thus involve efficient causes (the hitting);
material causes (ivory, not clay; the surface of the cloth); and formal causes
(the laws of semi-elastic impact and of sliding/rolling, as well as the
geometrical forms involved). If we want to understand what goes on we
will also have to notice that somebody plays billiards and wants B to move
(perhaps as it does, perhaps otherwise), ultimately wishing to win the game
and to gain the stake; both of these are final causes125.
The modern “positivist” view of causation may presuppose the naïve
correspondence theory of truth for the observation of single events and
hence replace Hume’s habit of mind by a regularity of nature; or it may
accept the criticism of naïve correspondence and stick to the subjective
expectation. In both cases it identifies the cause as an efficient cause, an event
which is invariably followed by another event. It is generally held that this
kind of causal thinking is the one about which the physical sciences speak,
and therefore the one which should be emulated by social and human
sciences126.
The premiss is blatantly wrong (and the conclusion thus no conclusion
but an assertion which must be assessed independently). Firstly, a physical
description (in case, by Newtonian mechanics) of the billiard game will
involve all the aspects listed under the Aristotelian explanation: physical
configuration and shape, masses, friction, elasticity. The events “hitting”
and “starting to roll” are at best aspects of moments of a complex process
without any privileged status. Actually they are even less: they do not exist
as events. When B touches A both will be compressed, and increasing

125
The Medieval scholastic tradition and later anti-Aristotelianism have spoken in
the singular of the efficient, the material, the formal and the final cause. Yet according
to Aristotle’s point of view, “the modes of causation are many”, even though they
can be grouped in classes according to their character (Physica 195a28; trans. [Hardie
& Gaye 1930]).
126
Many of those who do not accept the relevance of Humean causality in social
and human science, on their part, claim that it was the one which prevailed in
Newtonian physics, and that it has been left behind by modern physics. Why –
they ask – should social and human sciences imitate a model which has shown
itself to be erroneous in the physical sciences?

110
compression will be accompanied (not followed in time) by increasing
mutual repulsion. This repulsion will accelerate A and decelerate B, and
after a short while A runs faster than B. After another short while the two
balls separate, and we see A first sliding and then rolling along alone.
In this description of the physical process, as we see, there is no event
“hitting” preceding another event “starting to move”. Both are processes,
and indeed the same process. In the idealization where both balls are
absolutely hard the processes contract to a momentary event, it is true;
but then to the same moment. This is nothing specific for billiard balls; every
description of classical physics has the same property. So has also every
description according to relativistic physics and quantum theory. The
reason that Newtonian forces have to be given up is precisely the unaccept-
ability of delayed causation. Hence relativistic physics will not speak of
one electric charge q1 acting at another q2, which it could only do with a
delay corresponding to the velocity of light; instead, q1 is told to produce
a local electromagnetic field; this fields propagates, and acts on q2.
Hume was indeed quite right when connecting his “events” to mental
habit: his “hitting” and “rolling” are not moments of the process described
by physical science; they are moments of the awareness of the observer.
He sees and hears B approaching and hitting A, and afterwards he notices
that A has started rolling.
Hume’s scheme thus does not lay bare the underlying structure of the
physical description, and is certainly no critique of Newtonian causa-
tion127. It is a formalization of the sensori-motor scheme of practical
causation and of strategic action in general: First I get the idea of how to
get hold of the vase, and start drawing the table cloth; afterwards, the vase
falls to the floor. Here as in the Humean explanation, everything in the
configuration is taken for granted: the table, the table cloth on which the
vase stands, etc. Formally, of course, Hume has left aside the anthropomor-
phic notion of decision and planning. Fundamentally, however, the
emphasis on temporal separation shows that the “effect” is an end result
and thus an aim – no step in an ongoing process is ever the end unless you

127
That Hume attempted such a critique but did not produce one was already
pointed by Kant in the Critik der reinen Vernunft [B19; in Werke II, 59].

111
define it to be because you are not interested in what comes afterwards.
Similarly, the privilege of the hitting over the other aspects of the process
is that it corresponds to an action which you may undertake intentionally. The
table and the elasticity of the balls are given; as a player you push.
Yet Humean causality, even if a formalization of strategic action, is not
relevant for technological thinking. What I need to know (intuitively, in
this case) in order to win the billiard game is the degree of elasticity of
the impact, the way friction transforms sliding into rolling movement, etc.
The knowledge which serves technological planning is indeed “Newtonian”
or “Aristotelian”, not “Humean”: what you do is to interfere with or
determine specific features of a process (acting as an “efficient cause”) –
yet you can only determine the ultimate outcome if you understand how
the features on which you act interact with and are conditioned by other
(formal, material, and structural) features.
Humean causality is hence neither relevant for theoretical natural
science nor for technological thinking. Remains the question whether it
can be given any meaning within the humanities.
The answer is easily seen to be negative. We cannot notice it to be a
regularity that all Roman Empires collapse, since there was only one. Nor
can we test the hypothesis that relatively few people read Thomas Mann
because he is difficult by seeing if more people will read authors who are
in all respects like Thomas Mann except that they are more readable. As
everybody knows who has enjoyed Joseph und seine Brüder, an author whose
prose is easily read would be different from Mann on almost all accounts.
What one can do is to look in general at The Collapse of Complex Societies
(the title of a book published a few years ago [Tainter 1988]), or to
investigate the public of a variety of authors, thereby finding similarities
and divergences in possible “cause” and possible “effect”. This is often
believed to be the closest one can get to (Humean) causality in the
humanities.
It may well be the closest on can get – but still it has nothing at all to
do with the Humean concept. In the moment we single out a specific class
of societies as “complex” or single out an array of features by which
authors can be characterized, we have already introduced a screen of
theoretical or at least pre-theoretical thinking between the events and the

112
way in which we interpret them – in a more radical sense than that in
which even simple observations are by necessity tainted by theoretical
presuppositions.
The only kind of causality which is meaningful within the humanities
(and the social sciences, for that matter) is the one which also makes sense
in the physical sciences: correlation with, or explanation in terms of a theoretical
framework – though evidently a theoretical framework which is not the one
of the physical sciences, and more often an open framework than a finite
and formalized set of statements.128 A less shocking formulation of the
claim is that causality is never an extra component of the scientific explanation
of phenomena beyond theory, neither in the natural nor in the social or human
sciences. It is (at best) a way to formulate the theoretical explanation which singles
out one element as the most interesting and considers other elements and
features as a background with which the element in focus interacts.
That causality tells us nothing beyond theory may be a dismal
conclusion for those who accept that theories are fallible and therefore want
to base their science or their technology on more firm foundations. As we
have seen, however, this aim is not attained by attempting to adopt a
Humean idea of causality, and giving up illusions is no real loss. Moreover,
as we have also seen, the inescapable “fallibility” of theories only means
that no theory can pose as absolute truth. Inasmuch as a theory can be
claimed to be true in a materialist sense, as argued above, causal explana-
tions in the sense of correlation of phenomena with theory are also true.
When one element of a theoretical explanation is singled out as a cause
it may be so because we want to use our knowledge in some kind of
technology: that which it is of interest to consider as a cause is what we
can influence or determine, that which is technologically manipulable; those
features of the situation which we can do nothing about are then under-
stood as a background, the conditions to which our action is subject. But
even purely theoretical investigations may speak of and discuss causation.

128
The historian Fernand Braudel, who was interested in global and long-lived
structures but knew of course about the effects of particular human actions (dynastic
marriages, wars, etc.), tried to specify a less open framework by distinguishing
long-, medium- and short-term determination in history – cf. [Kinser 1981]. What
he achieved, however, was conceptual clarification and no formalizable framework.

113
In such cases, the choice of the cause is obviously rather free, and may
depend on which features of the situation we find most interesting, or
which aspects of the process we want to scrutinize more closely while
relegating others to the background.
One of the more grotesque (or, if you are in that mood, tragic) facets
of scientific life is the jealous and hostile passion with which scholars
discuss in such situations which is the cause. Was the tuberculosis which
killed several of my grand-uncles caused by bacteria or by their living
conditions? These competing causalities are evidently complementary
aspects of the same theoretical explanation: the disease cannot occur unless
one is infected; but infection is much more likely to result in disease when
he is badly fed and his lodging never dry. This example may be outdated
except when one or the other sanitary policy is advertized (i.e., when it
comes to technological choices), precisely because a single theory now
explains how the two explanations concur. If the two derive from different
theoretical approaches (which, like the early structural and the neo-
grammarian approach to linguistics may both be “parts of the truth”, even
though their mutual relation is not, or not yet, elucidated), similar
discussions are still seriously meant. In the humanities, where theories
stricto sensu are rare and open-ended frameworks the rule, discussion may
easily degenerate into calumny or non-discussion because both parts regard
the other as mistaken beyond hope of salvation by argument129.
Sticking stubbornly to your own paradigm may well be an efficient
way to make the sciences progress, as held by Kuhn: as an advocate in
court, you should do what can be done to find arguments in favour of your
client. But like the advocate you should also listen to the other part, and
acknowledge that the final verdict may not be totally in your favour.
Dismissing that possibility, and denying a priori that approaches which
differ from one’s own may be legitimate, is no more productive in science

129
This was indeed the situation in my own discipline, the history of science, until
a few decades ago. As late as the early 1980s the discussion between “externalism”
(the explanation of scientific development as “caused” by general cultural, social,
technological and economic factors) and “internalism” (causation by the inner
dynamics of the sciences) was still regarded by most members of the field as
meaningful.

114
than in court, and no less preposterous.

Objectivity, subjectivity, and particularism

If one’s choice of the element to be singled out as a cause is “rather
free”, does that mean that causal descriptions are subjective and not
objective? In general, is scientific knowledge subjective or objective? This
question is more trendy than the question of truth, maybe because it is
easier to assert that the other part in a discussion is subjective than to prove
that he is wrong.
Is science objective? Certainly not if we understand objectivity as
coincidence with the object. Nobody can carry an objective picture of a house
in the brain or the mind, since the house is larger than the skull and the
mind contains no bricks. Or, as formulated by a student of mine when I
objected to an explanation he had given at an examination that it was
incomplete130: Any complete model of reality is – by definition – reality.
Criticizing a scientific description for failing objectivity in this sense
is thus either absurd or foul game. Science, like any other knowledge, is
by necessity subjective. Even the way a house appears on a photograph
depends on the optics of the lens and on whether the film is black-and-
white or in colours. What science sees also depends on the conceptual
structure through which it looks: Priestley made oxygen but saw
phlogiston-free air.
But science, as knowledge in general, is also subjective in a sense which
corresponds to the dependency of the picture on the place from where it
was taken and the dependency of an answer upon the question. Knowledge
never comes from passive reception of whatever passes before one’s
indifferent eyes; it is always the result of an active practice – as a minimum,
of selective attention, but often of much more intentional operation. In
science, knowledge comes from experiment, investigation, critical reflection,
etc. As our practice vis-à-vis the reality we want to know about is not in
itself part of that reality, whatever knowledge we earn is by necessity
subjective and not determined by the object alone.

130
Mark Madden, Spring 1990.

115
 This much is common sense. It should have been evident at least since
Kant, and does not illuminate the relation between objectivity and
subjectivity very much. An interesting quality of the Piagetian and Kuhnian
epistemology is that it allows us to discern a sense in which the subjective
aspect of knowing may in itself be more or less objective.
In order to see how we shall first return to the discussion of the status
of the schemes and categories of our general cognitive structure (see
p. 28ff). When we conclude, for instance, that “the material world is consti-
tuted in a way that allows an adequate practice if we order our experience
as representations of permanent objects”, then we have concluded
something about the material world – namely about its response to a
particular practice. This predictable response is a property of the material
reality in which we live, i.e., of the object of our knowing; the category of
the permanent object, however much it is a constituent of our subjective
cognitive equipment, is also objective.
The basic categories and logical schemes, however, though arising
through cognitive development, are end points. Once my daughter had
acquired the scheme according to which there has to be more children than
girls in a mixed kindergarten (see note 22) she could understand nothing
but that, and once we have organized our way to experience in permanent
objects, we cannot avoid doing so. It is left to further empirical studies to
find the degree of permanency of actual objects, and to construct for
instance a theory which allows us to discuss in which sense salt is
conserved when dissolved in water. Irrespective of their genesis through
biological evolution and individual development (which make them
synthetic, i.e., informative about reality), the basic categories and schemes
remain synthetic a priori with regard to the ongoing process of knowing;
they thus belong to another cognitive species than the synthetic a posteriori,
the actual outcome of our observations. Once the synthetic a priori have
arisen, it is no longer possible to distinguish in their domain between more
and less objective subjectivity.
This distinction is only pertinent when it comes to discussing the
synthetic a posteriori, in particular scientific knowledge. To see this we
may compare the Kuhnian cycle with the hermeneutic circle – I quote the
explanation given in [Høyrup 1993: 172]:

116
 At our first approach to a foreign text (in the wide sense, i.e., to any spoken,
written, sculptured, painted or similar expression of meaning) we interpret
it in agreement with our own presuppositions and prejudices, which are in
fact our only access to the universe of meanings. But if the foreign text does
not fit our interpretive expectations on all points (which it rarely does), and
if we investigate the points of non-compatibility seriously, we will be led to
revise our prejudices. The revision will enable us to understand the presupposi-
tions of the foreign mind (or understand them better) and hence even to
understand ourselves from the foreign point of view. Understanding the other
leads us to better insight into our own expectations to universes of meaning,
and hence allows us to approach the foreign text (or other texts) with better
prejudices.
Some features of this structure are certainly different from what we see
in the Kuhnian cycle of normal and revolutionary phases. The relation
between the scientist and the object of a science is less symmetric than that
between the interpreting mind and the interpreted mind131. The “better
prejudices” with which we approach the object after a change of paradigm
do not come from understanding ourselves from the point of view of the
object. But they remain objectively better, i.e., they reflect the features of the
object more precisely or at a deeper level, according to the arguments on
pp. 73 and 106. The progress of a scientific discipline through scientific
revolutions (when these follow the Kuhnian ideal scheme), which was
spoken of above as progress toward greater truth, can also be understood
as progress toward greater objectivity of its subjectivity.
So far so good. Yet “subjectivity” is not only used about the inescapabi-
lity of knowing only in response to specific questions and in terms of a

131
Anthony Giddens [1976: 146ff], when discussing the similarities between the two
circles, characterizes the process of interpreting a foreign text or social world as
a double hermeneutic:
Sociology, unlike natural science, stands in a subject-subject relation to its ‘field
of study’, not a subject-object relation; it deals with a pre-interpreted world,
in which the meanings developed by active subjects actually enter into the
actual constitution or production of that world; the construction of social theory
thus involves a double hermeneutic that has no parallel elsewhere [...].
What this hermeneutics looks for is thus an understanding of (e.g.) Rousseau’s Émile
and its impact which builds on how Rousseau and his contemporaries understood the(ir)
world.

117
particular conceptual framework. These two kinds of subjectivity were
assimilated (in reverse order) to the making of a photograph on a particular
film by means of a camera with a particular lens and from a specific
perspective. Given these conditions (together with the shutter speed, etc.),
what will appear on the picture is determined by the object.
Or it should at least be. If it is not, the picture has been retouched.
Retouching corresponds to those kinds of scientific knowing which are
affected by other factors than the object, the kind of question (purportedly)
asked, the instruments used, and the conceptual framework which is
referred to or which is shared by the community of workers in the field132.
This kind of subjectivity is better spoken of as particularism. To some degree
it cannot be avoided – you always have motives beyond those which you
confess to yourself or reveal to your psychoanalyst, even when it comes
to knowing. But since (by definition) private distortions cannot be
controlled by others, they detract from the value of knowledge to the same
extent as they are present. Scientific knowledge is shared, or should at least
be shareable; but knowledge expressed in a code which cannot be
deciphered by others can not be shared. Particularism, though to some
degree inescapable, should hence be minimized through critical discussion.

132
It may be a profitable aside to point out that the seemingly innocuous phrase
“which is referred to or which is shared” hides a serious dilemma. Many social
scientists, noticing that social thought is always involved in conflict and cannot
avoid taking sides at least by asking the questions of one party, have held that this
inescapable subjectivity should be brought under control by each social scientist
telling his side, his employer and his sympathies explicitly and honestly. This stance
(which is in particular associated with Gunnar Myrdal) may seem attractive, and
certainly has a point. But apart from the naïveté of the expectation that people (and
social scientists are also people) should tell honestly when their sympathies and
aims contradict those of their employer, Myrdal’s cure against unbridled subjectivity
suffers from the same weakness as Lakatos’s formalized conception of his “hard
core” (cf. p. 91): too much of the framework that is shared by a scientific community
consists of tacit knowledge, and will not be revealed if one asks workers for their
political position – the paradigm under which they have been trained may well
be built around questions and concepts asked and formulated from a quite different
position.
Participating in a community which shares a paradigm, including the tacit
knowledge which goes with it, is therefore just as important as honesty if
subjectivities are to be kept as objective as possible.

118
 VI. THE ROLE OF NORMS

In 1965, a symposium was held in Oxford, at which among others

Popper, Lakatos and Kuhn were present, and the topic of which was a
discussion of Kuhn’s Structure of Scientific Revolutions133. As Popper and
Lakatos saw things, the question was whether Kuhn was right in replacing
the(ir) logic of research with a description of the social psychology of
scientists; Kuhn, on the other hand, asked whether Popper and Lakatos
had really brought forth a logic, and answered that their actual output was
an ideology.
Much may speak in favour of Kuhn’s reply. Still, both Popper1 (and
even Popper0) and Lakatos/Popper2 are too close to aspects of the scientific
production process to be dismissed as nothing but ideologues in the vulgar
sense. Also Kuhn’s own work, as we have seen above (p. 93), results in
a kind of logic for the social production of scientific knowledge. Even
though there may be many similarities between science and organized crime
(to paraphrase a discussion running through the same conference proceed-
ings; both are indeed social activities perpetrated by relatively closed and
highly specialized communities), there are also noteworthy differences.
We can therefore only come to understand the nature of scientific know-
ledge if (so to speak) we grasp how the process of science, in spite of its
similarities with organized crime, can be described approximately in terms
of a “logic” through which it manages to produce some kind of reliable
knowledge.

133
The revised contributions from the symposium were published in [Lakatos &
Musgrave 1974]. Many of them were cited above.

119
 Traditional hagiography explained the specific character of science by
the exemplary character of scientists. Science is reliable because scientists
are eminently reliable; science is objective because scientists are heroically
objective and dedicated to their cause; etc.
Bankers will probably not agree that scientists are significantly more
reliable than average people when it comes to their use of a cheque account.
If they are within their professional work, it must be explained in other terms
than through general moral perfection. This will lead us into an investiga-
tion of the relation between scientific practice and norms – and in general,
between knowledge and morality. As it will turn out, the analysis will
provide us with important insights, both as far as the nature of morality
is concerned and regarding the scientific process. Since the discussion has
to move back and forth between these levels, it may appear rather intricate.

Logic and norms

A “logic” is a fixed pattern. If the development of scientific knowledge
in social process follows a specific logic (or just follows it to some extent),
we must presume that this social process is itself governed by a set of
general “laws” or regularities.
Two questions can then be asked. Firstly, whether the logic of scientific
development is exact and formalizable, or rather to be described as a
“dialectical logic of development”. Secondly, which kinds of social
regularities are involved, and how they succeed in bringing about a logic.
Kuhn’s offer is not formalizable, and is in fact dialectical. In the closing
passage of his article on “The Function of Dogma in Scientific Research”
he points out that
scientists are trained to operate as puzzle-solvers from established rules, but
they are also taught to regard themselves as explorers and inventors who know
no rules except those dictated by nature itself. The result is an acquired tension,
partly within the individual and partly within the community, between
professional skills on the one hand and professional ideology on the other.
Almost certainly that tension and the ability to sustain it are important to
science’s success.
[Kuhn 1963: 368f].
Elsewhere he points out that the whole process only functions because what

120
he speaks about here as “established rules” are indeed not explicit and
unambiguous rules but shared norms or values which individual workers
interpret differently:
[...] individual variability in the application of shared values may serve
functions essential to science. The points at which values must be applied are
invariably also those at which risks must be taken. Most anomalies are solved
by normal means; most proposals for new theories do prove to be wrong. If
all members of a community responded to each anomaly as a source of crisis
or embraced each new theory advanced by a colleague, science would cease.
If, on the other hand, no one reacted to anomalies or to brand-new theories
in high-risk ways, there would be few or no revolutions. In matters like these
the resort to shared values may be the community’s way of distributing risk
and assuring the long-term success of its enterprise.
[Kuhn 1970: 186134]
The Kuhnian framework, however, is not the only place where we have
encountered the need for a dialectical understanding and for application
of the concept of norms. An obvious dialectical tension manifested itself
in the contrast between the empiricist imperative:
Scientific explanations are only allowed to make use of concepts and to
postulate relations and structures which can be rooted in experience,
observation or experiment. Mythological explanations referring to entities with
no such empirical underpinning are inadmissible: they only obstruct genuine
scientific insight.
and its falsificationist counterpart:
We are allowed to use in our explanations whatever self-invented concepts
and hypotheses we like; but we should be aware that our hypotheses are indeed
nothing but hypotheses, preliminary explanatory models, and not the truth.
We should therefore constantly check our hypotheses as thoroughly as we can,
and we must reject them as useless as soon as they enter into conflict with
our observations of reality – i.e., as soon as they are “falsified”.

134
When this was formulated, the idea that a system may function better if its
components are allowed a margin of unpredictability (and the notion that real
systems function that way) was highly untraditional (for which reason Kuhn resorts
to the metaphor of “risk distribution”). During the last decade, of course, the
metaphor of “chaos theory” has popularized the idea.

121
As they stand, the two rules of scientific conduct solved some of each
other’s problems, as we remember – but in mutually unacceptable ways.
None of them, however, could be rejected as plainly irrelevant. They stand
in much the same relation as these two passages from Deuteronomy:
Thou shalt not kill. (5:17)
and
But of the cities of these people, which the Lord thy God doth give thee for
an inheritance, thou shall save alive nothing that breatheth. Namely, the
Hittites, the Amorites, the Canaanites [...]. (20:16f)
These rules, like the empiricist and falsificationist maxims, express a moral
dilemma, the admissibility of which is perhaps the most important distinc-
tive characteristic of a system of (moral) norms or values as opposed to
a set of binding juridical rules or a theory.
One of the things we demand from a theory is that it should be free
of inner contradiction (to be sure, the existence of a recognized contradic-
tion is no reason for automatic and immediate rejection, as exemplified
in the history of quantum physics; but it is at least an anomaly which one
should try to solve): if the same theoretical system predicts on one hand
that a bridge which we try to build will stand and on the other that it will
fall down, we should obviously try to find out what is actually going to
happen, or at least to find out why our theory cannot tell. One aim of theory
construction (though not necessarily its actual scope) is that it should be
fit to serve strategically rational action (Weber’s Zweckrationalität), which
it cannot do if producing contradictory predictions.
The same demand for internal consistency we make to juridical laws –
for this reason, the existence of capital punishment presupposes a clear
distinction between that sort of homicide which is murder and hence to
be punished, and the executioner’s work for which he gets his salary. No
juridical system could live undisturbed by a clear contradiction like the
one exemplified by Mosaic law; when a contradiction occurs in real life
(i.e., not only as the outcome of a thought experiment), the judge or some
other instance of authority has to decide which norm is primary; if the
contradiction presents itself recurrently, the legal system has to be adjusted.

122
Explanations of morality
Norm systems do live with contradictions, whether we like it or not;
that they do so is one main reason why different social actors choose
differently – cf. Kuhn as quoted on p. 121 regarding the “individual
variability in the application of shared values” (another obvious reason,
certainly no less important is that norms may be contradicted and
effectively blocked by our drives and desires135; often the two are inter-
twined in a way which in the concrete case makes it difficult to distinguish
moral dilemma from pretext). To understand why norm systems have to
live with contradictions we may take a brief look at the nature and origin
of norms and morality136.
One very influential explanation (which is historically coupled to
empiricist philosophy, and which shares its matter-of-fact attitude and its
bent toward atomistic analysis) is utilitarianism: Behaviour is morally good
if it is useful, i.e., if it promotes general human happiness.
Three problems at least inhere in this understanding of morality. Firstly,
it reduces morality to strategic rationality with an undefined aim: what,
indeed, is general human happiness if not itself a moral issue? Secondly, it
makes no sense of the experience of the moral dilemma: if two alternative
actions are both prescribed, that one is obviously best which according
to a cost-benefit analysis is most useful. Thirdly, it presupposes that the
consequences of a course of action are finite137 – if not, cost-benefit

135
As observed by St. Paul (Rom. 7:19): “... the good that I would I do not; but the
evil which I would not, that I do”.
136
It will be seen that I treat morality (or ethics) and norms as one thing. This is
not done by everybody. Certain authors would see norms as that which people think
is right, whereas morality is concerned with what is right in itself. Others would
reserve the term morality for serious matters and use only norms to denote, e.g.,
norms concerning good manners. (The existence of cultures where a person’s
unwillingness to apologize for pushing somebody involuntarily is a sufficient reason
to kill him in a duel suggest the latter distinction not to be cross-culturally valid).
137
Or rather that the actual value of their sum is finite, in the same way as the
present value of £1 to be paid each year from now to eternity is £21 at an interest
rate of 5% per year – i.e., if consequences which our children and neighbours have
to bear are less important than those which hit ourselves, and those which hit our
grandchildren and our neighbours’ neighbours count even less.

123
analysis is impossible.
Equally influential is Kant’s approach through the distinction between
the hypothetical and the categorical imperative in Kritik der praktischen
Vernunft (cf. [Høyrup 1993: 137]): My present action is morally good if,
when generalized into a rule, it is of absolute (“categorical”) validity138.
Strategically rational action (“under the hypothesis that you want to achieve
X you should do Y” – Kant’s hypothetical imperative), on the other hand,
is thereby neither morally right nor wrong. From a Kantian point of view,
it is hence not the action in itself which is judged morally right or wrong:
the moral judgment can only deal with the action in its context of
justification and intention.
In any case, it will be seen, there is no more place for the moral
dilemma here than inside the framework of utilitarianism (with a reserva-
tion to which we shall return). One, at least, of two rules in mutual conflict
cannot be of general validity; one, at least, is hence no moral rule.
Under the impact of Kantianism, utilitarianism has been split conceptu-
ally into act utilitarianism (the classical stance, which judges acts individual-
ly, from their particular consequences) and rule utilitarianism, which does
not ask whether single acts, but only whether rules of conduct are useful.
The difference between the two positions is larger than it may seem at first.
Firstly, they may judge the same action differently: if I need the money
and you already got all you want, act utilitarianism may find it justified
if I omit to pay my debt to you. But rule utilitarianism would see that the
rule “you need not pay your debts” would be the end of lending, and thus
damage those in temporary economic distress (a real-life instance of the
same divergence is cited below in note 159).
Secondly, rule utilitarianism, while fitting our immediate feeling that

This is, of course, precisely the presupposition of every cost-benefit analysis:

consequences which are beyond my horizon do not exist; consequences within my
horizon are only counted to the extent that those who suffer them count for me.
138
Such a generalization can of course be performed at different levels. For most
of Kant’s contemporaries it would be obvious that the execution of certain criminals
was morally right – not because of the generalization “you should kill other people”
but, e.g., because “you should protect society against the damage which could be
wrought by incorrigible criminals”.

124
morality is concerned with rules and not with the expediency of single
acts, misses the “positivist” simplicity which constitutes the merit of act
utilitarianism to such an extent that it is dubious whether it still deserves
the utilitarian label. It can never be given within a particular action under
which general rule it should fall (cf. note 138): is “hanging X” an instance
of “killing a fellow human being”, of “annihilating an enemy”, of “execut-
ing a war criminal”, or of “celebrating our victory by liquidating a war
criminal belonging to the enemy’s side”? Classification of acts under rules
already presupposes some kind of moral theory or proto-theory telling the
pattern of possible rules. No more than in descriptive science is there any
smooth road from “positive fact” to generalization139.
Rule utilitarianism, if it is to be believed, has no more space for moral
dilemmas between rules than Kantianism. If it is at all possible to calculate
the utility of a rule (which is certainly no simpler than calculating the utility
of an action), then one can also calculate which of two conflicting rules
is “more useful” and thus primary. It is only through its theoretical
shortcoming – namely because the ascription of an action to a rule is itself
ambiguous – that a specific action may present a dilemma. This weakness
(or force through weakness) it shares with Kantianism.
It is not possible to derive actual morality from utilitarianism of either
one or the other kind; but from a small set of socially accepted core values
(inner solidarity within the Twelve Tribes of Israel, combined with the right
to conquer and subdue, if we think of the Deuteronomy example), we may
derive a complex of norms which looks so similar to the standards of real
morality that the utilitarian idea can hardly be totally wrong140. Kant, on

139
That this is so was indeed pointed out by Kant in his third Critique, Kritik der
Urteilskraft (A VII, B VII; A XXIIIf, B XXVf; A XXIV, B XXVI): Only the faculty of
judgment, the very same which allows us to judge art and which cannot be reduced
to strict proof, allows us to decide whether or not a particular act falls under a
specific rule, and, in general, to perceive the particular as a case of the general (be
it rule, or principle, or [natural or moral] law). And only this faculty allows us to
“find” the general from the particular [or construct it, or whatever we do; changing
our terminology does not change the problem – JH].
140
Evidently, the “socially accepted core values” also belong to the category of
morals. The immediate scope of utilitarianism is thus only to reproduce the process
by which a few core values (in the traditional formulation summed up as “general

125
the other hand, is certainly right when emphasizing that moral norms are
characterized by claiming general validity. The kind of utilitarianism which
offers the best prospects as a “rational reconstruction” (or retracing) of the
emergence of morality is thus indubitably rule utilitarianism, its inherent
difficulties notwithstanding.

Morality, language and social practice

The phenomenon of the dilemma may advance our understanding by
another step. Norms should be seen as social phenomena. Human beings
living together in a society socialize each other (in particular, of course,
their children) and make sense of their existence by describing their
situation and actions in language141. Patterns of action which serve shared
interests will be formulated in general terms, as norms – and patterns which
go against shared interests will be forbidden, again in general normative

human happiness”) unfold as a complete moral system.

If a similar argument was to be used in order to explain the emergence of the
core values, it would probably have to involve something like Habermas’ “universal
pragmatics” (see below, note 141) as moulded into a less universal shape by the
necessarily restricted horizon within which any human society lives, together with
other aspects of human biological nature (“happiness” is certainly a cultural
variable, but only in exceptional cases reconcilable with starvation or extreme
physical distress). A complete philosophy of morality would then, furthermore,
have to take into account that the core values, qua participants in the general system
of moral norms, are themselves reshaped by the processes through which this
system unfolds.
141
Habermas, in his “universal pragmatics”, derives certain basic aspects of morality
not from the contents but from the very presence of this communicative situation:
If communicative messages could not be presumed to be true, they would not be
messages (even the lies of commercial advertisement only function because their
character of communicative messages make us accept their overt or tacit implications
at one level of consciousness, in spite of what we may know at other levels).
Similarly, the mutual character of communication entails at least rudimentary
human equality.
The pattern of the argument is obviously (though tacitly) borrowed from
Merton’s notion of “institutional imperatives”, to which we shall return below
(p. 135ff). At present, however, we are concerned with other aspects of the
communicative practice.

126
terms142. These normative prescriptions and proscriptions will be practical
knowledge, knowledge about how one should act. To the extent that norms
are made explicit there will be a certain amount of bargaining between
them at higher communicative levels, for instance through their integration
into religious systems, through the “thought experiments” of myth, drama
and literature (cf. p. 180), and through their interaction with legal systems.
The outcome will be total structures which are categorically different from
the structures of practical action and which possess a certain degree of
autonomy.
In these structures, as already more modestly in the enunciation of rules
of conduct in language, the general formulations of norms will be more
general than those patterns of action which they formulated; conflicts which
never occur in the living existence of society will arise if the norms are
taken to the letter and not in agreement with that social practice which they
formulated in the first place (even though bargaining and restructuration
tend to clear the most obvious conflicts away). In this way, dilemmas may
arise, at least as virtual or philosophical problems – as Wittgenstein [1968:
§38] tells, “philosophical problems arise when language goes on holiday”.
In its real life, a culture will possess a large amount of “tacit knowledge”
concerning the range within which norms should be applied (the Ancient
Israelites will have had no difficulty in recognizing whom “thou shalt not
kill” and whom to exterminate; nor have the fighters of countless Christian
armies, nor the host of eminently Christian judges). None the less, because
conflicting interests may interpret the range of shared norms differently
and tacit knowledge is necessarily blurred, the theoretical dilemmas may
also turn up in practice.
No social situation, moreover, is completely stable. Norms, on the other
hand, once they have come into being, become embedded in language,
religious ritual, myths, literature etc., and are thereby provided with a fair

142
Indeed, formulation in language cannot avoid being in general terms: general terms
constitute the backbone of any language, proper names are by necessity peripheral
and could never serve to tell anything on their own.
Already our use of language (e.g., the concept of “lending”) thus provides us
with an a priori pattern of possible rules with regard to which we interpret actions,
i.e. with a first version of the proto-theoretical foundation that rule utilitarianism
presupposes without recognizing it.

127
amount of temporal inertia. For both reasons, and in particular through
the two in combination, norms may end up governing behaviour far
beyond the range of experience from which they grew originally – and
what was once merely potential conflicts between norms may thereby
suddenly be actuated in possible behaviour143.
Change has been a condition of human existence as long as human
society has been human and communicative. The distinction between the
two sources for dilemmas is thus merely analytical. It is not possible to
point to a stage of innocence where only theoretical dilemmas existed.
Human societies, as long as they have been in possession of moral norms,
have always been troubled by moral dilemmas.
But not only total societies: Shared experience, specific patterns of
discourse, conflicting interests, and change over time also affect single
institutions and communities within society at large. These will therefore
develop a specific ethos of their own, which will share many of the
characteristics of the normative structure of general culture.
This brings us back to the creation of scientific knowledge. The
empiricist and falsificationist rules of conduct are, in fact, nothing but
community-specific norms, generalizing the experience and pragmatic rules
of conduct of working scientists, and brought into mutual conflict when
formulated and taken to the letter by philosophers. The shared anomaly
of the two imperatives – the absence of purely observational knowledge
which can serve verification or falsification – on its part, is of the same
kind as the dilemma presented by indefinitely life-saving medical technolo-
gies. If not created, it has come to the fore through the development of
sophisticated instrumentation, from the invention of the telescope and the
microscope onward.
It is thus not only in Kuhn’s interpretation that the “logic” of the
scientific process will have to be non-formalizable and dialectical. The
underlying pattern (or “logic”) of any social process governed to an

143
The actuation of conflicts because of changing social (here technological) realities
is exemplified by one of the central norms in traditional medical ethics: the patient
should be kept alive as long as possible. New medical technologies make “as long
as possible” indefinitely long, while transforming the concept of “alive” in a way
which changes the meaning of the norm through and through.

128
appreciable extent by norms or values has to be so.

Knowledge, norms and ideology

Should we then conclude (as Kuhn tends to do) that Popper’s and
Lakatos’s attempts at formalized non-dialectical methodologies are
descriptively irrelevant ideologies? Or, more brutally, that they have to
be descriptively irrelevant qua ideologies?
The question may be approached via a discussion of the concept of
ideology in the light of the theory of morality. Ideologies may be understood
as systems which possess both normative and descriptive aspects but which
cannot be fully analyzed into normative and descriptive elements – systems
which, in the same breath, tell how and what the world is and how one
should (therefore) behave, which merge theoretical and practical know-
ledge. Whether the Popperian and Lakatosian methodologies can be
descriptively relevant is thus a specific case of a general question: whether
the normative aspect of an ideology by necessity invalidates its descriptive
aspect (or by necessity invalidates it completely) – and then, in the second
instance, since the normative aspect was implied by the description,
nullifies even this?
Utilitarianism of either kind explains norms as strategically rational
prescriptions (with a tacitly presupposed aim of undefined “happiness”);
since strategic rationality cannot exist without knowledge about the
connection between our actions and their ensuing consequences, utilitarian-
ism presupposes that its norms translate knowledge. Quite as much cannot
be said if norms are generalizing reconstructions of patterns of action which
are adequate within a specific and more restricted socio-cultural horizon.
None the less, they generalize from what was adequate action within the
restricted domain, given the shared interests of the participants in the socio-
cultural system as understood by these. In terms of the materialist notion
of truth introduced in Chapters III (p. 46) and V they therefore contain
a core of truth reflecting features of this social situation – a core which may
be larger or smaller, depending upon the degree of generalization and on

129
the kind of reconstruction which has taken place144. Norm systems are
thus already ideologies, possessing an aspect of implicit descriptive know-
ledge; they are “proto-knowledge”145. Even presumed descriptive know-
ledge, on the other hand, presupposes a framework, a cognitive structure
prescribing implicitly the way in which the subject-matter should be
understood; since the selection of interesting features corresponds to a
particular practice (remember Australian aboriginal children developing
map understanding well before conservation, in contrast to European
children – see above, p. 24), a prescription of how to know also involves
an implicit delineation of a range of possible actions. Both from this point
of view and through its prescription of how to know it involves practical
knowledge. Knowledge systems, like norm systems, are already ideologies and
thus “proto-morality”. No absolute distinction between the two (or three)
domains can be made; differences are of degree although – it must be
stressed – the span from one extreme of the spectrum to the other is immense146.

144
Reconstructions aiming at inner consistency and utilitarian explainability may,
like the search for coherence of scientific knowledge, increase the truth value of
a norm system. Reconstructions aiming at agreement with mythical frameworks
need not do so.
145
An implication of this that cannot be pursued systematically in the present context
is the possibility that the cognitive aspect of a norm system may then, like any piece
of knowledge, be wrong, mistaken or (if we refer to the materialist notion of truth
that is presupposed in the argument) less true than possible.
So much remains of utilitarianism, even when norm systems are seen as
reconstructions of generalizations, that erroneous cognitive presuppositions
(inasmuch as they can be traced) can be used to judge the norm system.
146
The non-separability of description and prescription, it is true, contradicts the
implications which Hume drew from a famous observation made in the Treatise
of Human Understanding (III(i)1, ed. [Mossner 1969: 521]), and which have been
widely accepted within philosophy since then: there can be no logical derivation
leading from sentences built around “is” or “is not” to sentences built around
“ought” or “ought not”. Since the former sentences are descriptive and the latter
prescriptive, Hume’s observation seems (and is generally taken) to imply that
knowledge and norms are not only separable but actually separate.
The arguments that knowledge is proto-normative and norm systems proto-
cognitive does not invalidate Hume’s logical observation; but they do go against
what he concludes from it. What they say is that both norm systems and seemingly
neutral descriptions share the character of the question “When have you stopped
beating your wife?”: they make presuppositions which are not stated in the sentence

130
 It is, no doubt, legitimate to characterize Popper’s and Lakatos’s views
as ideologies for scientists; but that does not prove in itself that they have
not got a descriptive point. In Lakatos’s case this was argued quite
extensively above. As far as Popper is concerned, the question is rather,
which point? Is he right when declaring his methodological prescriptions
to be purely utilitarian norms, a guide to how science will progress most
rapidly? Or are they misleading in this respect, which would mean that
the purported cognitive (and hence also the normative) contents of the
ideology is wrong in the sense suggested in note 145? That the latter
possibility is to be preferred was also argued in some depth above. But
Popperianism also has another level of cognitive content with normative
implications – viz that science progresses because it follows Popper’s
precepts (cf. note 46), which implies that scientists have to be respected

itself, but without which the formulation of the sentence becomes meaningless.
It can be argued that the very distinction between “is-” and “ought-sentences”
is of the same character. The understanding of the descriptive statement presup-
poses our familiar “naïve” correspondence theory of truth; that of the prescriptive
statement builds tacitly on a no less naïve understanding of the freewill problem:
A prescriptive sentence “You ought to do X” presupposes, at least as understood
in Hume’s argument, that “you” are in possession of a Free Will and that you are
hence able to decide sovereignly to do or not do X. But the relation between free
will and determination is more complex than this. The naïve conception of the Free
Will renders the momentary feeling of deciding freely (e.g.) to shout at your
neighbour because the gangster deserves it. But thinking back tomorrow at your
present rage you may think that you overreacted on Mr. Jones’s jesting provocations
because you had slept badly; if it is your husband who shouts you may think so
already in the moment when it happens. What one decides to do is thus not as
independent of what is as he feels in the moment. But further, reversely: what is
and influences one’s way to act encompasses not only lack of sleep and actions
performed by his neighbour but also norm systems – norms which Mr. Jones has
transgressed as well as norms which allow one to scold the scoundrel or which
constrain one more than he can bear.
From the statement that “Mr. Jones has stopped beating his wife” we may
conclude that Mr. Jones is or has been married, since the statement would be
meaningless without this presupposition. From statements building upon an
indefinite array of presuppositions we cannot decide a priori which conclusions
can and which cannot be made; in particular, if an “is-statement” has normative
presuppositions (as “Mr. Jones is a criminal”, where the notion of a criminal is
defined legally and may encompass norms about what should be done to criminals),
then Hume’s argument fails.

131
as “objective beasts” by the surrounding society, and in particular have
to be more highly respected than Marxists and psychoanalysts and their
kin. This norm certainly is useful for the scientific community – but mainly
so with regard to its similarities with the world of organized crime147.

Value relativism and value nihilism

Another look at the categories of the general philosophy of morality
will permit us supplementary insights into the discussions about scientific
development.
It was asserted above that morality is able to live with dilemmas. This
is only partly true. Firstly, of course, the reconstruction of incoherent rule
systems through bargaining and various kinds of thought experiments have
the function (and, as far as many literary thought experiments are
concerned, the deliberate aim148) to expose and thereby to solve, to
surmount or to reconcile them. Secondly, awareness of the existence of
inconsistencies in the moral system may (in particular in secularized and
enlightenment periods, when the Solomonic wisdom of religious authorities
is questioned) lead to value relativism: norms there must be, but we choose
them freely – man is the measure of all things, as once formulated by
Protagoras (and held by utilitarianism). Or the conclusion may be value
nihilism: “If God does not exist, then everything is permitted” (Ivan
Karamassov) or “Good is only what is good for the strongest”, as Plato
expressed the ultimate consequence of Protagoras’ relativism in Gorgias
and the Republic.
Both attitudes can be found within the philosophy of science. Lakatos’s
whole methodology of “research programmes” is, indeed, a relativist
reaction to the breakdown of Popper’s claim that an absolute methodology
could, and should, be applied if we want to know in spite of the fallible
character of all knowledge: The community of experienced scientists is,

147
Members of the Sicilian Mafia speak of themselves, and want to be spoken of,
as uomini d’onore, as men of honour.
148
Think, e.g., of Sophocles’ Antigone, which confronts the norms of the city state
and of political society (represented by Creon) with those of ancestral morality and
human love (represented by Antigone and her fiancé Haimon).

132
in fact, the “measure of all things” within their science, those who decide
which methods are to be accepted. Another runaway Popperian has taken
the nihilist standpoint: Paul Feyerabend, whose discovery that Popper’s
proclaimed “rational method” does not work made him publish a book
with the title Against method, in which it is claimed that “there is only one
principle that can be defended under all circumstances and in all stages
of human development. It is the principle: anything goes”149.
Value nihilism is a tempting inference from the discovery that absolute
norms have a hollow ring. It looks like another version of the Socratic
principle that “the only thing which I know is that I know nothing”. But
the probable outcome of practiced value nihilism is not Ivan Karamassov’s
gentle desperation but Rodion Raskolnikoff. Outside literature, and in a
somewhat less drastic illustration: In local democracy it is unavoidable that
those who decide know some of those whom they decide about. An
absolute prevention of local favouritism and similar corrupt behaviours
is therefore only possible if local democracy (and local government
altogether) is abolished. The nihilist supporter of local autonomy will
therefore have to drop the prohibition of favouritism – any norm which
cannot be upheld absolutely cannot be upheld at all. The result, of course,
will be that vaguely endemic corrupt manners become epidemic.
Norm systems, indeed, are not only reconstructed reflections of
adequate patterns of behaviour. They are also what we usually take them
to be: regulators of behaviour (and only thereby, of course, reflections of
adequacy). Conflicting norms – in this case the norm of democratic
government as close to those concerned as possible, and the norm of decent
behaviour and of equal opportunities irrespective of kin, friendship and

149
[Feyerabend 1975: 28]. Feyerabend himself declares the philosophy to be anarchist
in his subtitle. Many currents, it is true, can be found within anarchism. Still,
Feyerabend’s principle reminds most of all of that which later anarchists have held
to be malicious lies about Bakunin. The political quotation in the chapter leading
forward to the principle is actually from Robespierre: “Virtue without terror is
ineffective”. Feyerabend is an anarchist because it sounds so nice, and in spite of
his confession (p. 20) to detest most anarchism and most anarchists past and present
because of their seriousness and their lack of respect for human lives. In language
and in the interest of philosophical provocation, too, anything goes.

133
protection – are so too. They cannot be absolute prescriptions150, but
through acts of balancing (affected, among other things, by the socialization
and tacit knowledge of the range to be given to each norm) they may still
put certain limits to behaviour in a situation which is strained by contradic-
tory claims and interests. Irrespective of the anomalies which were
discussed above, this also holds for methodological norms like the
empiricist and the falsificationist imperatives.

Institutional imperatives
Terms like “prescription”, “methodology” and “rule” are often used
in the vicinity of Popper, Lakatos and Kuhn (and in many other quarters
of the philosophy of science). Principles from the philosophy of morality
and norms should therefore be applicable to the process of scientific
development and work – as they are indeed applied in the present pages.
As a matter of fact, however, this approach is far from traditional. The
aspect of the sciences which is traditionally discussed through the concept
of norms is the sociology of science.
The seminal (indeed paradigmatic) work which launched this norm-
based sociology of science was an article by Robert K. Merton from 1942
on “science and democratic structure”151. Concerning the “ethos of

150
For this we do not need norms in conflict. The validity of rules in practice can
never be more absolute than the linking of single acts to particular rules.
151
This was the title given to it when it appeared as a chapter in [Merton 1968].
Originally it was entitled “A Note on Science and Democracy”. Both versions of
the title point to the origin of the essay in the antifascist debates of the late 1930s
about the role of science. The setting is explained by Robert Merton as follows:
[...] A tower of ivory becomes untenable when its walls are under prolonged
assault. After a long period of relative security, during which the pursuit and
diffusion of knowledge had risen to a leading place if indeed not to the first
rank in the scale of cultural values, scientists are compelled to vindicate the
ways of science to man. Thus they have come full circle to the point of the
reemergence of science in the modern world. Three centuries ago, when the
institution of science could claim little independent warrant for social support,
natural philosophers were likewise led to justify science as a means to the
culturally validated ends of economic utility and the glorification of God. With
the unending flow of achievement, however, the instrumental was transformed
into the terminal, the means into the end. Thus fortified, the scientist came to

134
science”, Merton explains that

The institutional goal of science is the extension of certified knowledge. The

technical methods employed toward this end provide the relevant definition
of knowledge: empirically confirmed and logically consistent statements of
regularities (which are, in effect, predictions). The institutional imperatives
(mores) derive from the goal and the methods. The entire structure of technical
and moral norms implements the final objective. The technical norms of
empirical evidence, adequate, valid and reliable, is a prerequisite for sustained
true prediction; the technical norm of logical consistency, a prerequisite for
systematic and valid prediction. The mores of science possess a methodological
rationale but they are binding, not only because they are procedurally efficient,
but because they are believed right and good. They are moral as well as
technical prescriptions152
The final sentence has a clearly Kantian ring. “Moral prescriptions”
are those which (are held to) have absolute character, while “technical
prescriptions” are merely tools for strategic rationality. But the actual
understanding of the nature of norms comes close to the one proposed
above. “Institutional imperatives”, in fact, are understood as norms which
at least to a certain degree must be respected if the institution is going to fulfill
its presumed role – in casu, the production of “certified” (as opposed, e.g.,
to revealed) knowledge.

regard himself as independent of society and to consider science as a self-

validating enterprise which was in society but not of it. A frontal assault on
the autonomy of science was required to convert this sanguine isolationism
into realistic participation in the revolutionary conflict of cultures. The joining
of the issue has led to a clarification of the ethos of modern science.
The reasons for public distrust of science have evidently changed since 1942. The
ecological crisis, for instance, was still below the horizon, and the involvement of
social science in the management of minds through scientifically designed
advertisement and propaganda were not yet conspicuous (although it had begun
as early as the 1920s, and Aldous Huxley had made it a prominent theme in his
Brave New World from 1932). But the phenomenon of public distrust remains, for
which reason clarification of the actual ethos of science is still important – not
primarily for purposes of self-defence but rather as a basis for self-critical reflection.
152
[Merton 1968/1942: 606f]. Two terminological details should be taken note of:
Merton speaks of certified, not certain knowledge. And he speaks about empirical
confirmation, not verification. Merton was never close to the logical empiricists, and
what may remind vaguely of their idiom is nothing but a concise common sense
description of the aims and methods of scientific work.

135
 These imperatives are not codified explicitly in any catechism for future
scientists; they become visible, as Merton points out, in the “moral
consensus of scientists as expressed in use and wont, in countless writings
on the scientific spirit and in moral indignation directed toward contraven-
tion of the ethos”153.
Merton himself, however, codified the system of institutional impera-
tives, finding four of them:
1. Communism, “in the non-technical and extended sense of common
ownership of goods”. Apart from eponymity expressing recognition
(“Boyle’s law”, “Rorschach test”), nobody has property rights to scientific
knowledge. Scientific results should be made public, firstly so that others
may use them, secondly in order to be subjected to criticism (prerequisites
for cumulativeness and certification, respectively). And further: “The
communism of the scientific ethos is incompatible with the definition of
technology as ‘private property’ in a capitalist economy. Current writings
on the ‘frustration of science’ reflect this conflict. Patents proclaim exclusive
rights for use and, often, nonuse. The suppression of invention denies the
rationale of scientific production and diffusion”154.
2. Universalism, which “finds immediate expression in the canon that
truth-claims, whatever their source, are to be subjected to preestablished
impersonal criteria: consonant with observation and with previously
confirmed knowledge. The acceptance or rejection of claims entering the
lists of science is not to depend on the personal or social attributes of their
protagonist: his race, nationality, religion, class and personal qualities are
as such irrelevant”. Universalism thus deals with knowledge, but no less
with persons. The optimal progress of knowledge requires that nobody who

153
[1968/1942: 605f]. As an aside on Merton’s own “context of discovery” it can
be told that Merton had intensive first-hand knowledge of these moral attitudes.
He wrote his PhD dissertation under the guidance of the historian of science George
Sarton, who continually taught him about what and what not to do – as Merton
told in a lecture at the George Sarton Centennial Conference, Ghent 1984.
154
Obviously, this norm has run into heavy weather since the Reagan-Thatcher era –
patenting of results has become endemic in every discipline which offers the
possibility, and financing authorities do what they can to accelerate the process.

136
is competent is excluded from the scientific institution155.
3. Disinterestedness is the norm which explains those features of scientific
activity which traditional hagiography derives from the particular moral
qualities of scientists (altruism, honesty, “objectivity”), or from their
personal motives: curiosity, thirst for knowledge. Scientific disinterestedness
requires that the scientist should not distort his science or his results in
order to gain personal advantage or in the service of particular interests
(in the terminology introduced on p. 118, disinterestedness thus imposes
the elimination of particularism). Transgressions in this field are probably
the ones which are most severely punished. As a rule, the scientist who
has been caught in deliberate fraud can start looking around for a different
career156. The same thing, of course, will happen to an accountant who
has betrayed his employer. Other professions, in contrast, have quite
different norms. The strictness with which the accountant is treated can
be compared with the lenience with which his employer is handled when
caught in insider trade at the stock market; the bad luck of the cheating
scientist can be compared to the praise bestowed upon the fraudulently
imaginative journalist157.

155
This was evidently, when written in 1942, a reference to the expulsion of Jews
and Social Democrats from German universities. The effect had been described
bitingly by the old David Hilbert (too much of a Nestor of mathematics to be
maltreated by the Nazis at that date), as NS-Reichsminister Rust asked him in 1934
whether it was really true that the Mathematical Institute in Göttingen had suffered
from the expulsion of Jews and their friends: “Jelitten? Dat has nicht jelitten, herr
Minister. Dat jibt es doch janich mehr!” (quoted from [Neuenschwander & Burmann
1987:25]).
156
In one Danish case from the 1950s, the Rector Magnificus of Copenhagen
University resigned from his office, not because he had committed fraud himself,
but because he had been unwilling to believe evidence that his son-in-law had done
so. The son-in-law when discovered gave up his scientific career, changed his too
characteristic name, and settled down as a practicing physician. A number of more
recent cases from the US are described in [Broad and Wade 1982].
157
One example: Some years ago the Danish journalist Jan Stage was forced to admit
in court that he had invented himself an interview with Bület Ecevit, endangering
thus this Turkish social democratic politician whom the military government had
forbidden to make any public announcement. Shortly afterwards, Stage’s employer
Politiken ran an advertisement campaign featuring precisely Jan Stage – much better

137
 Disinterestedness does not prohibit (and no norm can prevent)
misunderstandings of experiments, blindness to adverse results, and overly
dogmatic trust in established theories. What it proscribes are cases like the
physics professor from Copenhagen going to Thule in January 1968 “in
order to prove” that the crash of a B 52 carrying a number of H-bombs
had produced no radioactive pollution. Jørgen Koch’s slip (occurring during
an interview in the Danish radio news) demonstrates that the norm is not
universally respected. But the start one feels when hearing statements like
this demonstrates the existence of the norm as a norm – and overt admis-
sions of the kind can only undermine the public trust in the scientific
institution. The rule thus is an institutional imperative: rampant disrespect
endangers the institution.
4. Organized scepticism is the claim of the scientific institution that it
should not be subjected to the interests of other institutions or submitted
to their control, nor be bound by prevailing opinion and prejudice158.
This norm is certainly useful for the cognitive functioning of science by

known and much better suited for advertisement purposes than his honest
colleagues, it seems.
There are good reasons that the scientist and the accountant are more strictly
regimented than most other professions: In both cases, controls are almost
automatically applied; and in both cases, the fraud undermines the very raison-d’être
of the profession. The journalist, on his part, is rarely paid solely for telling the
truth; entertainment value is quite as important for newspapers getting an
appreciable part of their income from advertisement and the rest from customers
paying rather for entertainment and relaxation and for the subjective impression
of being informed than for information itself (in the case of papers which get
support from political parties, employers’ or trade unions, etc., other reasons ask
at least for a specific perspective on truth).
Professional honesty thus depends on the situation and the rationale of the
profession. So does the particular character of that honesty. The accountant has
to be honest about money. The scientist who mixes up private and institutional
money may be rebuked, fined, or perhaps dismissed – but may in even the worst
of cases hope to get another job within the profession. Public trust in science, and
the confidence with which others may use his scientific results, are not undermined.
158
It will be remembered from Chapter IV that this question was central to one
of the important differences between Popper1 and Popper2/Lakatos: Popper1’s
methodological conventionalism tends to make scientists bend to conventional
wisdom; Lakatos’s methodology of research programmes, on the contrary, will
protect research challenging accepted opinions as long as it remains fruitful.

138
offering moral support to scientists who risk conflict with those in power
by staying loyal to what they (suppose to) know, and by censuring
opportunism; but the attitude is one which will easily bring scientist into
conflict with their surrounding society159:
[...] Most institutions demand unqualified faith; but the institution of science
makes scepticism a virtue. Every institution involves, in this sense, a sacred
area that is resistant to profane examination in terms of scientific observation
and logic. The institution of science itself involves emotional adherence to
certain values. But whether it be the sacred sphere of political convictions or
religious faith or economic rights, the scientific investigator does not conduct
himself in the prescribed uncritical and ritualistic fashion. He does not preserve
the cleavage between the sacred and the profane, between that which requires
uncritical respect and that which can be objectively analyzed.
– or, to be more precise: his professional ethos tells him that he does
not need to preserve it. As we know, scientists are not only scientists but
also members of society, and many of them split their allegiance between
the norms of their profession and those of society (or their social group)
at large. Obviously, the Atlantic allegiance of the physics professor
mentioned above outweighed not only his allegiance to the norms of his
profession but even his awareness that there might be a problem (and, as
it turns out now that the US archives have been opened, his scientific
honesty).
Organized scepticism, it should be noted, has nothing to do with the
customary concept of philosophical scepticism. It does not imply scepticism
toward the possibility of obtaining reliable scientific knowledge – on the
contrary, the latter kind of scepticism is often promoted by those who wish

159
Maintaining this ideal in spite of pressure is thus, in the isolated instance,
contrary to act utilitarianism. It would be much more remunerative for scientific
institutions to agree with government officials, newspaper magnates, etc. But it
may be prescribed by rule utilitarianism: a science which has bent too obviously
to the desires or requests of authority tends to be decried at the next turn of the
road (in Spring 1990, the whole concept of “social science” was abolished in the
late German Democratic Republic, as a consequence of the too apologetic behaviour
of too many social scientists!).
Much hagiographic history of science probably serves the purpose of mediating
between act and rule utilitarianism at precisely this point: if science can be shown
to have been right in resisting heroically the now defamed authorities of former
times, it might well be right in continuing to defy authority.

139
to domesticate the provocative self-assurance of science encroaching on
the sacred domains of other institutions or to wash their hands when
convicted of having acted in bad faith160. It is therefore totally mistaken
to cite as evidence against the norm of organized scepticism five scientists
out of seventeen who would not accept reports of flying-saucers “no matter
who made the observations” ([Sklair 1973: 154], quoting a study from 1960);
precisely these five, indeed, illustrate the norm, being so sure about the
assertions of their science regarding the impossibility of interstellar travel
and being so knowledgeable about the susceptibility of even fellow
scientists to mass illusion that they felt entitled to contradict every report,
be it published in New York Times or made by the president of the National
Association for the Advancement of Science. “Organized scepticism” does
not contradict Kuhn’s findings about “dogmatism”; the two are, in fact,
sides of the same coin.

Theoretical versus applied science

Merton’s article became a Kuhnian paradigm for a generation of
American sociology of science, though mostly in a sadly trivialized reading.
As a reaction, a later generation has been eager to show that it is all wrong.
Many of the objections are irrelevant in the present context, but one point
of the discussion is not.
If scientific activity is regulated by the Mertonian norms, then scientists
in general, or at least the majority, should subscribe to these norms when
interviewed by sociologists – thus an alluring start of the argument. Most
of those whom the sociologists regard as scientists, however, are active
in industrial and other applied sciences; a survey which is meant to be
representative will hence be dominated by applied scientists; some surveys
have indeed looked solely at industrial scientists. The outcome of such
studies has been that many industrial and similar scientists do not subscribe
to the Mertonian norm system, in particular not to the communist

160
Thus the psychologist H. J. Eysenck, when it turned out that the research on
which he drew for proving that intelligence was determined by inheritance alone
was one immense fraud, started a campaign to prove (Popperian methodology at
hand) that astrology was as good a science as any other.

140
imperative.
Though often represented as counter-evidence (e.g., in [Sklair 1973:154]),
this is actually an essential underpinning of Merton’s argument, which
connects it to the above discussion. Industrial and other applied scientists
do not work inside an institution whose primary “institutional goal [...]
is the extension of certified knowledge”. The aim of their work is the
adaptation and application of new or (often) old knowledge (cf. chapter
VII); they are paid for producing knowledge which can end up as privately
owned technology161. That they do not follow all Mertonian norms (or
do so only in a restrictive interpretation, cf. below) merely illustrates that
these are institutional imperatives: they have little to do with the personal
character and history of scientists, and they do not belong to the corpus
of already existing scientific knowledge as an inseparable attribute; they
crystallize within an institution, i.e., a network of social interactions
organized around a particular core value (as it was called above, see note
140): “the extension of certified knowledge” in a collective process.
It can still be objected that even many scientists working within the
institutions of theoretical science (as well as whole institutions) do not obey
the norms. The Ancient Testament, however, also abounds with stories
about members of the Twelve Tribes killing each other, which does not
invalidate the norm “Thou shalt not kill” (Judg. 19-21 tells a nauseating
story of treason, group rape and massacres culminating in the virtual
extermination of one of the tribes at the hands of the remaining eleven).
An analogue of the Popperian argument from p. 57 will be more justified
in the present context than was the original version: the more often and
the more strongly the norms are broken, the less efficient is the work of
the institution, and the more likely is it to run into blind alleys. Keeping
all knowledge as private property (i.e., secret or encoded in a way which
impedes other from building on it) would mean the end of science;
prohibiting the teaching of “Jewish” physics in Germany (1933 to 45) or

161
These formulations only fit industrial scientists precisely. Applied social scientists,
for instance, rarely produce artefacts that may be bought and sold; their employer,
however (be it the state or a private institution), will control their work and use
its results in much the same way as an industrial corporation controls and uses
the work of an industrial scientist.

141
(supposedly “bourgeois”) genetics in the Soviet Union (1948 to c. 1955)
delayed research significantly in both countries; etc.

Further norms, contradictions, contradictory interpretations

In another respect, the Mertonian norms seem to agree less well with
the above analysis of the nature of norm systems: the scheme seems too
simple, clear-cut and free of inner contradictions. It looks more like a set
of “ethical rules” for the profession laid down by a supervisory committee
than as a piece of real-life morality162.
That the normative regulation of scientific practice is indeed much more
equivocal was already pointed out by Robert Merton in a humorous and
less widely read paper on “the ambiguities of scientists”. He lists nine pairs
of mutually contradictory norms, beginning as follows [Merton 1963: 78f]:
1. The scientist must be ready to make his new-found knowledge available
to his peers as soon as possible, BUT he must avoid an undue tendency to rush
into print. [...].
2. The scientist should not allow himself to be victimized by intellectual fads,
those modish ideas that rise for a while and are doomed to disappear, BUT
he must remain flexible, receptive to the promising new idea and avoid
becoming ossified under the guise of responsibly maintaining intellectual
traditions.
3. New scientific knowledge should be greatly esteemed, BUT the scientist
should work without regard for the esteem of others.
4. The scientist must not advance claims to new knowledge until they are
beyond reasonable dispute, BUT he should defend his new ideas and findings,
no matter how great the opposition. [...].

162
The difference between such precisely stated “ethical rules” (for journalists,
advertisement firms, etc.; not to be confounded with the general concept of ethics)
and morality is that morality tells (more or less unambiguously) how one should
behave. “Ethical rules” tend – in particular in cases where they do not derive from
the aim of the profession but go against it, as in the case of, e.g., advertisement
ethics – to state (like law) the limits of the permissible, ultimately thus telling how
far one may deviate from decent behaviour without risk of condemnation or penalty – or,
if “positive” and non-committal, they tend like Popper’s methodology to advertize
the honesty and altruism of the profession.
(The trivialized reading of) Merton’s scheme has indeed inspired the ethical
rules for scientific behaviour which are administered by academic authorities in
the US.

142
 5. The scientist should make every effort to know the work of predecessors
and contemporaries in the field, BUT too much reading and erudition will only
stultify creative work. [...].
[...]
“and so, on and on”, as stated at the end of the list of “moral dilemmas”.
That a normative system may function even though ridden with such
contradictions (and all practicing scientists know both the dilemmas and
how to deal with them in single cases) hinges on the tacit knowledge of
the participants in the social pattern that is regulated by the norms. Much
of this tacit knowledge is part of the paradigm which governs work within
a discipline; familiarity with this paradigm (and awareness of the number,
the character and the severity of the anomalies with which it is confronted)
allows the worker to decide whether a new suggestion should, e.g, be
dismissed as a mere “intellectual fad” or hailed as a “promising new
idea”163. The rest (inasmuch as two segments of tacit knowledge can be
distinguished) structures the “merely social” interactions within the
profession (e.g., whom to honour, and how).
Even the four “institutional imperatives” are ambiguous, and not
interpreted in the same way by everybody. To some, for instance, the
“communist” norm only means that final results should be made publicly
known at some adequate moment without becoming thereby public
property which can be used freely by everybody164. To others, the “com-
munism of the scientific ethos” is indeed “incompatible with the definition
of technology as ‘private property’ in a capitalist economy”, as claimed
by Merton165.

163
That not all workers make the same choice is one of the points in an activity
governed by norms and not by rules – cf. Kuhn [1970: 186] as quoted on p. 121.
164
In recent years, this attitude has been demonstrated in glaring and dismal
dimensions by the behaviour of the molecular biologist Robert Gallo, purportedly
co-discoverer of the HIV-virus. Within the humanities it is widespread in all fields
where a monopoly on the right to publish findings can be upheld – e.g., in
archaeology and Assyriology.
165
As the General Union of Danish Students fought its battle against contract
research at universities in the early 1970s, it took care to appeal to both versions
of the norm. Arguments that research paid by contracts could not be published
freely was correlated with the “narrow” interpretation; the slogan “Research for

143
 How the institutional imperatives are interpreted also changes – and
has to change – from one discipline to the other. To see why we may look
at the implications of “universalism” and “disinterestedness”. Both are
rejections (in somewhat different terms) of particularism, and thus objectivity
norms. Yet how science achieves actual objectivity (and thus implements
the two imperatives) depends on the problems and methods of the single
discipline. In those branches of medical research which test individual
cures, double-blind testing functions excellently; but to claim that this is
the only way to guarantee scientific objectivity is evidently preposterous
(already in medical disciplines like epidemiology or preventive medicine,
but a fortiori in sciences like astronomy, sociology and historical research).
In sociology, where the value system of scientific workers may overlap
with the value system that regulates the social unit under investigation,
it may give sense to claim that sociological science should be value-free;
as demonstrated by numerous hilariously absurd discussions taking place
in the late 1960s, claims that science in general (and not only sociology
and sociology-like human and social sciences) is/is not/should be/should
not be value-free give little meaning.
Even though discussions about the responsibility and political involve-
ment of the sciences are often expressed in different terms nowadays, the
theme can be pursued with profit. The idea that sociology should be value-
free was formulated by Max Weber; already in his writings the use of the
concept is ambiguous, and later interpretations are no less divergent. The
sociologist Alvin Gouldner [1973: 11-13], in a lucid and stylistically
sparkling essay on the meaning and function of “the myth of a value-free
sociology”, starts by presenting the opportunities offered by the concept:
[The value-free doctrine] enhanced a freedom from moral compulsiveness; it
permitted a partial escape from the parochial prescriptions of the sociologist’s
local or native culture. Above all, effective internalization of the value-free
principle has always encouraged at least a temporary suspension of the
moralizing reflexes built into the sociologist by his own society. From one
perspective, this of course has its dangers – a disorienting normlessness and
moral indifference. From another standpoint, however, the value-free principle
might also have provided a moral as well as an intellectual opportunity. [...].

the people, not for profit” was an appeal to the broad version.

144
 The value-free doctrine thus had a paradoxical potentiality; it might enable
men to make better value judgements rather than none. [...].
The value-free doctrine could have meant an opportunity for a more
authentic morality. It could and sometimes did aid men in transcending the
morality of their “tribe”, [...], and to see themselves and others from the
standpoint of a wider range of significant cultures.
The value-free doctrine could thus, we see, push sociologists from
particularism toward universalism in their value judgements, and enhance
the organized scepticism of the discipline. But Gouldner goes on with
harsher words:
But the value-free doctrine also had other, less fortunate results as well.
[...] many [...] used the value-free postulate as an excuse for pursuing their
private impulses to the neglect of their public responsibilities [...]. Insofar as
the value-free doctrine failed to realize its potentialities it did so because its
deepest impulses were [...] dualistic. [...].
On the negative side, it may be noted that the value-free doctrine is useful
both to those who want to escape from the world and those who want to escape
into it. It is useful to those [...] who live off sociology rather than for it, and
who think of sociology as a way of getting ahead in the world by providing
them with neutral techniques that may be sold on the open market to any
buyer. The belief that it is not the business of a sociologist to make value
judgements is taken, by some, to mean that the market on which they can vend
their skills is unlimited. From such a standpoint, there is no reason why one
cannot sell his knowledge to spread a disease just as freely as he can fight it.
[...].
In still other cases, the image of the value-free sociology is the armour of
the alienated sociologist’s self. [...]. Self-doubt finds its anodyne in the image
of a value-free sociology because this transforms [the sociologist’s] alienation
into an intellectual principle. [...].
There is on way in which those who desert the world and those who sell
out to it have something in common. Neither group can adopt an openly critical
stance toward society. [...].
When the value-free principle is understood in these ways – not as a
statement that sociology need not repeat or apply prevalent value
judgements but as a claim that the activity of the sociologist is itself above
moral judgement—it has thus become a disguise for breaches of the norms
of disinterestedness, organized scepticism and communism (in the broad
interpretation which emphasizes the general social responsibility of the
sciences). It is only in this variant (when no longer a norm of objectivity

145
but only of marketability) that the value-free postulate can be generalized
to all sciences.
Other fields possess their own more or less idiosyncratic specifications
of the “objectivity norms”, which we may pass over without further
discussion. The examples of the double-blind technique and the value-free
doctrine should suffice to make a general point: that norms at the level
of Merton’s “institutional imperatives” function not only directly as
regulators of scientific behaviour but also as “core values” around which
more specific norms crystallize during the practice of the single scientific
field. The conflicting interpretations of the value-free principle also illustrate
that the norms regulating scientific work (occasionally down to the specific
norms of the paradigm – cf. note 77) are susceptible to interaction with
the norms of general social life.

VII. THE THEORY OF INTERDISCIPLINARY

AND APPLIED SCIENCE

The discussion of Chapters IV and V presupposed (numerous examples

at hand) that scientific work is made within separate disciplines; Chapter
VI assumed that science may be either “theoretical” or “applied”. Both
suppositions refer to common sense knowledge; none the less, both are
contradicted by many traditional philosophies of science, not least
empiricism166. At best, they regard the division of science into disciplines
and the split between theoretical and applied knowledge as purely

166
Even Habermas, while setting aside the humanities and emancipatory social
science, follows Peirce in his identification of natural science as a mere means to
produce technology – cf. [Høyrup 1993: 171, 178].

146
pragmatic divisions of labour with no epistemological or philosophical
basis – “Knowledge is one. Its division into subjects is a concession to
human weakness” (H. J. Mackinder, quoted from [Mackay 1977: 99]).
The aim of the present chapter is to show that the divisions are
epistemologically founded, and to explore the relations between disciplinary
and interdisciplinary science and between these and the pair “basic”/“ap-
plied” science. After having looked in the preceding chapters at the nature
of (scientific and general) knowledge and at that specific practice which
aims at the production of scientific knowledge, we shall thereby be brought
to consider the production of scientific knowledge as an aspect of general
social practice.

Know-how and know-why

That scientific insight (“know-why”) can be used in strategically rational
practice as “know-how” is both a commonplace and part of our daily
experience – as illustrated by an advertisement slogan painted on a van
belonging to a pharmaceutics firm which I noticed in the street some years
ago: “Today’s theory – tomorrow’s therapy”. The commonplace is as old
as the scientific revolution – Thomas More as well as Francis Bacon claimed
that natural philosophy had the double aim of producing useful technology
and honouring God by studying his creation167.

167
On the other hand, the idea that technical practice is or should be derived from
scientific theory is not much older in Europe than the late Renaissance (in the
Islamic world, it can be traced back to the eight or ninth century C.E.). Aristotle,
in particular, tells in the Metaphysics that first the practical arts were invented in
order to provide us with the necessities of life. Later, theoretical science developed,
which was honoured more highly because it had no technical purpose (cf. [Høyrup
1993: 23]). Aristotle’s slogan would thus rather have been “yesterday’s therapy,
today’s theory” – or, even more radically, and denying that the two have any inner
connection, “Yesterday therapy. Today’s progress: theory”.
We may find Aristotle’s point of view outdated, and perhaps be repelled by
the philosopher’s disdain for those who provided him with food and shelter. Yet
what he tells was historically true in his times, and corresponds exactly to an
institutional and epistemological split between practitioner’s knowledge and
theoretical “know-why” which only broke down definitively in the nineteenth
century. Here, the typical Early Modern (More to Bacon) legitimation of science
as adoration of the creator and civically useful should not mislead us: It refers to
the use by practitioners of the elementary items from the theoretician’s tool-kit.

147
 The commonplace was translated into an analytical tool for research
statistics around 1960. At that moment, the OECD had come to consider
the development of adequate theory a crucial fundament for technological
and social progress (and thus for the supremacy of the “West” over the
Soviet Union168). As a consequence, the OECD embarked upon a massive
patronage of “science policy” programs in the member countries, and in
order to monitor the programs and the distribution of research funds, a
handbook containing prescriptions for the production of research statistics
(known as the Frascati Manual) was prepared. The manual contains the
following definitions of “basic” and “applied” research [Measurement of
Scientific and Technical Activities, 19f]:
– Basic research is experimental or theoretical work undertaken primarily to
acquire new knowledge of the underlying foundations of phenomena and
observable facts, without any particular application in view.
Basic research analyses properties, structures and relationships with a view
to formulating and testing hypotheses, theories or laws.
– Applied research is also original investigation undertaken in order to acquire
new knowledge. It is, however, directed primarily towards a specific practical
aim or objective.
Applied research is undertaken either to determine possible uses for the
findings of basic research or to determine new methods or ways of achieving
some specific and pre-determined objectives [...].169
In agreement with the commonplace, the OECD thus sees the realms
of “theory” and “therapy”, i.e., basic research and practical technology,
as belonging together, linked through applied research (and, in fact,
through a further bond constituted by “experimental development”, which

168
As we now know, this proved right, in the sense that the socialist countries were
unable to transform their massive investments in research and scientific education
into productive and institutional innovation. In the capitalist world, on the contrary,
the outcome of the first oil crisis, where the need for technological change had
imposed itself, was accelerated development on all levels.
Technological development, it is true, does not in itself create or destroy social
systems. But unequal economic development provided the basis for what came
to happen on the political and institutional levels.
169
It should be clear already from this definition that “applied research” may be
a much more creative process than the trivial “application of science” assumed
by Popper and others (cf. p. 93 and note 112).

148
makes science-policy makers speak of R&D, “research and development”) –
but still as clearly different activities. Why this has to be so is a question
to which we shall return below (p. 156 onwards).

The acquisition of theoretical knowledge

From a common-sense point of view, OECD’s strategy may seem
awkward. If technological progress and know-how is the aim, why then
finance activities aiming at the acquisition of “new knowledge of the
underlying foundations of phenomena and observable facts, without any
particular application in view” (the kind of knowledge of which I speak
as “theoretical”) and not merely research looking for the relevant kinds of
knowledge? Or, if basic research is relevant, why is there any need for a
particular stage of “applied science”?
The answer is connected to the dynamics of the Kuhnian cycle as
described above, but may also be approached from the common dynamics
of student projects as made at Roskilde University.
Such a project normally takes its starting point in an interesting
phenomenon (Greek for “that which is seen”) providing a first formulation
of the problem and a first idea about the relevant research strategy.
However, work guided by this formulation will normally set the original
phenomenon in a new light, through concepts which only appear to the
participants in the project due to the first phase of the systematic investigation.
Then they redefine the problem (that is at least what they should do), and
continue work according to the new definition of the problem and the
strategy. This spiral process may continue; in the end the treatment of the
problem (and indeed the problem itself) may have very little direct
connection to the phenomenon which originally inspired the work, and
much more with the internal but hidden structures which were revealed
by the systematic investigation (the “underlying foundations of phenomena
and observable facts” of the Frascati Manual).
The Kuhnian cycle of successive paradigms repeats the same structure
in larger scale. In the pre-paradigmatic phase, scholars start investigating
phenomena which present themselves directly, and which are either
intriguing or in some sense important – maybe the curious fact that glass,
when rubbed by a woollen cloth, attracts small pieces of paper, maybe the

149
effect of taxation principles on the productive activity and the distribution
of wealth and poverty within a country. The initial investigation results
in the discovery of new connections between phenomena giving rise to
new questions. In the ensuing phase, phenomena from seemingly quite
different domains (lightning!) may enter the field, whereas others from
the original area of investigation are discarded from view as after all
irrelevant with regard to the underlying regularities which have come in
focus (as when Mesmerism changed from a supposedly magnetic phenome-
non into hypnosis); in this phase (and more radically in later repetitions
of the cycle) questions will come to the fore which could not even have
been imagined when the process started – as we have seen, the basic
problems of Das Kapital could only be asked when Adam Smith had split
society into classes according to kind of income (cf. p. 62). But even though
the questions could not be imagined directly from common sense observa-
tion of the phenomena which constituted the starting point, the answers
remain relevant for explaining these phenomena – the Marxian analysis
of the economic process tells better than both physiocrats and Adam Smith
were able to do the dynamics of English economic life in the decades
preceding the appearance of The Wealth of Nations.
What begins as an investigation of a particular range of phenomena
belonging concretely together thus ends up as an investigation of the
underlying regularities or laws for everything which can be explained (e.g.)
from competition and monopolization of resources, or from Maxwell’s
equations for the electromagnetic field; what has begun as scrutiny of an
arbitrary section of the world (whether delimited spatially or by association
with a particular practice) is stepwise transformed into a general examina-
tion of reality under the particular perspective of the techniques and concepts
developed by the discipline. In this way we are led to much more
fundamental insights not only into the phenomena which set us on the
track but also into others – but only into particular aspects of these
phenomena170.
This contradicts the general way of speaking according to which the

170
That empiricist philosophy of science is blind to the epistemological foundation
for the splitting of knowledge into disciplines is thus a specific instance of its
general blindness to the role of theory.

150
different sciences deal with differ-
ent sections of reality: biology deals
with living beings (zoology with
animals, botany with plants); phys-
ics with non-living matter; econom-
ics deals with the economy; socio-
logy with our society and anthro-
pology with the functioning of
primitive societies171; electromag-
netic theory of course deals with
electricity and magnetism.
Yet electricity and magnetism
are not specific sections of the phy-
sical world – they are all-pervasive.
Similarly, no single human action
is solely economic. When I buy a
piece of cheese in the supermarket, my action can of course be analyzed
as a step in the economic cycle. But how I move around when searching
and paying may also be described by mechanical physics and anatomy;
that I feel a need to eat is a matter for the physiologist, and that I wish
to eat cheese could give an cultural anthropologist the occasion for an
explanation along quite different lines; my choice of a particular brand
instead of another would certainly be of interest for the advertisement
psychologist; etc.
That we tend none the less to accept the common way of speaking
illustrates to which extent we have come to see the world in which we live
through the perspectives of the established sciences and to think in terms
of their fundamental concepts. We have come to accept as self-evident,
for instance, that society consists of an economic, a juridical and a political
sphere; only second thoughts allow us to rediscover that these spheres are

171
The latter point, alas, is no parody in bad taste; it renders the objection actually
raised by a fellow member of the profession when the anthropologist A. Shiloh
wondered [1975; objection 1976: 554–555] that his discipline had done nothing to
understand how the National Socialist system of extermination camps had been
possible.

151
wholly abstract172.
The simple Kuhnian cycle (as translated here into the metaphor of
perspectives in dynamic development173) is not the whole truth about how
disciplines develop. Above (p. 74), the “neo-classical” or “marginalist”
revolution in academic economics was mentioned as a change of perspec-
tive which was inspired by political convenience rather than the inner
dynamics of classical political economy as based on the labour theory of
value. Similar examples are numerous174. More cognitively productive,
however, are two other processes: branching and integration.
Branching of disciplines can be explained at several levels. Since the
later seventeenth century, the number of active scientists has doubled every
fifteen years175; since it is impossible to follow in even modest detail the
technical work of more than a few hundred colleagues, larger disciplines
tend to split into specialties which loose contact (cf. p. 98 on the magnitude
of the group sharing a paradigm in the narrow sense). But the process can
also be explained at the epistemological level: Every time new conceptual-

172
It may be objected that the three spheres are embodied in separate institutions:
The stock exchange (etc.), the judicial system, and the parliament-and-party complex.
But the transactions going on at the stock exchange are undertaken on legal
conditions established by political authorities (and their very meaning indeed
defined in commercial and corporation law); similarly in the case of other seemingly
sphere-bound institutions. The spheres remain abstractions and analytical tools.
173
Where, if we continue the metaphor, each picture reveals where the camera
should be moved next in order to take an even more detailed picture – a feature
of the development of a discipline which was formulated on p. 117 as “progress
toward greater objectivity of its subjectivity”.
174
Without having researched the matter directly it is for instance my definite feeling
that the shift from Mertonian to anti-Mertonian sociology of science (see p. 140)
is better characterized as an “academic patricide” than as a Kuhnian revolution.
In disciplines without a firm inner structure (similar in many ways to Kuhn’s pre-
paradigmatic sciences), such patricides may be an easy way for the young wolves
of a new generation (in the present case my generation) to show their academic
valour.
175
This rule of thumb was stated by D. Price [1963]. It is certainly only a rough
approximation, and the process cannot go on in this pace for much longer (it seems
indeed to have decelerated somewhat over the last 20 years), but it is not without
value. A similar growth rate turns out to hold for the number of scientific
publications appearing each year and for the population of scientific journals.

152
izations or new techniques emerge within a discipline (i.e., at least when
a new paradigm takes over), new open problems turn up, and specializa-
tions materialize when groups pursue different questions (and tend to loose
contact quickly if further conceptualizations and techniques developed
around one question are not obviously relevant for other question
types176.
Integration or convergence take place when the methods and concepts
developed within one field turn out to be relevant for the understanding
of another, and when two fields provide compatible explanations of a
common object; they are furthered by the norm that scientific explanations
should be consistent177. At times they lead to a genuine integration of
separate disciplines into one, at times the process is better described as

176
Explanations at still other levels can also be given, e.g. through the specific
recruitment structure of a scientific institution. One such explanation has been given
of the explosive development of new disciplines in nineteenth-century Germany:
Since there could be only one (ordinary) professor for each discipline at a university,
the only way to argue for a new professorial position was the invention of a new
sub-discipline.
The various explanations do not contradict each other. They relate to each other
in much the same way as do the pushing, the material and shape of the billiard
ball, and the configuration of the table.
177
The ultimate consequence of this norm goes further than the integration of single
disciplines. Since reality (or physical reality or the reality of human life) is one, the
descriptions provided by the different sciences ought to be compatible and in the
final instance to be put together in a large coherent system of unified science. This
was the dream of the Romanticists, and again of the logical empiricists (Kuhn’s
Structure of Scientific Revolutions was actually published in the “International
Encyclopedia of Unified Science”, a series directed by the most important logical
empiricists). The neo-humanist movement in early nineteenth-century Germany
dreamt of unified Geisteswissenschaft. A rather recent formulation (concerning the
social sciences alone) runs as follows
Economics and anthropology, and for that matter sociology and political science
as well, are all – insofar as they are scientific – ultimately the same science. As
economics broadens its horizons, it will increasingly seem to be invading realms
of behavior that have in the past been reserved to other social disciplines. As
other social sciences become more rigorous, they will increasingly grow to
resemble economics (and, indeed, to be economics).
[Hirschleifer 1977: 133]
The formulations are rather asymmetric, and social scientists outside economics
will probably find them outright imperialist. But the principle remains the same.

153
an assimilation (in a Piagetian sense) of a whole domain to the paradigm
of another discipline (as when the neogrammarian sound shift laws were
explained by structuralist linguistics – cf. p. 104); at times a new discipline
emerges from the process alongside those which entered it; mostly (since
the delimitation of disciplines is always blurred) the outcome is somewhere
in between. In any case (as a colleague once summed up my view in a
metaphor), the interplay between branching and integration assures that
“the tree of science is no spruce but a banyan tree”.

The “Scientific-Technological Revolution”

The spiral development of scientific disciplines leads, thus it was
asserted above, “to much more fundamental insights not only into the
phenomena which set us on the track but also into others – but only into
particular aspects of these phenomena”. But whoever wants to build a
bridge or to improve the educational standard of the young generation
is interested in the bridge or education as functioning wholes, not only in
aspects. A bridge should be able to carry its load when built; but it should
also be stable toward wind and perhaps (depending on its location)
earthquakes, it should not corrode, and at best it should also correspond
to certain aesthetic norms; an educational policy should involve what to
teach to whom, but it does not function if teaching is badly made, if those
who should be taught cannot afford participation, or if teachers cannot
be recruited. Practice is concerned with many-sided sections of reality (by
necessity, if a “section of reality” is understood as what belongs together
within a particular practice).
This difference between the orientation of “know-why” and “know-
how” is the reason that they were not only carried by separate professions
but to a large extent by non-communicating professions until the early
nineteenth century: The higher levels of theory were carried by university
scholars, members of scientific academies, etc. Practitioners, on their part,
were taught as apprentices by other practitioners. Most of their knowledge
had originated in direct practice; what was borrowed from theoreticians
consisted normally of specific techniques (e.g., the use of logarithms in
navigation) and not of connected insights; typically these techniques would
belong on the basic and not on the advanced levels of the sciences of the

154
time178.
The first major change of this pattern was the appearance of the
scientifically taught engineer in the early nineteenth century (cf. [Høyrup
1993: 143]). Around the engineering schools, a particular “engineering
science” developed (cf. [Channell 1982]), the aim of which was so to speak
to translate and combine the knowledge of the “aspect-oriented” sciences
into practically relevant information.
Efforts to integrate theory developed with an eye on application and practice
making use of actual research results (and not just of the results that the
engineer had been taught in his youth by a teacher who had learned about
them during his early years in university) began around c. 1850, first in
organic chemistry and soon in electrotechnology (Siemens) and metallurgy
(Krupp etc.). This step has been spoken of as the beginning of the
“Scientific-Technological Revolution”. It was contemporary with parallel
attempts to develop behavioural sciences for use in “social engineering”,
the most important examples being probably Galton’s eugenics and
Lombroso’s physiognomic criminology (cf. [Høyrup 1993: 157]).
World War I, along with sonar, poison gas and other new technologies
developed by physical scientists on the basis of their theoretical insights
and their best research techniques, gave rise (in alliance with contemporary
industrial needs) to the development of “engineering psychology”, that
is, scientific investigation of how to design machinery in agreement with
what psychology could known about the perception, discrimination and
reaction capabilities of the human operator (cf. [Chapanis 1968])179. Alfred
Binet’s and Cyril Burt’s creation of the IQ-test-technique (cf. p. 7) occurred
in the same period and exemplifies the integration of psychological science

178
It goes by itself that this is only a rough approximation. It is, however,
significantly less distorted than the reverse simplification: that the Technological
Revolution taking place since the late Renaissance was derived from the Scientific
Revolution.
179
The establishment of engineering psychology thus exemplifies that direct
application of science in industry and warfare which provoked the Cambridge
mathematician Hardy to formulate in 1915 that “a science is said to be useful if
its development tends to accentuate the existing inequalities in the distribution of
wealth, or more directly promotes the destruction of human life” (quoted in [Hardy
1967: 120]).

155
with other divisions of general practice (in casu the educational system).
It was followed in the inter-war period (if we restrict the list to human,
social and organizational sciences) by the creation of industrial sociology,
by mass media studies aiming at advertisement efficiency, and welfare
economics (John Maynard Keynes, Gunnar Myrdal); further, during World
War II, by operations research, enhanced propaganda studies, and by
studies of the “cultural anthropology of the enemy” (undertaken in
particular by Ruth Benedict; see [Ember & Ember 1977: 42f]). After 1950,
the OEEC and its successor organization OECD promoted the generalization
of science-based “policies” (the very use of the word in this sense is indeed
a post-war phenomenon): economic policy, educational policy, criminal
policy, population policy, technology policy, science policy, etc. The postwar
era can hence be regarded as inaugurating the mature phase of the
Scientific-Technological Revolution180. This situation, and its problems,
is what produced the Frascati Manual; “science policy” is indeed a meta-
policy meant to produce the scientific knowledge needed in the other
policies.
Some of the sciences created during the scientific-technological
revolution started from a low level of theoretical knowledge and developed
the necessary knowledge directly for the purpose of application. The whole
field of mathematical statistics (which has had an splendid career since
then) was founded by Galton in this way as a tool for eugenics; Binet’s
IQ tests represent a similar instance. Both cases are characterized by the
absence of developed theoretical disciplines which could have served. The
general experience has been, however, that the theoretical sciences were
in possession of relevant knowledge of importance. If we refer to the

180
Evidently, speaking of this phase as “mature” does not imply that nothing new
is going to happen. As a matter of fact, another phase shift is already taking place.
The early integration of scientific knowledge into machinery and practical processes
was, so to speak, put in once and for all: Burt constructed a number of tasks and
put them together and then standardized the test; afterwards, the test was ready
for use and would not be changed (only replaced by a new version when it proved
inadequate). Increasingly since the 1960s, scientific knowledge is put into the
machinery and the process itself so as to allow servo-control and other kinds of
autoregulation (certainly with better success in car construction than in medical
service and other social processes).

156
questions formulated on p. 149 it is therefore clear why the science policy
experts of the OECD would find it appropriate to invest in basic research.
It had also been common experience (since the development of
“engineering science”) that the knowledge possessed by the theoretical
disciplines could not be used directly. If practice regards a section of reality,
and theoretical disciplines only provide a particular perspective on this
and other sections, no single theoretical discipline can do the job (whether
we think of bridge building or the planning of an educational policy); as
an engineers’ saying tells, “the difference between theory and practice
consists in condensed water”181. Moreover, if communication between
successive paradigms within the same field is only partial, the same holds
by necessity in stronger form for the paradigms of different fields which
must be combined in order to give a sufficiently complete understanding
of the practical problem to be dealt with. The combination of several
theoretical disciplines is therefore no easy process but one requiring active
transformation of the conceptual frameworks involved, and active analysis
of how the different frameworks relate to each other. This is the task of
applied science. Often, applied science may have the further task to
investigate questions left open by all theories – in the terms of the saying
quoted above, when other disciplines have provided the knowledge needed
to build a house which is stable and thermally and acoustically as well
isolated as required, then the applied scientist has to find out how to
modify their application in order to eliminate the unforeseen problem of
condensed water.
A special task of applied science comes from its direct coupling to
strategically rational action. Applied science should tell how to achieve
certain effects by deliberate action (cf. the discussion on p. 112 of Humean

181
A “real-life” example is the statement that anthropological investigation of
societies that base their agriculture on artificial irrigation may well contribute to
improvement of the social efficiency of irrigation systems – but only on the
condition “that some anthropologists, some of the time, take their problems not
from theories of social organization and social evolution but from the concerns
of the bulk of mankind – problems of food production, productivity, income
distribution, and employment – and work backwards from there” (R. Wade,
commentary in [Hunt & Hunt 1976: 405]). Basic science can only function as applied
science if it borrows the characteristics of applied science.

157
causality as a formalization of strategic action), and it should thus single
out those factors which can be influenced by a human agent (identical with,
or acting on behalf of, those who want the effect to be achieved): an applied
educational science which tells pedagogical skill to be a natural gift that
cannot be taught is only of interest for educational authorities if it also tells
how they may find the pedagogically gifted candidates; a science which
tells the skill to be teachable is only of interest if it also tells how to teach
it.
An applied science, it should be clear, is no trivial collection of results
from the theoretical sciences. It is no less of an active process than
theoretical science. It may also to some extent run through transformations
similar to the Kuhnian cycle; but it cannot move from “section-” to
“perspective-orientation” as long as it remains an applied science; its
problems belong, so to speak, not to the science and its scientists but to
those authorities, organizations or corporations that want to apply their
knowledge182.

Interdisciplinarity
Integration and convergence of theoretical disciplines is a familiar
phenomenon since long, and might well have been spoken of as
“interdisciplinarity” as long as “disciplines” in the modern sense have been
known. As a matter of fact, however, nothing in that direction occurred –
maybe because it was rarely clear whether genuine integration or cross-
disciplinary inspiration was involved, maybe because the traditional
normative ideal was “unified science”, not the unification of a few
disciplines.
What goes nowadays under the name of interdisciplinarity evolved

182
Cf. note 161 and the surrounding text. An informative discussion of the
implications of this ownership is [Schmid 1973], where it is argued that those who
want to make “science for the people” should accept that “the people” as organized
in unions or other organizations really formulate the problems (instead of devising
their own “people” and determining what should be its problems), and the author’s
own objection [1981] that existing social science is not fit for that model – whether
in a popular movement or in a firm or an organization, all the sociologist can
honestly do is to offer participation as a critical intellectual.

158
instead (without yet carrying the label) around the engineering and similar
schools and the emergence of applied sciences. In a general sense, an
“engineer” is a practitioner who has been taught, and makes use of, the
results of actual science in his work (with some imprecision, an engineer
is thus somebody who practices an applied science; actually, there is no
one-to-one correspondence between separate engineering professions and
actual applied sciences). The “engineers” of the nineteenth century were
mostly engineers in the received, specific sense, and they were taught as
engineers were still taught at the Technical High School a few decades ago
(the changes since then are modest, but they are authentic): The curriculum
consists of a number of “basic disciplines” – mathematics, physics,
chemistry, more or less adapted to the particular needs of the profes-
sion183 – and a number of “application disciplines”.
During the twentieth century, many other professions have become
“engineering-like” (e.g., nursing), and others have emerged as wholly new,
still “engineering-like” professions (e.g., social work). Derek Price, alongside
his doubling of the number of scientists every fifteen years since the later
seventeenth century, suggests [1963: 6f] that the number of people who
apply scientific knowledge in their daily practice (“engineers” in the present
pages) has doubled every ten years since 1900. Until the 1960s, they were
trained in much the same way as the classical engineers. Similarly, the new
applied sciences that arose (e.g., communication studies, industrial
sociology) followed the pattern of “engineering science”: combining a fixed
set of theoretical disciplines, adjusting and correcting their perspectives
until condensed water and similar problems have been minimized. In
neither context was any need felt to give the system a distinctive name.
Instead we may speak of this particular kind of unacknowledged
interdisciplinarity as an “engineering model”, which is characterized by
finite and well-defined multi-dimensionality.

183
Adapted at least through the selection of pertinent topics. When it comes to the
adaptation of perspectives, the teachers of the basic disciplines often have
difficulties. From my own experience at an engineering school I remember two
colleagues (B, a nuclear physicist, and H, trained as an engineer) producing a
common course in electromagnetic theory for future building engineers. After a
couple of years H observed that “B eliminates one Maxwell equation per year, but
the outcome does not change”.

159
 The name “interdisciplinarity” only surfaced (soon to become high
fashion) in the 1960s, as the finitist presupposition failed. It did so, on one
hand, in the training of the new “engineering” professions (and, for that
matter, even in the training of engineers stricto sensu). It was no longer
possible to teach young people the disciplines and the particular knowledge
from these disciplines they would need in the ensuing forty years of
professional activity.
Two of the strategies which were invented to circumvent the problem
are irrelevant to the epistemological issue: the production of human
“dispensable items”, trained for a very specific activity and dismissed when
that activity was superseded (customary in the lower ranks of computer
operation), and “lifelong education”, primarily of members of the traditional
professions by means of supplementary courses. A third strategy, cham-
pioned by the policy makers of the OECD, was the creation of more flexible
educational institutions encompassing both traditional university subjects
and the training of members of the practical professions, based on
“interdisciplinarity” and involving the students in “project work”184. Such
projects should simulate or exemplify the confrontation with practical
problems whose elucidation requires the integration of an array of
disciplines that cannot be specified in advance and once for all.
But the finitist assumption also broke down in the applied sciences
themselves. Classical engineering science had been concerned with a
particular and rather well-defined part of reality, and the formation of the
early applied sciences was based on the assumption that practical reality

184
A number of such teaching institutions connecting active basic research and the
education of practitioners were indeed created in the late sixties and the early
seventies. In Great Britain, in the wake of a reform plan “Education for the Future”,
a large number of teacher-training colleges were upgraded as “polytechnics” (they
have now been reclassified as “universities”, but the substance remains the same).
In West Germany, a number of such institutions were erected anew as Gesamthoch-
schulen, while others of the same character were given the name of universities
(e.g., Bremen and Bielefeld). (Outside the OECD, but with similar aims, the GDR
had its Dritte Hochschulreform). In Denmark, the new universities in Roskilde (1972)
and Aalborg (1974) represented the new idea.
Not all institutions were equally organic in their interdisciplinarity. In many
cases, the slogan covered realities not wholly different from those of the traditional
engineering schools.

160
could be cut up in pretty well-defined slices. A field like “educational
studies”, however, is not at all well-defined; as insights grow (and as
unforeseen condensed water turns up), new approaches are included in
the field, and old ones perhaps discarded as unsuccessful, at a pace which
had not been known in the earlier phases of the scientific-technological
revolution.
Worse perhaps than the indefinite number of disciplines that may be
involved is their character. The classical engineering sciences drew on
disciplines whose mutual relation was relatively clear – the Kuhnian
incommensurability between the paradigms of (say) mechanical physics
and metallurgical chemistry (both involved in our bridge) only becomes
serious at their most advanced levels; on lower levels, it is normally not
too difficult to establish who is responsible for condensed water. Even as
complex a project as “Manhattan”, the project that created the first atomic
bombs and which involved in total some 250,000 collaborators, followed
that model: mathematicians, physicists and chemists made the research
and the fundamental design, military people took care of secrecy, and large
industrial corporations built and ran the factories.
The mature phase of the scientific-technological revolution, however,
asked for scientific answers in realms where disciplines with much less
well-defined perspectives were involved185. The many OECD-sponsored
policies, if they were to build on scientific knowledge, would all involve
economics, legal studies, organizational theory and sociology, together with
the sciences involved in their particular objective186. “Global” problems

185
It is therefore somewhat paradoxical that participants in a research project on
the Lebensbedingungen der modernen Welt directed by Habermas formulated the thesis
in 1973 that science was entering a “post-paradigmatic” phase of “finalization”,
where fundamental sciences could be oriented toward any practical problem where
they were needed ([Böhme, van den Daele & Krohn 1973]; cf. [Schäfer (ed.) 1983]
and [Pfetsch 1979]). The examples which were set forth in the argument were
precisely basic sciences like classical mechanics whose perspective was well
understood (because active research had stopped), and applied sciences like
agricultural science which drew on such basic sciences.
186
The communication difficulties arising in these situations can be illustrated by
a dialogue which took place in 1960 at one of the OECD-sponsored conferences
which prepared the “new math” reform in mathematics education:

161
of population growth, resource conservation, climate and ecological balance
also became urgent during the 1960s, and since neither traditional
techniques nor common sense and laissez-faire had much to promise,
scientific insight seemed to be necessary if anything was to be done about
them. Even here, however, the mutual relation between the perspectives
of the relevant disciplines was not clear – how much had to be presup-
posed, e.g., about social, sexual and nutritional habits, and about ploughing
techniques in their interaction with the quality of the soil, etc., if Esther
Boserup’s optimistic research results about the tolerability of population
growth in Java were to be transferred to other contexts? In these applied
social and human sciences187 and in the scientific approach to the global
problems, interdisciplinarity thus turned out to be a problem which had
to have a name if it was to be discussed; very soon, the name of the
problem got the status of a slogan which in itself was believed to procure
the solution.

Hans Freudenthal [mathematician, but primarily a main authority on mathematics

education]: We could teach anything, drive the children in any direction. But there
exist other things at school. We must see the whole together.
Jean Dieudonné [mathematician, and the leading figure in the formalist “Bourbaki”
transformation of mid-twentieth-century mathematics]: Non, nous parlons de
mathématiques. Le reste je m’en fous. [“No, we speak about mathematics. Fuck
the rest!”].
[Grubb & Kristensen 1960: 12f]
No wonder that the OECD was forced to discover the problem of interdisciplinarity.
187
Speaking of educational studies etc. as “applied social and human science” is
in itself an illustration of the problem. Many of the policies erred because of a belief
that everything could be planned from organizational theory, economics and
sociology alone. But any strategy to regulate our behaviour (and the policies are
such strategies) must take into account both our social existence and our existence
as cultural beings, producing and reacting on meaningful communication in a
historical context. On the level of applied science, the distinction between human
and social science is absurd.

162
Interdisciplinarity in basic research
Very soon, too, the concept spread to “non-applied science”, and it may
be here rather than in the genuine applied sciences that it really gained
its spurs as a slogan. One current leading in this direction was that kind
of “science for the people” which, in polemical formulation, tended to
devise its own “people” and determine what should be its problems (cf.
note 182). In less polemic words: That widespread current in the radical
scientific environment of the late sixties and the seventies that tried to
develop knowledge of direct relevance for actual societal and political
issues: peace research, women’s studies, black studies, critical studies of
education188, critical science policy studies, etc.
The Frascati Manual has a term for this kind of fundamental research
[Measurement of Scientific and Technical Activities, 19f]:
[...] in some instances basic research may be primarily oriented or directed
toward some broad fields of general interest. Such research is sometimes called
“oriented basic research”.
Not all research belonging to this class is and was of course politically
radical. Fields like “Soviet studies” and “China studies”, whose aim it was
to know anything of relevance for understanding the Soviet Union or China,
were often sophisticated espionage in disguise189, and some of the scholars
and institutions making peace and conflict research were more interested
in how the US might establish their own peace in Vietnam than in avoiding
wars, or they were paid by institutions with such interests. Research
laboratories financed by particular industries (Philips, Carlsberg, etc.) tend
to ask for research connected to possible products (“research dealing with
beer is in favour”, as a colleague summed up what he had learned when

188
Not only my knowledge of the debates surrounding the new math reform but
also my familiarity with (and my particular interpretation of) Piagetian genetic
epistemology goes back to engagement in critical studies of mathematics education.
189
The Danish Institute of Eastern European Studies, which was no cover for such
intelligence work, had great difficulties in overcoming the suspicion of East
European authorities [Andreas Jørgensen, personal communication].

163
examining the purportedly “basic” research of the Carlsberg Labora-
tory190).
There are strong institutional and financial reasons that China studies
and Beer studies retain their “orientation”; in this respect they show
themselves to be more closely related to the applied sciences than to basic
research. Fields defined by the personal engagement of the workers, on
their part, have turned out to exhibit much of the dynamics of scientific
disciplines in the pre-paradigmatic phase; if the vocabulary had been at
hand, it would indeed have been possible for the early economists of the
eighteenth century to speak about their science as “interdisciplinary studies
of the problems of wealth”, involving social statistics, political and social
philosophy, and history. In some cases, e.g. in women’s studies, it has also
been possible to observe something like a Kuhnian circle – not only in a
sense that certain books (at one moment Germaine Greer’s Female Eunuch)
have acquired a paradigmatic status, only to be replaced by another
exemplar after a while, but also through assimilation and accommodation.
Thus, at a certain moment approaches which had been used in women’s
studies turned out to be relevant in various kinds of minority studies;
thereby features of the female situation turned out to be specific instances
of something more general, and many workers in women’s studies began
speaking about women as “a minority” regardless of actual numbers. In
a commonsensical statistical interpretation, this is evidently absurd.
Epistemologically, it is not: from that perspective it tells that deeper work
had shown the fundamental structure of the social minority situation not
to be mere statistics. As it happened when Newton took the common sense
term “force” and gave it a specific interpretation (at odds with daily usage)
and when Freud did the same to the “sexual drive”, the concept of a
“minority” was reinterpreted so as to reveal deeper structures of reality –
structures that only come to the fore through the dynamics of a theoretical
discipline.
In the end, the interdisciplinary interests of the 1970s have resulted
in that kind of processes which were spoken of above (p. 153) as “integra-

190
“Det må gerne handle om øl” [Uffe Sæbye, personal communication]. That was
in 1977. Since then, the official aim of the laboratory has been redefined as applied
research.

164
tion or convergence”. Instead of being a universal solution,
“interdisciplinarity” in the theoretical sciences (whether spoken of as such
or not) has turned out to be one moment in the global dynamics of
scientific knowledge, a complement of the Kuhnian cycle and no alter-
native – the mediation which takes care that the deeper knowledge which
is built up in the development of theoretical science is never totally cut
off from general practice. Dialectic, a fundamental feature of the individual
acquisition of knowledge and in the development of a discipline, is also
to be found at this level.

VIII. ART AND COGNITION

Further investigations of the sociology of the scientific-technological

revolution might be fruitfully contrasted with the deliberations of Chapter
VI concerning the role of norms in the regulation of the scientific process.
It is clear, for instance, that “big science” – the activity of large, hierarchi-
cally organized research laboratories and organizations – leaves little space
for individual, value-based decision on the part of most participants.
Decisions of importance are taken by the managers of the projects, and
their primary loyalty is not necessarily directed toward the scientific value
system. It is also clear that scientists who depend critically for the funding
of their research (and ultimately for their living) on research councils or
similar bodies may tend to let their research interests be determined not
by the “prescriptions” of the paradigm as to what is important but from
what they suppose the granting authority will appreciate. Quite often this
authority can be safely supposed to favour some kind of “social” utility –
relevance for export industry, not too critical understanding of social and
cultural change, etc.; scientists may therefore be driven toward presenting

165
a more “finalized” picture of their scientific insights than warranted (cf.
note 185), and work accordingly, i.e., on levels where theoretical develop-
ment has stopped191.
Instead of pursuing such issues, however, we shall let these suggestions
suffice and return to philosophical questions, examining what the episte-
mology developed so far has to tell in relation with aesthetic theory – in
other words, we shall address the relation between art and cognition, which
is one of the central questions of aesthetics, though only one among several.
Raising this question may look like barking up the wrong tree, but the
conclusion will lead us back to our main topic – it is indeed absurd (as
once remarked by Lenin) to deny the role of fantasy in even the most exact
science.

Knowing about art

Let us imagine that we open the radio and hear the beginning of Die
schöne Magelone. When encountering this or any other piece of art, we know
that it is there. In the present case we perceive the sounds, we notice that
they constitute music, we distinguish a piano and a voice; we may discover
that the words are German, follow the words, perhaps we even recognize
the work or at least the composer.
If instead the music is an ethnographic recording from Burundi in
Central Africa, we may have greater difficulty in bringing the work from
perception to classification. We may be unable to identify the instruments,
and we may feel puzzled by its complex rhythm. But we still recognize
that it is there.
In this sense, the problem of art and cognition is relatively trivial. We

191
As Danmarks Grundforskningsfond (Danish Fund for Basic Research) distributed
its first 800,000,000 DKr to 23 research projects, critics pointed out that most projects
were applied rather than basic; the chairman found no better defence than a claim
that all “contained important elements of basic research”, which seems to imply
that all were at most oriented basic research (Weekendavisen, 21.5.1993 p. 5). The
selection was also strongly biased toward “mainstream research”. This might in
itself seem a justifiable choice if we believe Kuhn’s arguments in favour of normal
science; but since oriented basic research tends either to be similar to pre-
paradigmatic science (where “normal science” does not exist) or to be finalized,
the cocktail is dubious.

166
may also take Saussure’s Cours de linguistique générale in our hands and
notice that it is a book, that it is written in French, etc. This is not what
we mean by the cognitive dimension of a piece of scientific theory. What
epistemology investigates is the relation between the theory set forth in the
book and the purported object of this theory (language, in the actual case).
However, that a piece of art, if we are to understand it as such, has
to “be there”, is not quite as trivial as it would seem. This follows if we
ponder the relation between two traditional conceptual pairs.
There is widespread agreement that the concepts of “beauty” and “art”
are closely linked. Some define one from the other, others do not go so
far but claim that “beauty” in some sense (at least through conspicuous,
intentional and provocative absence, which we may characterize as a
“negative aesthetic”) distinguishes the work of art192. The explanation
of “beauty” with reference to (sensual and/or intellectual) “pleasure” is
also conventional. It is clear, however, that the pleasure of senses which
have not been integrated in the “unified space” (gustation and olfaction,
and the senses of pain, heat and cold – cf. note 17) is never referred to as
“beautiful” in what we feel to be the genuine sense. Sensations that cannot
be apprehended as sensation of something cannot be “beautiful” however
pleasant they are. But art, if connected to the category of beauty, must then
be something which can be apprehended by the senses that give access
to unified space.
However, since the beauty of a poem, whether apprehended through
reading or through the ear, does not depend on its actual location, “being
there” cannot itself be at stake. But then a necessary condition for some-
thing to be art must instead be that we grasp it by that kind of intellect which
makes use of the senses of unified space, i.e., that kind of intellect which
sets itself apart from what it perceives, from the things that “are there”
physically or conceptually193.

192
Useful general surveys are [Dieckmann 1973] and [Beardsley 1973].
193
It could be added that a work of art, qua work, i.e., something which has been
produced by somebody (cf. [Heidegger 1977/1936]), must necessarily be there. But
what has been produced and is indubitably somewhere may non the less be
perceived without being grasped by the senses which locate in a there – as illustrated
by the work of the perfumer.

167
Knowing in art

Though not fully trivial, this remains so far a modest conclusion, and
we may still ask whether the work of art stands in a relation to something
else which is similar to the way Saussure’s book (theory) relates to language
(reality). May we claim that a work of art encompasses or transmits
knowledge about something?
At times the answer is an obvious yes. Many works have a clearly
descriptive dimension (for which reason all aesthetic theory from Aristotle
until the eighteenth century spoke of mimesis or “imitation” as an essential
aspect of art). For instance Malinowski’s poem “Kritik af kulden” (“Critique
of frost” [1980: 5])194:
Tidligt i marts vender vinteren In early March winter returns
tilbage And the newly pruned apple trees
Og havens nybeskårne æbletræer Float like furred ghosts
Svæver som pelsklædte spøgelser In austere frost and full moon. In
I streng frost og fuldmåne. Her- here
inde A twig stands in bloom.
Blomstrer en gren.
As it stands, this seems to be a naked description of a situation. But this
situation is of no interest in itself. This contents cannot be the reason that
the poem is printed and sold (nor that Malinowski would spend his all
too short life on writing poems).
In certain cases, analysis along these lines would even lead us to
characterize the work as a lie; no wonder that the era which considered
mimesis a central characteristic of art repeated time and again that Homer
was the greatest of liars195. An illustration of this point is provided by

194
My translation, as usually (with apologies to Malinowski, and to all the poets
whom I disfigure in the following).
195
Thus Aristotle, in De poetica 1460a19f (trans. Bywater 1924): “Homer, more than
any other has taught the rest of us the art of framing lies in the right way. I mean
the use of paralogism”.
That the mimesis of the poet does not have to depict an actual (or historical)
state of affairs is what (etymologically) makes him a poet, somebody who produces
something; we might translate him literally into “a maker”. As the term came to
designate the maker of verse, it acquired connotations not too far from our idea
of “creativity”. However great the changes in the character, function and under-

168
Cecil Bødker’s poem “Generationer” (“Generations” [1964: 103])
Faderen stod på den bare jord The father stood on naked ground
og gjorde et hus making a house
med egne hænder. with his own hands.

Sønnen steg op på faderens skul- The son climbed his father’s shoul-
dre ders
og satte nye etager på adding new stores
med andres hænder. with the hands of others.

Sønnesønnen lå på ryggen The grandson lay on his back

på tagterrassen on the roof terrace
og tog solbad. sunbathing.
Evidently, the situation described has never existed. The poem, none the
less, seems utterly meaningful – but this must be in an unfamiliar sense
(“metaphorical”, we would normally say, and believe that a term solves
the problem; it doesn’t really before we have made clear(er) its relations
to other terms and concepts).
At times, the contents of the work is not a description but a message,
a (moral or similar) opinion about the state of the world, directly expressed
or implicit. Let us consider Gelsted’s “Døden” (“Death” [1957: 78]):

Aldrig mere skal jeg se et træ Nevermore shall I see a tree

– hvile i det svale bøgelæ. – repose in the beeches’ cool shel-
ter.

Forårsskyer går i himlens blå, Spring clouds drift in the blue of

aldrig mere skal jeg se derpå. heaven,
nevermore shall I watch them.

Timen kommer, som jeg ikke ved, The hour arrives that I do not
hvor i glemsel alting synker ned.. know,
where everything slides into obli-
vion.

standing of art during the last 2500 years, it is hence also possible to point to
continuities that make it meaningful to speak as generally about the artistic sphere
as done in these pages.

169
 Dækket til af jord som nu af nat Covered by soil as now by night
ligger jeg af lys og liv forladt. I lie, deserted by light and life.

Glemt er hver en drøm og hver en Forgotten every dream and every

sang, song,
intet er jeg som jeg var engang. nothing I am which once I was.

Intet er den jord jeg dækkes af, Nothing the soil that covers me,
ingen hvile er i ingen grav. no rest is found in no grave.

Vilde, røde hjerte, alt du led, Fierce red heart, all you suffered
intet er det i al evighed. is nothing for ever and ever.

The poem may be read as a commentary to a famous Epicurean maxim

(ed., French trans. [Solovine 1965:139]):
Death is nothing to us: since that which is dissolved is deprived of the ability
to feel, and that which is deprived of the ability to feel is nothing for us.
Since Gelsted was well versed in Greek literature, and familiar with ancient
philosophy, the poem probably is a commentary, and an objection. A work
of art may thus also speak at the same level as a philosophical argument.
This allows us to formulate with more precision the question which also
follows from the observations that Malinowski’s “facts” are uninteresting
and Bødker’s wrong: why is Gelsted’s objection expressed in a poem, and
not in another philosophical statement that “what I cannot accept about
death is that my sufferings shall be forgotten”? Why will Malinowski tell
in verse instead of prose that defeat is never complete and definitive? Why
should Bødker express her critique of liberalist ideology as poetry? This
question will be pivotal for the argument below.
At times the obvious answer to the question whether a work of art
transmits knowledge about something is no. A generally accepted instance
is Bach’s Musical Offering, where neither description of things (not even
in a metaphorical sense, whatever that is) nor moral or similar opinion
is to be found196. Vivaldi’s concert The Spring from The four seasons,

196
Except, of course, in the sense that a work of art which follows a certain stylistic
canon “convincingly”, can be taken as an indirect argument in favour of this canon,
and hence also – inasmuch as the canon reflects norms and attitudes belonging

170
however, is no less adequate in spite of its title and programme. Only
because we know that it should correspond to typical situations belonging
to a particular season are we able to recognize them: they do not corre-
spond to them in accordance with any code of more general validity – so
far from it, indeed, that Vivaldi could use precisely the same music in his
opera Montezuma for the scene where the victorious Spaniards celecrate
their triumph. Words, of course, are also different from what they describe
(cf. the screwdriver and the screw of p. 46); but in this case correspondence
follows from a more general code.
What makes Vivaldi differ from Bach is thus not presence versus
absence of descriptive contents. It is rather that Vivaldi’s style allows us
to organize what we perceive with less mental effort, and that we are less
disquieted by it. The Musical Offering forces us to concentrate. Anton
Webern’s reinstrumentation of its “Ricercare”-movement asks even more
from us. Vivaldi’s concert does not force us to concentrate but has sufficient
content and density (again: whatever these metaphors may cover) to make
attention and repetition rewarding.

Fresh eyes

The issue of concentration and attention brings us to one of the central

twentieth century views on the role of art in relation to the problem of
cognition: The role of art is to bring us beyond cognitive routine, to make
the familiar unfamiliar in order to let us see it with fresh eyes. A radical
yet quite plain example taken from the world of music is John Cage’s Credo
in US, which starts by flattering the listener’s conventional musical
understanding, quoting the beginning of Dvořák’s Ninth Symphony From
the New World, and then suddenly jumps to something which in this
conventional understanding is totally cacophonous (and continues to jump

in the domain of general culture – to express a view with general implications.

In this sense, Bach’s Offering to Friedrich II of Prussia can be seen as a justification
of courtly formalism.
But as when we spoke of the “metaphorical description”: this statement opens
up a problem, and does not settle it. Namely: how does a stylistic canon reflect
norms and attitudes? So far we may only conclude that the “obvious answer” is
not necessarily the final word.

171
back and forth between the two). But the point of view exists in many
variants.
Very important is Brecht’s theory of Verfremdung (to be translated
“estrangement”, “making unfamiliar”, not “alienation”, which translates
Entfremdung). When Brecht undertakes to shatter the illusions and the
identification produced by naturalist stage play, his intention is that we
shall not be allowed to indulge in trivially sentimental pity with the poor
Mutter Courage as she looses all her children in the 30 Years’ War197.
He wants us to recognize her role as co-responsible for what happens to
herself as well as to her children – we should judge and hence learn and
not only empathize. Similarly, when Peter Weiss lets the contradictory
forces of the French Revolution be embodied by lunatics198, he prevents
us from identifying with one of the parties and forces us to accept the
dilemma – and having de Sade embody moral disengagement he prevents
us from dodging the dilemma by taking this facile position.
Another exponent for the position that the role of art is to liquidate
easy routine is modernism. Whereas Brecht and Weiss want to make us see
a state of affairs of general importance with fresh eyes, much main-stream
modernism (not all! and not all modernism is mainstream!199), when it
has to state a theoretical position (and in particular when it is explained
by its academic advocates), asserts that the aim of art is to force upon us
the fresh eyesight abstractly, without any engagement in morally important

197
Mutter Courage und ihre Kinder [Gesammelte Werke, vol. 4].
198
Peter Weiss [1964], Die Verfolgung und Ermordung Jean Paul Marats, dargestellt
durch die Schauspielgruppe des Hospizes zu Charenton unter Anleitung des Herrn de Sade.
199
One strong exception, borrowed from Paul Celan’s Schneepart, which I shall
abstain from translating:
Ich höre, die Axt hat geblüht,
ich höre, der Ort ist nicht nennbar,
ich höre, das Brot, das ihn ansieht,
heilt dem Erhängten,
das Brot, das ihm die Frau buk,
ich höre, sie nennen das Leben
die einzige Zuflucht.
[Celan 1975: II,342].

172
real life issues which might lure us into believing that this real life is what
the work of art is about (we might say that they see Bach’s Musical Offering
as a model for art in general).
A theoretical formulation of the view was provided by the “Russian
formalists” (c. 1915-1930), who influenced both Brecht and Eisenstein’s
montage technique. René Wellek [1973: 173b] sums up their views as
follows:
They had grown up in a revolutionary atmosphere which radically rejected
the past, even in the arts. Their allies were the futurist poets. In contemporary
Marxist art criticism art had lost all autonomy and was reduced to a passive
reflection of social and economic change. The Formalists rejected this reduction
of literature. But they could accept the Hegelian view of evolution: its basic
principle of an immanent dialectical alteration of old into new and back again.
They interpreted this for literature largely as a wearing-out or “automatization”
of poetic conventions and then the “actualization” of such conventions by a
new school using radically new and opposite procedures. Novelty became the
only criterion of value. [...].
Jan Mukařovský (born in 1891), a follower of the Russian formalists in
Czechoslovakia who developed their theories more coherently [...] formulated
the theory very clearly: “A work of art will appear as positive value when it
regroups the structure of the preceding period, it will appear as a negative
value if it takes over the structure without changing it”. [...] In literary history
there is only one criterion of interest: the degree of novelty.
Two things may be added. Firstly, that the view of the formalists
emphasizes the intellect and the role of sober-minded reflection. Secondly,
that it gives an explanation why artistic styles are worn out and have to
be replaced when they themselves have become routine: If it is the style
itself and not an irreconcilably contradictory tension between form and
contents in the particular work (Bach!) that is supposed to crush standard
expectations, a work cannot fulfil this role once its style has itself developed
into a standard expectation.
Another example from the world of music may illustrate the point (and,
at the same time, show its shortcomings). In the 1950s and early 1960s,
Karlheinz Stockhausen and Pierre Boulez were recognized on a par as
leading members of the Darmstadt School of serial music. As long as this
extrapolation and automatization of Schönberg’s dodecaphonic principles
was unfamiliar, one (I!) would indeed find that the compositions of the

173
two were equally forceful as sources for “fresh hearing”. As time went
on and the “conventions” became familiar, Boulez’s music stopped
triggering intense attention, and it came to be vaguely boring; Stockhausen
conserved his interest though now, one might say, as music and not as an
enhancement of acoustical awareness200.
The formalist theory may also be applied to kinds of art (and quasi-art)
which do not fulfil its requirements: trivial art, but perhaps in particular
sub-genres like pornography, horror and violence (in the following I shall
speak in a generalized sense of these and their kin as “pornographic”
genre’s). They draw on what is already there in the receiver’s mind,
preferably on strong dreams, drives and prejudices, in ways which prevent
rational deliberation and even the mere establishment of qualitatively new
cognitive relations. The customary in customary amalgamation, in
particularly that which is already so strongly felt that it is easily drawn
upon and thus relatively resistent to reason and to integration in higher
synthesis, is enhanced and stabilized201.

Form versus contents

In order to better understand the implications of the formalist view

we may return to Chapter II – more precisely to Piaget’s concept of schemes
and to the dialectical relation between assimilation and accommodation.

200
These observations and impressions are mine and of course subjective; but they
seem not to be quite private, since Boulez concentrated on a conductor’s career
while Stockhausen has remained a productive composer. The young Danish serial
composer Thomas Koppel, to whom the serial technique was so much of an
automatic and calculated technique that he made his compositions while listening
to rock music, quit serial music altogether and became a rock musician.
201
Drawing preferentially on what is already there as strong dreams, drives and
phobias, often to the exclusion of rational deliberation, also characterizes expression-
ist currents. But expressionist art (at its best) aims at creating new cognitive
relations, and therefore draws less on preestablished prejudice and on the customary
and the customary blend than the pornographic genres.
But the difference may be subtle. The leap is short from certain kinds of
expressionism to very efficient propaganda (another “pornographic” genre) – for
instance from the use of the rats in Murnau’s film Nosferatu to certain scenes in
the Nazi movie The Eternal Jew.

174
 The point of this dialectic was that actual knowledge about something
(the contents of our cognition) is always organized in a specific form, the
scheme and that cognitive structure in which the scheme participates.
Formalism carries its name because it sees the form of cognition as the aim
of art202, not the creation of supplementary contents (who, in fact, is
interested in knowing about the temperature in Malinowski’s garden some
forgotten day in March?).
Yet even if the Piagetian framework makes the central principle of
formalism more clear, it also highlights its failings and problematic features:
A form, firstly, can no more be separated from contents than contents
from form. You cannot, as Jean-Luc Godard parodically makes Ferdinand
propose in the movie Pierrot le Fou, write a novel on “la vie simplement”.
Novels telling about life must necessarily deal with “la vie des hommes”,
i.e., must necessarily be that novel which Ferdinand refuses to write.
Piagetian schemes, in the same way, are only built up as generalizations
of actions or interiorized actions, i.e., as the form of something. If poetry
is not allowed to deal with “morally important real life issues” it will have
to treat of unimportant issues. Or, as stated by Kandinsky, innovator of
artistic form if anybody ever was, in a strong attack on marketable art for
art’s sake: “Form without content is not a hand but an empty glove filled
with air. An artist loves form passionately just as he loves his tools or the
smell of turpentine, because they are powerful means in the service of
content”203.
Specifically, formalism possesses no instrument allowing us to
understand the function of Gelsted’s and Bødker’s poems. These poems,
indeed, do not force upon us a new way in which to see the use of words

202
Of course, one need not have read Piaget to use these terms, which do not even
correspond to Piaget’s own way to express himself.
One may also notice that the term “form” is used in aesthetics in various senses,
not all of which are relevant for a discussion of “formalism”. A useful discussion
is [Tatarkiewicz 1973], which however tends to distinguish mechanically (between
“Form [concept] A”, “Form B”, ..., “Form G”) rather than seeing the connections
between the different meanings, and which also includes notions which might have
been called “form” but which are actually labelled differently.
203
“L’arte d’oggi è più viva che mai”, in [Kandinsky 1976: II, 190]; I quote Tatar-
kiewicz’s translation [1973: I, 221].

175
and phrases – their artistic form is quite conventional. Instead they convey
a new understanding of the world (the nature of death, the reality behind
the myth of “free initiative”). Formalism is unable to explain why these
insights are better formulated as (traditionalist) poems than in thesis form.
More generally, formalism can be claimed to see innovation in artistic
and in cognitive form as identical. Even though this may be warranted in
many cases, it is no universal truth. In fact, becoming familiar with the
rapidly changing artistic forms of the twentieth century is largely an
assimilative process: “Oh, music may also sound like this” / “Indeed, this
is also an impressive painting”. As any assimilation, this involves accommo-
dation, but mainly of our concepts of musical or visual beauty. Our interest
in music, however, is not explainable through the observation that music
changes our understanding of music.
How are we then to understand the interest of art which does not live
up to the formalist requirements? And how are we to understand the actual
effect of trivial literature, pornography, etc.? We need an understanding
of art which does not merely discard them but on the contrary understands
the (empirically) different effects of “real art” and “trivial art”.

Gelsted and Epicurus

The Epicurean maxim expresses a chain of purely analytical discourse.

A familiar phenomenon is investigated stepwise: What does it mean to
fear death? What does it mean to fear anything? Since this peculiar “thing”
is something which I cannot perceive, I cannot fear it.
It is easy to loose the thread of a complex analytical argument, which
deprives it of its power to convince. Epicurus’ argument has therefore been
summed up in an aphorism, “Why fear death? When I am here, death is
absent. And when death is here, I am absent”. This pulls together the main
lines of the argument in a way which allows you to apprehend them in
one glance. One reason that the aphorism can be grasped in this way is
the doubly symmetric structure of the argument (presence/absence, I/
death), which, firstly, binds the two clauses constituting each of the last
two periods together, and, secondly, joins these two periods; another reason
that the aphorism is grasped and remembered is the artistic-humorous

176
form. None the less, even the aphorism does not convince if one is scared
of death.
Gelsted’s poem is a protest against Epicurus’ acceptance of death. Stated
analytically, the argument might run something like this: What I cannot
concede to death is that it reduces me to pure absence. I can only bear my
existence if I am able to ascribe to it a meaning, if beauty is real, if suffering
is real, etc. The fact that I shall die, that everything which in this moment
is beautiful or bitter matters nothing under the perspective of Eternity, is
unacceptable to me; what I fear is this fact of ultimate meaninglessness
of the present and not an abstract future “thing” called death.
Yet instead of presenting this analytical string, which will easily be as
existentially non-committal as Epicurus’ argument (although, I discover,
I actually put it into words and phrases which are more heavily loaded
than a purely analytical argument should be), the poem reminds the readers
of what existence is, and thus suggests how absence can be grasped. It draws
upon the connotations of words and upon what readers can be supposed
to know about the experience of spring under newly green beeches, upon
their knowledge that “spring clouds drifting in the blue of heaven” imply
sunshine, fragrance and breeze but no storm; as a climax it draws upon
the readers’ own experience of suffering, and on the extra pain added to
suffering if it is recognized to be meaningless.
All this is not put into any logical or analytical framework. The
coherence of the poem is rhetorical and rhythmic, using the contrasts
“nevermore/the beeches’ cool shelter” and “spring clouds/nevermore”
(a double contrast which keeps the two first stanzas together, enhancing
their weight in the argument); then comes a middle part, strung together
by “oblivion”/“forgotten” (glemsel/glemt) and leading to a final climax
produced (at the rhetorical level) by a triple nothing (evidently, this
structural analysis could be expanded). Just as the text draws upon the
readers’ total understanding of what it is to live in order to make it clear
what it is to die, it is left to the readers to take bearing of the rhythmic and
rhetorical structure in order to build up an ordered totality, an implicit
synthetical argument. And it is, indeed, an ordered totality which is built
up.
The words of the text do not serve in a sharply defined sense, as they

177
would (ideally) do in a technical manual, where the role of the reader is
reduced to understanding (or, perhaps, not understanding) the terms
correctly. The words carry their whole, open-ended load of connotations,
and should do so. By means of its rhythmic and rhetorical structure,
however, the text puts these bundles of connotations into mutual relations
which readers would not automatically produce on their own. Assimilating
these relations to their own understanding, readers will accomplish an
accommodation, whereas the technical manual only gives information about
familiar entities, i.e., assimilation relatively free of accommodation. The
manual, if well written, can be used. A work of art cannot: “using” it
involves one in co-producing.
This is a general characteristic of the artistic product, and explains why
a work of art is not exhausted by one reading (or whatever kind of
reception is involved)204. On one hand, assimilation normally presupposes
repeated experience. On the other, one and the same work will be seen
differently from different readers’ perspectives, and by the same reader
in different moments. This is a simple consequence of the open-ended and
non-overlapping ranges of connotations (not only of words but also of
rhythmic and other structures) produced in different readers and in
different situations. Although one interpretation of a work can often be
argued to be “better” than another, i.e., to make better sense of more
features of the work205, it is not possible to translate a work into a single

204
Since even the forms of artistic expression carry a load of implied meanings and
connotations derived from their use it also explains why it is impossible to resurrect
the styles and genres of former times. As expressed by Kandinsky in Über das
Geistige in der Kunst [1976: II, 69]:
We cannot re-create the sensibility and the internal life of the ancient Greek;
therefore, even if we try for instance to apply the Greek principles in sculpture,
we shall create only shapes which are similar to the Greek ones, but the work
will forever stay without spirit. Such imitation is similar to that of a monkey.
Externally, the movement of the monkey are fully similar to those of a
human.The monkey sits down, takes a book to its nose, turns over the leaves,
seems to be fully absorbed; yet the internal meaning of these movements is
completely absent.
205
This may mean that the better interpretation is also in better agreement with the
intentions of the artist. But this need not be the case. Erich Maria Remarque
intended to write a patriotic novel, and believed to have done so. Only when the

178
definitive interpretation – even the artist will not be able to do so, since
even the artist’s range of connotations is open-ended.
On this account, the difference between trivial art and what we might
call complex art (which is not the same as complicated art – complication
may follow from confusion, but complexity not206) can be seen in a new
perspective. Trivial art does not put things and concepts into unexpected
relations, and therefore does not produce accommodation. Yet what is
utterly familiar for one person may be an unexpected and inciting discovery
seen from another’s perspective. Accommodation of the totally unfamiliar,
furthermore, is not likely to occur (cf. the leaden ball of p. 9). Whether a

manuscript of Im Westen nichts Neues, after having been rejected by some ten
patriotic publishing houses, was accepted by a left-wing publisher and became an
immense success, did he discover that his patriotism was not that of the patriots
but somehow pacifist [Nils Rickelt, personal communication].
206
The issue can be illustrated by Schönberg’s comparison of Bach’ style with that
of his predecessors and contemporaries [1989: 89f]:
What is new in Bach’s art can only be understood if, on one hand, one
compares it with the style of the Netherlandian school and, on the other, with
the art of Händel.
The secrets of the Netherlanders, to which the uninitiated had absolutely
no access, relied upon a complete knowledge of the possible contrapunctual
relations between the seven tones of the diatonic scale. This allowed the initiates
to produce combinations allowing many kinds of vertical and horizontal
transposition and other similar transformations. But the remaining five tones
were not included in these rules, and if they occurred at all then outside the
contrapunctual combinations and as occasional variation.
In contrast Bach, knowing more secrets than possessed by the Nether-
landers, extended the rules in such measure that they included all twelve tones
of the chromatic scale. [...].
If one has observed that the contrapunctual flexibility of Bach’s themes
in all probability depended on his instinctive thinking in multiple counterpoint
giving room for additional voices, and if one compares his counterpoint with
Händel’s, then the counterpoint of the latter appears impoverished and simple,
and his side voices are really mediocre.
Even in other respects is Bach’s art superior to Händel’s. As a composer
for the theatre, Händel was always able to start with a characteristic and often
excellent theme. But then follow, apart from the repetition of the theme, a
decline, [...], empty and trivial, étude-like broken accords. In contrast, even
Bach’s connecting and subordinate passages are always full of character,
invention, fantasy and expression.

179
particular work is “trivial art” or not is hence not merely to be determined
from the work itself; it also depends on reception and on the capacities
and preceding experience of the receiving mind.
On the same account, we may understand the frequent weakness of
didactic art. “Leftist detective stories”, “progressive fairy tales”, and
“morally edifying versions” will all too often be reduced to one level of
meaning. In order to make sure that the reader gets the “right” associations,
the challenge of open-ended connotations is reduced to a minimum. Yet
guiding the mind of the reader so that it performs no “wrong” movements,
and thus barring co-production, prevents it from performing that autono-
mous activity which is a presupposition for accommodation. Preventing
the occurrence of “wrong” new thoughts is tantamount to preventing the
occurrence of new understanding207.

Art as thought experiments

For the next step in our argument we may return to the difference of
opinion between Kuhn and Lakatos. For the latter, the hard core of a
research programme was precisely definable. For Kuhn, the paradigm was
constituted in part by tacit knowledge, skills, and context-defined concepts.
It is not possible to define exhaustively what, e.g., a text structure is, nor

207
Evidently, this is not the only reason that art that is morally or politically engaged
is claimed to be banal or of bad quality. Often it simply means that the critic does
not share and does not appreciate the values which are expressed.
A classical example of politically engaged art which is not banal art is provided
by Eisenstein’s Potemkin. There is no doubt about the political message. This,
however, does not make the movie banal. The political message, indeed, is not
presented in homiletic one-dimensionality, but through a highly complex use of
the pictorial medium and the temporal organization.
As an (unusually transparent) example of a critic disguising political disagree-
ment as art criticism one may cite a commentary to Shostakovitch’s Leningrad
Symphony written by Clive Bennett for a record edition (Decca D213D 2). If, as the
composer’s notes tell, the symphony is inspired by the war during which it was
written, and if it is meant as a requiem for the victims of the Nazi atrocities, it is
obviously nothing but vulgar propaganda, “a film score without a film”. But “the
symphony, if we believe [the] interpretation [that the inspiration is rather Stalin’s
purges], becomes transformed from a propaganda work into a canvas of universal
significance”.

180
to prescribe exactly a universal method for finding it. What you acquire
when learning textual analysis is unsharply defined knowledge and skill
in analyzing. Your experience within the field will be associatively
connected, like Gelsted’s “tree”, “spring clouds” and “suffering”.
As it was argued, this difference is what makes Kuhn’s approach a more
adequate description of the actual scientific process: Scientific practice is
to a considerable extent based upon intuitive knowledge, knowledge
organized in totality and in analogies.
This is the reason for the importance for the thought experiment. A
prototype is the argument against the rotation of the Earth which was
raised by the Paris philosopher Buridan in the fourteenth century: as we
all know, the stars of heaven seem to circle around the Earth once every
24 hours. Wouldn’t it be more economical if the Creator had made the
Earth rotate and had left the immense sphere of stars at rest? Perhaps, but
hardly the case, Buridan explains. Reflect upon what happens if you shoot
an arrow vertically upwards. It will fall down upon you own head. But
if the Earth rotated, you would have moved a considerable distance (some
1500 feet per second) while the arrow was in the air, and it would fall to
the ground far from where you are. This is obviously not the case.
According to later physics, Buridan’s intuition is wrong. He is unaware
of the law of inertia, according to which an arrow shot from a horizontally
moving bow will receive a horizontal component of movement. This,
however, is only important in so far as it shows that intuitive knowledge
need not be correct. More central is the observation that scientific argumen-
tation presupposes a certain measure of global knowledge (correct or
incorrect) about the behaviour of its object.
Works of art can also be regarded as thought experiments. In some
cases this is obvious. If we consider a novel like George Orwell’s 1984, what
makes the book influence us is the fact that it is psychologically plausible:
The world which is depicted could be a world, with all the complexities
of a real world, for all we know about social life and human beings.
Evidently, “all we know” is a historical product (as was Buridan’s
knowledge); in the moment when our intuitive knowledge makes the world

181
of the novel seem implausible, the novel itself will loose its actuality208.
Other works of art are only thought experiments in a transferred sense.
A piece of pure music is neither plausible nor implausible with regard to
our experienced daily world. Its “plausibility” depends on its inner
coherence. But even a piece of pure music is a testing of possibilities and
consequences within a space of plausible solutions (a form or style); the
“transferred sense” thus is a sense.
The reason that Buridan needed his thought experiment (and that Niels
Bohr and Einstein needed theirs!) is that analytical thought does not exhaust
everything we know. Thought experiments allow us to gauge what we
can formulate analytically against what we know tacitly (i.e., to check those
pieces of knowledge which we can isolate against the totality of the world
as we know it). Similarly, the genuine thought experiments of art allow
us to gauge specifiable moral (etc.) convictions, to see whether they will
work acceptably in a specific situation created (fictionally) for that purpose.
In this way, Shakespeare demonstrates in Romeo and Juliet that the morality
of the family feud and honour are unacceptable. The thought experiment
of the tragedy leads to better knowledge about how we should live, to
superior practical knowledge. That the questions explored by art are (as
all questions about practical knowledge) normally much less accessible
to explicit analysis than those investigated by the sciences explains why
thought experiments play a much more central role in art than in scientific
discourse.

208
When explaining in De poetica (1451a36f: trans. [Bywater 1924]) the difference
between the mimesis of the artist and that of the descriptive historian, Aristotle
states that “the poet’s function is to describe, not the thing that has happened, but
a kind of thing that might happen, i.e. what is possible as being probable or
necessary”. And later (1454a35ff), “whenever such-and-such a personage says or
does such-and-such a thing, it shall be the necessary or probable outcome of his
character; and whenever this incident follows on that, it shall be either the necessary
or the probable outcome of it”. Again (1460a26), “a likely impossibility is always
preferable to an unconvincing possibility” – viz because a work of art does not
function if its world appears to us as an implausible postulate.

182
“Realism”
In mechanical interpretation, what was just said about Romeo and Juliet
might be taken as an argument in favour of realism. Before we go on this
term has to be explained. “Realism” as I use it here is not identical with
“naturalism” – it may be its opposite. The pair realism/naturalism is rather
a transposition into aesthetics of the epistemological pair realism/nominal-
ism. Nor is “realism” meant here as the antithesis of “embellishment”
(which it sometimes is) – actually, the two concepts are close
neighbours209.
Positively stated, realism (or aesthetic realism as I shall call it in the
following in order to avoid misunderstandings, and since we are dealing
with art) is the view that a work of art should lead to understanding of
the essence of things. It may do so by depicting phenomena naturalistically,
but whether it does so or not, the important thing is that it should lead
to insight in something more fundamental than these phenomena; in this
sense, aesthetic “realism” is akin to philosophical realism – whether
“strong”, objectively idealist as Plato’s variant, or “moderate” as Aristotle’s.
An example of definitely non-naturalist realism is Futurism. Futurism
did not use the realist label; but the label “Futurism” is itself a claim that
art should tell the essence of the new world: speed, aggressiveness and
fight, breakdown of classical harmony. Surrealism (sometimes non-naturalist
as with Max Ernst, sometimes deceitfully naturalist as with Dali), often
inspired by psychoanalysis, aimed at showing the higher reality of the mind
as uncensored by reason.
Within a philosophical context which itself is clearly different from
classical (Platonic or Aristotelian) realism and its concepts of “truth” and

209
The reason that these conceptual fences have to be erected is of course that the
word is used in so many different senses. Taken as a synonym of naturalism,
however, the term is superfluous; and taken as a token for the view that art should
show the world as ugly or as cruel as it is instead of postulating a harmony which
isn’t to be found in reality, the term becomes epistemologically uninteresting (its
relevance in political discussions of the responsibility of art and artists in a specific
historical context notwithstanding).
What I try to do here is to specify a sense which underlies some of the views
on art which proclaim themselves “realist” (not least “Socialist Realism”), and a
number of others which use different banners.

183
“reality”210, a suggestion of aesthetic realism can still be found in
Heidegger’s “Ursprung des Kunstwerkes” [1977/1936]. “Instating itself
in work [of art], truth [which is ‘the truth of being’] appears” (p. 69). Van
Gogh’s painting of a pair of peasant’s shoes uncovers not only the shoes
as things but the whole of the peasant’s lived experience: the hard toil of
monotonous ploughing, the hostile wind, the loneliness of the work in the
fields, the anxiety for the daily bread, ... (p. 19).

What was said above regarding Romeo and Juliet is a similarly “realist”
interpretation. The tragedy does not tell merely about a particular sequence
of events arousing fear and pity; it also tells what are the real consequences
of the prevailing code of honour and the practice of family feuds – no less
than van Gogh’s shoes as lived experience. Even Aristotle moves on a
comparable level (though with reference to a different overall philosophy)
when he asserts (De poetica 1451b5-7; trans. Bywater 1924) that “poetry is
something more philosophic and of graver import than history, since its
statements are of the nature rather of universals, whereas those of history
are singulars”. In spite of what it is tempting to read into the term mimesis,
Aristotle’s ideal for art is thus no naturalistic imitation of phenomena (cf.
also the quotations in note 208).
Both Heidegger’s and Aristotle’s aesthetic realism is cautious and
unpretentious in the sense that none of them is coupled to a statement of
what the underlying truth revealed or displayed by the work of art should
be. What one might call extreme aesthetic realisms do not share this
restraint – certainly not that Socialist Realism which was proclaimed as
a programme at the Congress of Soviet Writers in 1934, and which may
be the most outstanding example of a declared aesthetic realism from this

210
Heidegger [1977/1936: 14f] indeed points out that the dichotomy matter/form
is derived from tools created with a certain purpose; and concludes that seeing
everything that is as the result of an imposition of form on matter presupposes that
it is the outcome of an act of creation similar to the one by which men create their
tools – and thus in the final instance, and however much theologians and
philosophers try to deny the parallel – faith.

184
century211. It was coupled to a version of Marxism which already knew
not only the past and the present essential conditions of class struggle but
also its inescapable outcome212; art was therefore to reflect the movement
of (this) history past and future and to lay bare how particular situations
were explainable in terms of the general laws of history.
The way this programme was coupled for a while to political power
falls outside the scope of a discussion of the relation between art and
cognition. What falls inside is the observation that the programme, if
followed to the letter, makes the expression in artistic form superfluous213.
If the essence of things is already known so precisely that it can be
translated into prescriptions, then it can no less easily be explained as
theory, and art which follows the prescriptions becomes theory (and
ultimately ineffective as art, cf. the above observations on didactic art); what
can, and what needs to be expressed as art is open-ended knowledge.
Aesthetic realism (not least this realism) is often, and justly, seen as
antithetical to formalism. The reason is not that realist art does not care

211
The programme of Socialist Realism is certainly not the only example of political
control of art from the twentieth century. Other instances of this phenomenon,
however, have not been supported by a similarly elaborate aesthetic philosophy.
212
This (Stalinist) version of Marxism was thus in itself (like all the brands of
Marxism which subscribe to strict economic determinism or to a closed Hegelian
dialectic where “history has been but is no more”) a brand of extreme philosophical
realism or objective idealism – as caustically pointed out in Sartre’s Questions de
méthode – cf. [Høyrup 1993: 206].
213
This “letter”, it should be said to do justice to the better theoreticians of the
movement (e.g., Georg Lukács), was not theirs explicitly – they had too much
pragmatic sense and artistic feeling to reject everything which was valuable art,
and could avoid doing so by moving imperceptibly between “realism” as here
understood and “naturalism”. What I do in these pages is to draw some con-
sequences which the fathers of Socialist Realism had too much insight to draw,
but which are none the less inherent in their ideas, as demonstrated in not a few
of the works that came out of the programme. Lukács himself [1969/1948: 78]
notices their “monotony [...]. One has barely begun the reading of most of these
novels before he knows everything that is going to happen: in a factory, vermin
is at work; everything is chaotic, but finally the Party cell or the GPU discovers
the nest of wrongdoers, and production flourishes [...]”. As far as predictability
is concerned no worse than much trivial art – but certainly not what a philosopher
of art whose ideals were Tolstoy and Balzac would like to sponsor.

185
about form. On the contrary: that Madame Bovary kills herself by
swallowing arsenic does not in itself tell anything about the obtuseness
of provincial bourgeois society; if Flaubert’s novel manages to relate this
“essential truth” about bourgeois life it is through the way it selects, orders
and tells its material – i.e., by means of the form of the novel. Aesthetic
realisms (and not just this realism) are passionately absorbed in putting
art into the correct form, that which corresponds to their assumed under-
lying order of reality. What distinguishes realism from formalism is not
the degree of absorption in the question of form; it is that formalism rejects
the idea that any particular form should be correct and hence definitive,
arguing instead that every artistic form is wrong when it has become
customary or trivialized.

Synthetical understanding and practical knowledge

The experience of artistic thought experiments and the analysis of
Gelsted’s poem suggested that the formalist understanding was insufficient,
however much its “realist” counterpart exaggerates its own point. It
remains, however, that a large class of artistic works refer to nothing
outside themselves and the stylistic canon to which they relate, neither
as thought experiments creating a world in agreement with our tacit
knowledge and scrutinizing it, nor with connotations derived from our
“lived experience” of spring clouds and suffering. If we are not satisfied
by the formalist explanation that they allow us to see with fresh eyes (and
as we have seen, there are reasons not to stop at that point), what are we
then to do about works that (like Bach’s Musical Offering) are only thought
experiments in a transferred sense?
Clearly they cannot gauge the validity of moral convictions etc., since
such convictions have neither presence nor representative in the work.
What they do to us when we put their “plausibility” on trial – in terms
of inner coherence and of the tension between form and material, stylistic
canon and actual use – is rather to sharpen our ability to comprehend
totalities, to perceive intuitively214. For instance: If we have become able

214
The following line of reasoning was originally inspired by [Feinberg 1977]. As
it unfolded, it developed a certain (unforeseen) affinity with Kant’s position in

186
to grasp (consciously or, just as well, subconsciously) the structure of
Beethoven’s piano sonatas, we shall have no difficulty in following the
implicit prescriptions given by the rhythmic and rhetorical structure of
Gelsted’s poem; and we shall also have enhanced our chances to grasp
a structure in complex real situations.
– Certainly not the structure: real-life situations are, no less than works
of art with their indefinite range of connotations and resonances, infinitely
complex, and we shall find no bottom if we dive into them; realist aesthetic
theory fails on both accounts. But precisely therefore training in grasping
as much and as essentially as possible and concluding from that is crucial. In
Brecht’s words215:

Was hilft zweifeln können dem What help is doubting for the one
der sich nicht entschließen kann! who is unable to conclude!
Falsch mag handeln Incorrectly may act the one
wer sich mit zu wenigen Gründen who is satisfied with too few argu-
begnügt ments
aber untätig bleibt in der Gefahr but unfit in danger remains the one
wer zu viele braucht who needs too many

Because no description of reality (no scientific description, nor certainly

any other) can be transcendentally and exhaustively true (i.e., identical with
what it describes), translating one’s apprehension of a situation into a
formal system from which a conclusion can be drawn always involves a
moment of synthetical judgment which integrates our analytical and tacit
knowledge but cannot itself be argued exhaustively in analytical argument.
Under which moral rule a certain act is to be counted, for instance, is itself
not to be derived from rules alone (cf. above, p. 125). Alf Ross, the Danish
legal philosopher, gave the strong formulation to this observation that the

Kritik der Urteilskraft – cf. below.

215
“Lob des Zweifels”, [Gesammelte Werke, vol. 9, 626-628]. The poem as a whole
is a recommendable treatise on epistemology in practice.

187
leap from legal premisses to action is irrational216. Knowing how to make
this leap (and all the analogous leaps) is practical knowledge.
We may conclude that: if skill in logical inference and analytical thought
is of any use in human life, then only if coupled to practical knowledge, to a
corresponding skill in synthetical or intuitive judgment. We may then continue
an argument begun on p. 30. There, the fundamental cognitive categories
were suggested – like the conserved object – to allow more adequate action,
and to have developed biologically for that reason. This, we now see, they
could only do if they developed together with a faculty for synthetical or
intuitive judgment.
Art, we have also seen, trains that faculty, and allows its development
into a genuine skill. Thereby, what looked initially as a side track returns
us to our main theme. As it was argued, scientific practice is largely
governed by paradigmatic knowledge, as social life by norm systems. Both
contain, and presuppose for their application, the wielding of intuitive and
integrative knowledge. Art, by training the faculty for integrative thought,
is thereby also a training of the very basis of scientific practice and social
life. By being less bound than real-life moral reflection and scientific work
to a specific content, it may provide a more thorough and comprehensive
training than the two activities provide occasion for themselves217.

216
Cf. [Jarvad 1993: 94].Evidently, the idea of rationality underlying the argument
is what was spoken of above (p. 3) as the “absolutist ideal of the strict proof”, not
the open rationality of dialogue.
Elsewhere (p. 43), Jarvad sums up Ross’s view in the words that “theories and
analyses do not lead to decisions, do not designate certain decisions as correct and
others as wrong, at most they pinpoint errors in the basis for the decision. Science
does not lead to decision and action”.
217
This point comes close to Kant’s Kritik der Urteilskraft. That faculty of judgment,
indeed, which we need in order to decide whether a particular act falls under a
specific rule and, in general, to perceive the particular as a case of the general (cf.
note 139); a faculty which is hence a cornerstone of theoretical as well as moral
reflection; that faculty is exerted in its purest form as judgment of art, and it is in
the domain of aesthetic judgment that its particular character is exposed. This
character Kant describes (A VII, B VII) technically as “want of a principle”, where
theoretical thought has the concept of nature (involving “the nature of things”,
regularity, causality, predictability etc.) as its principle and moral thought the concept
of freedom, i.e., in terms that correspond to his conception of theoretical and practical
reason. As argued forcefully by Garroni [1992], however, what is involved is the

188
 Paraphrasing what Gouldner says about the value-free doctrine (cf.
p. 145), the free activity of art may enable us “to make better value
judgements rather than none” – “better” in the sense of less parochial,
better since based on a broader understanding of our total situation and
the implications of our actions. This may be a first step in a justification
(and, at the same time, a critique) of the idea that understanding of art
leads to moral improvement, and that aesthetic education is the best moral
education.
From a slightly different point of view the discussion is connected to
the role of metaphors in scientific thinking (of which a particular instance
was discussed in note 48); understanding of such metaphors – which serve
the construction of general concepts from particulars – relies on the same
faculty as the understanding of the metaphors of art: no reasons of principle
would prevent the Cecil Bødker’s sunbathing grandson from serving as
the starting point for a theory about Torstein Veblen’s “leisure class”. Yet
once again, the understanding of art may provide more abundant oppor-
tunity than theory to exert – whence train – that faculty.
Formulated in less aphoristic terms: Apart from its function in concrete
intuitive reasoning and as a source for better practical knowledge, as
discussed inter alia in connection with Gelsted’s poem and the role of art
as thought experiment, art (and here, abstract music no less than figurative
painting or texts with a meaning) may also have a fundamental cognitive
function as training of that integrative or synthesizing competence without which
the best analytical abilities are empty – as training of the very faculty for
practical knowledge218.
This sounds like a conclusion to the question concerning art and
cognition, and in a way it is. Yet it is important to remember that these
functions of art as a way to practical knowledge and as training of practical
cognition per se do not answer (or at least do not exhaust) the question

role of the faculty of judgment as a general guide to the intellect in its wielding
of conceptual thought.
218
This fits an observation made on p. 167, viz that art must be grasped “by that
kind of intellect which makes use of the senses of unified space”. This is the kind
of intellect which needs to be trained if analysis and synthesis shall work together.

189
why we engage in art, as producers or as co-producers. The phenomeno-
logy of beauty – as told by two illustrious witnesses – may serve once more.
In his Confessions (X.xxx-xxxv – ed. trans. [Trabucco 1960: 123-147]),
St. Augustine aligns indulgence in the pleasures of the senses with erotic
concupiscence (and with interest in natural philosophy characterized as
“vain curiosity”). Pleasant smells are no problem for him (they produce
no beauty, we might say), but all the more is music. Though he has
improved, he still cannot ignore the musical beauty of a psalm melody
sung by a beautiful voice or avoid rejoicing in it – and no better is seeing
the sweetness of the world. Aristotle, when locating the essential beauty
of the tragedy, points to its arousal of “pity and fear” (De poetica 1453b12).
Less immediately than the “pornographic” genres but no less truly, all
impressions of beauty – and all engagement in artistic production or co-
production – appear to be rooted in our affections. How it relates to them
is a major question if we wish to understand what art is, and one might
postulate that the experience of “beauty” results from some kind of unity
of affection and integrative insight; but the problem does not belong within
the present line of argumentation, and we shall leave the argument here.
With respect to its root in affections, of course, engagement in art only
differs by degree and degree of immediacy and perhaps in the kind of affections
involved from other kinds of human conscious activity. No such activity
is undertaken without a motive, and motives are by definition rooted in
the affective. The sphere of art may therefore be less absolutely separable
from other spheres of life than presupposed for convenience in the
preceding pages. Still, if large enough, differences in degree remain
decisive, and art, if no absolutely separable sphere, remains a sphere of
its own no less than science and morality.

190
 IX. REFERENCES AND BIBLIOGRAPHY

Anderson, John M., 1971. The Grammar of Case. Towards a Localistic Theory. Cam-
bridge: Cambridge University Press.
Aristotle, Works. Translated into English under the Editorship of W.D. Ross. 12 vols.
Oxford: The Clarendon Press, 1908-1952.
Bacon, Francis, 1937. Essays. London: Oxford University Press.
Barber, William J., 1967. A History of Economic Thought. Harmondsworth, Middlesex:
Penguin.
Barnes, Barry, 1979. “Thomas Kuhn”, in Q. Skinner (ed.) 1979: 83–100.
Beardsley, Monroe C., 1973. “Theories of Beauty since the Mid-Nineteenth Century”,
in Dictionary of the History of Ideas I, 207-214.
Bickerton, Derek, 1983. “Creole Languages”. Scientific American 249:1, 108-115
(European edition).
Bloom, Alfred H., 1979. “The Impact of Chinese Linguistic Structure on Cognitive
Style”. Current Anthropology 20, 585-586.
Bødker, Cecil, 1964. Samlede digte. København: Hasselbalch.
Böhme, Gernot, Wolfgang van den Daele & Wolfgang Krohn, 1973. “Die Finali-
sierung der Wissenschaft”. Zeitschrift für Soziologie 2, 128-144.
Brecht, Bertolt, 1967. Gesammelte Werke. 20 vols. Frankfurt a.M.: Suhrkamp.
Brinkmann, Richard (ed.), 1969. Begriffsbestimmung des literarischen Realismus. (Wege
der Forschung, CCXII). Darmstadt: Wissenschaftliche Buchgesellschaft.
Broad, William, & Nicholas Wade, 1982. Betrayers of the Truth. London: Century
Publishing.
Burling, Robbins, 1993. “Primate Calls, Human Language, and Nonverbal
Communication”. Current Anthropology 34, 25-37, discussion 37-53, 165-170.
Bywater, Ingram (ed., trans.), 1924. Aristotle, De poetica, in Aristotle, Works, vol.
XI.
Celan, Paul, 1975. Gedichte. 2 vols. Frankfurt a.M.: Suhrkamp.
Casini, P., 1988. "Newton’s Principia and the Philosophers of the Enlightenment".
Notes and Records of the Royal Society of London 42, 35–52.
Chamisso, Adalbert von. Werke. 4 vols. Leipzig: Max Hesse, n.d.

191
Channell, David F., 1982. “The Harmony of Theory and Practice: The Engineering
Science of W. J. M. Rankine”. Technology and Culture 23, 39-52.
Chapanis, Alphonse, 1968. “Engineering Psychology”, pp. 81-87 in International
Encyclopedia of the Social Sciences 5. New York: Macmillan and The Free Press.
Cole, Michael, & Sylvia Scribner, 1974. Culture and Thought. A Psychological
Introduction. New York: Wiley.
Crombie, Alistair C. (ed.), 1963. Scientific Change. London: Heinemann.
Dictionary of the History of Ideas. Studies in Selected Pivotal Ideas. 5 vols. New York:
Scribner, 1968-74.
Diderichsen, Paul, 1971. Elementær dansk grammatik. 3. udg., 5. oplag. København:
Gyldendal.
Dieckmann, Herbert, 1973. “Theories of Beauty to the Mid-Nineteenth Century”,
in Dictionary of the History of Ideas I, 195-206.
Ember, Carol R., & Melvin Ember, 1977. Cultural Anthropology. 2nd edition.
Englewood Cliffs, New Jersey: Prentice-Hall.
Einstein, Albert, 1921. “Geometrie und Erfahrung”. Offprint from Sitzungsberichte
der Preussischen Akademie der Wissenschaften 5.
Feinberg, Evgeny L., 1977. “Art and Cognition”. Soviet Studies in Philosophy 15:4,
62-91. Russian original: Voprosy filosofij 1976 no. 7.
Feyerabend, Paul, 1975. Against Method. London: New Left Books.
Foucault, Michel, 1966. Les Mots et les choses. Une archéologie des sciences humaines.
Paris: Gallimard.
Garroni, Emilio, 1992. Estetica. Uno sguardo-attraverso. (Strumenti di studio). Milano:
Garzanti.
Gelsted, Otto, 1957. Udvalgte digte. København: Gyldendal.
Giddens, Anthony, 1976. New Rules of Sociological Method. London: Hutchinson.
Gouldner, Alvin W., 1973. For Sociology. Renewal and Critique in Sociology Today.
London: Allen Lane.
Grubb, G., & E. Kristensen, 1960. “First Draft of a Summary of the Discussions
Following the Lectures”. ICMI-Seminar, University of Aarhus, May 30 to June
2, 1960 Mimeo, Mathematical Institute, University of Aarhus.
Gustafsson, Bo, 1968. “Klassicism, marxism och marginalism”. Häften för Kritiska
Studier 1:1-2, 3-16.
Hardie, R. P., & R. K. Gaye (eds, trans.), 1930. Aristotle, Physica, in Aristotle, Works,
vol. II.
Hardy, Godfrey Harold, 1967. A Mathematician’s Apology. With a Foreword by C.
P. Snow. Cambridge: Cambridge University Press.
Heidegger, Martin, 1977/1936. “Ursprung des Kunstwerkes”, pp. 1-74 in Heidegger,
Holzwege (Gesamtausgabe, 5). Frankfurt a.M.: Klostermann.
Hicks, R. D. (ed., trans.), 1980. Diogenes Laertius, Lives of Eminent Philosophers. 2
vols. (Loeb Classical Library). Cambridge, Mass.: Harvard University Press /
London: Heinemann. 11925.

192
Hirschleifer, J., 1977. “Further Comments on Economics and Anthropology”. Current
Anthropology 18, 133-134.
Høyrup, Jens, 1993. “Institutions, Professions, and Ideas. An Approach to the
Theory of the Humanities through their History and Institutional Settings and
their Implicit Anthropologies”. Filosofi og Videnskabsteori på Roskilde Universitets-
center. 1. Række: Enkeltpublikationer 1993 Nr. 1. Second, slightly revised printing.
Hunt, R. C., & E. Hunt, 1976. “Canal Irrigation and Local Organization” Current
Anthropology 17, 389-398, discussion 398-411, 18 (1977), 116.
Jacobsen, Thorkild, 1988. “Sumerian Grammar Today”. Journal of the American
Oriental Society 108, 123-133.
Jarvad, Ib Martin, 1993. Ret og Stat. Tværvidenskab. København: Christian Ejlers.
Jerison, Harry J., 1973. Evolution of the Brain and Intelligence. New York & London:
Academic Press.
Jerison, Harry J., 1976. “Paleoneurology and the Evolution of the Mind”. Scientific
American 234:1, 90-101 (European pagination).
Kandinsky, Wassily, 1976. Tutti gli scritti. 2 vols. Milano: Feltrinelli.
Kant, Immanuel. Werke. 6 vols. Wiesbaden: Insel Verlag, 1956–1964.
Kinser, S., 1981. “Annaliste Paradigm? The Geohistorical Structuralism of Fernand
Braudel”. American Historical review 86, 63-105.
Kuhn, Thomas S., 1963. “The Function of Dogma in Scientific Research”, in A. C.
Crombie (ed.) 1963: 347-369.
Kuhn, Thomas S., 1970. The Structure of Scientific Revolutions. (International
Encyclopedia of Unified Science, Volume 2, Number 2). 2nd ed. Chicago:
University of Chicago Press. 11962.
Kuhn, Thomas S., 1974. “Reflections on my Critics”, in Lakatos & Musgrave 1974:
231-278.
Kuhn, Thomas S., 1978. Black-body Theory and the Quantum Discontinuity, 1894-1912.
New York: Oxford University Press.
Lakatos, Imre, & Alan Musgrave (eds), 1974. Criticism and the Growth of Knowledge.
Corrected reprint Cambridge: Cambridge University Press. 11970.
Lakatos, Imre, 1974a. “Falsification and the Methodology of Scientific Research
Programmes”, in Lakatos & Musgrave (eds) 1974: 91-196.
Losee, John, 1972. A Historical Introduction to the Philosophy of Science. London:
Oxford University Press.
Lukács, Georg, 1969/1936. “Erzählen oder Beschreiben? Zur Diskussion über
Naturalismus und Formalismus”, in R. Brinkmann (ed.) 1969: 33-85.
Mackay, A. L., 1977. The Harvest of a Quiet Eye. A Selection of Scientific Quotations.
Bristol & London: The Institute of Physics.
Malinowski, Ivan, 1980. Vinterens hjerte. København: Borgen.
McGarry, Daniel D. (ed., trans.), 1971. The Metalogicon of John of Salisbury. A
Twelfth-Century Defence of the Verbal and Logical Arts of the Trivium. Translated
with an Introduction and Notes. Gloucester, Massachusetts: Peter Smith.

193
Measurement of Scientific and Technical Activities. Proposed Standard Practice for
Surveys of Research and Experimental Development. “Frascati Manual”. Third
edition. Paris: OECD, 1976.
Merton, Robert K., 1963. “The Ambivalence of Scientists”. Bulletin of the Johns
Hopkins Hospital 112, 77-97.
Merton, Robert K., 1968. Social Theory and Social Structure. Enlarged Edition. New
York & London: The Free Press.
Merton, Robert K., 1968/1942. “Science and Democratic Social Structure”, in Merton
1968: 604-615. First published in Journal of Legal and Political Sociology 1 (1942).
Mossner, Ernest C. (ed.), 1969. David Hume, A Treatise of Human Nature. Harmonds-
worth, Middlesex: Penguin.
Neuenschwander, Erwin, & Hans-Wilhelm Burmann, 1987. “Die Entwicklung der
Mathematik an der Universität Göttingen”. Georgia Augusta. Nachrichten der
Universität Göttingen, November 1987, 17-28.
Ong, Walter J., 1982. Orality and Literacy. The Technologizing of the World. London
& New York: Methuen.
Pfetsch, Frank R., 1979. “The ‘Finalization’ Debate in Germany: Some Comments
and Explanations”. Social Studies of Science 9, 115-124.
Piaget, Jean, 1950. The Psychology of Intelligence. London: Routledge & Kegan Paul.
Piaget, Jean, 1967. Six Psychological Studies. New York: Random House.
Piaget, Jean, 1972. Psychology and Epistemology. London: Allen Lane.
Piaget, Jean, 1973. The Child and Reality. Problems of Genetic Psychology. New York:
Grossman.
Piaget, Jean, 1976. The Grasp of Consciousness: Action and Concept in the Young Child.
Cambridge, Mass.: Harvard University Press.
Popper, Karl R., 1972. Conjectures and Refutations. The Growth of Scientific Knowledge.
4th, Revised Edition. London: Routledge & Kegan Paul.
Popper, Karl R., 1972a. The Logic of Scientific Discovery. 6th Revised Impression.
London: Hutchinson. 1Wien, 1935.
Popper, Karl R., 1973. Objective Knowledge. An Evolutionary Approach. 2nd Revised
Printing. London: Oxford University Press.
Popper, Karl R., 1974. “Normal Science and its Dangers”, in Lakatos & Musgrave
1974: 51-58.
Price, Derek J. de Solla, 1963. Little Science, Big Science. New York & London:
Columbia University Press.
Propp, Vladimir, 1968. Morphology of the Folktale. 2nd ed. Austin & London:
University of Texas Press.
Propp, Vladimir, 1984. Theory and History of Folklore. Edited, with and Introduction
and Notes, by Anatoly Liberman. (Theory and History of Literature, 5).
Manchester: Manchester University Press.
Quine, Willard Van Orman, 1963. From a Logical Point of View. Logico-Philosophical
Essays. Second Edition, Revised. New York: Harper & Row.

194
Quine, Willard Van Orman, 1963/1951. “Two Dognas of Empiricism”, in Quine
1963: 20-46. First published in Philosophical Review, January 1951.
Robins, R. H., 1971. General Linguistics: An Introductory Survey. 2nd edition. London:
Longman.
Robinson, Joan, & John Eatwell, 1973. An Introduction to Modern Economics. London
etc.: McGrawhill.
Rotman, Brian, 1977. Jean Piaget: Psychologist of the Real. Hassocks, Sussex: Harvester.
Saussure, Ferdinand de, 1972. Cours de linguistique générale. Publié par Charles Bally
et Albert Sechehaye. Paris: Payot. 11922.
Schäfer, Wolf (ed.), 1983. Finalization in Science: The Social Orientation of Scientific
Progress. (Boston Studies in the Philosophy of Science, 77). Dordrecht etc: D.
Reidel.
Schmid, Herman, 1973. “Om forskning för folket”. Nordisk Forum 8, 291-305.
Schmid, Herman, 1981. “Tillämpad forskning som praktik”. Institut for Samfundsøko-
nomi og Planlægning, Roskidle Universitetscenter. Arbejdspapir nr. 16/1981.
Schönberg, Arnold, 1989. Stil und Gedanke. Leipzig: Reclam. 11950.
Selby-Bigge, L. A. (ed.), 1975. David Hume, Enquiries Concerning Human Understand-
ing and Concerning the Principles of Morals. Reprinted from the Posthumous
Edition of 1777 and Edited with Introduction ... . Third Edition with text
Revised and Notes by P. H. Nidditch. Oxford: Oxford University Press.
Shiloh, Ailon, 1975. “Psychological Anthropology: A Case Study in Culture
Blindness”. Current Anthropology 16, 618-620; discussion 17 (1976), 326f, 349f,
554f.
Skinner, Quentin (ed.), 1979. The Return of Grand Theory in the Human Sciences.
Cambridge: Cambridge University Press.
Sklair, Leslie, 1973. Organized Knowledge. St Albans Herts: Paladin.
Solovine, Maurice (ed., trad.), 1965. Epicure, Doctrines et maximes. Paris: Hermann.
Summers, Montague (ed., trans.), 1971. Heinrich Kramer & Jacob Sprenger, Malleus
maleficarum. New York: Dover. 1London: John Rodker, 1928.
Tainter, J. A., 1988. The Collapse of Complex Societies. (New Studies in Archaeology).
Cambridge: Cambridge University Press.
Tatarkiewicz, W., 1973. “Form in the History of Aesthetics”, in Dictionary of the
History of Ideas II, 216-225.
Trabucco, Joseph (ed., trad.), 1960. Saint Augustin, Les Confessions. Paris: Garnier.
Watkins, J. W. N., 1974. “Against ‘Normal Science’”, in Lakatos & Musgrave 1974:
25-37.
Weiss, Peter, 1964. Die Verfolgung und Ermordung Jean Paul Marats dargestellt durch
die Schauspielgruppe des Hospizes zu Charenton unter Anleitung des Herrn de Sade.
Frankfurt a.M.: Suhrkamp.
Wellek, René, 1973. “Evolution of Literature”, in Dictionary of the History of Ideas
II, 169-174.

195
Westman, Robert S., 1980. “The Astronomer’s Role in the Sixteenth Century: A
Preliminary Study”. History of Science 18, 105-147.
Wittgenstein, Ludwig, 1968. Philosophical Investigations. Translated by G. E. M.
Anscombe. Oxford: Basil Blackwell.

196