This article was published in an Elsevier journal.

The attached copy

is furnished to the author for non-commercial research and
education use, including for instruction at the authors institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy

Studies in History
and Philosophy
of Science
Stud. Hist. Phil. Sci. 38 (2007) 543554
www.elsevier.com/locate/shpsa

Falsicationism and the structure of theories: the

PopperKuhn controversy about the rationality of normal science
Jose Dez
Universitie of Barcelona/LOGOS Research Group, 08001 Barcelona, Spain

Abstract
Many controversies within philosophy of science have been attempted to be explained in terms of the metaphilosophical prescription/
description distinction over the goal of philosophy of science. The aim of this paper is to show that the controversy between Popper and
Kuhn about the ir/rationality of Normal Science cannot be fully explained in these terms, not even if we also take the truth/problemsolving distinction over the goal of science into account. It is argued that, to gain full understanding of this controversy, it is necessary to
take into account their dierences regarding a topic apparently not involved in it, namely the structure of scientic theories. The conclusion is that both Popper and Kuhn were descriptive/prescriptive at the same time in their attempt to make the rules of (a specic part
of) scientic practice explicit, yet Kuhn did it better because he had a richer and more accurate idea of what scientic theories are.
Although this work is mainly a piece of history of philosophy of science, it also aims at shedding some light on epistemological issues.
If the conclusion is sound, it also shows how structural aspects of scientic theories may be relevant to the epistemology of science.
2007 Elsevier Ltd. All rights reserved.
Keywords: Falsicationism; Normal science; Rationality; Theories; Thomas Kuhn; Karl Popper

When citing this paper, please use the full journal title Studies in History and Philosophy of Science

Though he is not a na ve falsicationist, Sir Karl may, I

suggest, legitimately be treated as one. (Kuhn, 1970a, p.
14)
This passage is really astonishing. It is exactly like saying: Although he is not a murderer, Sir Karl may, I suggest, legitimately be treated as one. (Popper, 1983
[1956], p. xxxiv)
1. Prescription and description: making rules explicit
The aim of this paper is to show that, contrary to what
has been widely accepted, the controversy between Popper
E-mail address: [email protected]
0039-3681/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.shpsa.2007.06.007

and Kuhn about the ir/rationality of Normal Science cannot be fully explained in terms of the metaphilosophical
prescription/description distinction, not even if we also
take the truth/problem-solving distinction over the goal
of science into account. To gain full understanding of this
controversy, it is necessary to take into account their dierences regarding a topic apparently not involved in it,
namely the structure of scientic theories. I begin with
some introductory remarks about the task of philosophy
of science showing that, in the specic eld relevant to
the controversy, description and prescription are not in
opposition, and that, a fortiori, their alleged opposition
cannot explain crucial episodes of such controversy.
Secondly, I introduce the main textual references of the

Author's personal copy

544

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

controversy and discuss their puzzling character. Third, I

present the main traits of Kuhns analysis of scientic theories in contrast to Poppers traditional axiomatic account.
Finally, I show that, taking into account the dierent
notions of scientic theory they implicitly have in mind,
the puzzling aspects of the controversy vanish and that,
so understood, although Poppers falsicationism has a
clear insight, his criticisms of Kuhns Normal Science are
misguided.
Analytic philosophy of science consists in analysing,
among other things, two kind of entities: scientic constructs (concepts, theories . . .) and scientic practices (testing, explaining. . .). Analysing scientic constructs is
basically a reconstructive/interpretative endeavour, which
has to be evaluated both intrinsically and through the benet of the reconstruction applied to other tasks, mainly the
analysis of scientic practices. For instance: a reconstruction of scientic concepts that does not allow one to say
that there is a(n important) sense in which scientic concepts remain the same throughout the evolution of a single
theory, is worse than another which does; a reconstruction
of scientic theories which does not make room for the distinction between normal and revolutionary changes is
worse than another which does.
Analysing scientic practices consists basically in making its rules explicit. Every practice (argumentation, speaking a language . . .) is constituted by certain rules, and
making these constitutive rules explicit is both descriptive
and prescriptive. For example, a specic grammar makes
the rules of a specic language explicit: it describes its
rules. But rules are constitutively normative.1 Thus,
describing the grammar rules of, say, Spanish is at the same
time prescribing how people must/must not speak, that is,
to distinguish between grammatically well formed (rule following) and grammatically ill formed (non-rule following)
utterances. In analysing practices, description (of the rules
that constitute them) and prescription are not in opposition
but two sides of the same coin.2 This task must be evaluated by historical/empirical data about practice performance, taking paradigmatic examples specially into
account.3 On the other side, the normativity that follows
from making the constitutive rules of a practice explicit,
that is, evaluating whether a particular performance is or
is not in accordance with the rules, is compatible with leaving certain performance alternatives open. That is, this
normativity does not imply a command of every possible
course of action, it simply rules out some courses as non1

rule-following and allows others. The rules of many practices do not determine a unique course of action, but this
does not mean that they are not normative/prescriptive.
All this applies to scientic practices as well. Scientists
perform certain practices like testing, explaining, experimenting, working within a theory, and so on. These practices are constituted by their own rules, and philosophers of
science try to make these rules explicit by analysing the corresponding scientic practices. The analysis must be evaluated in the light of historical/empirical data about scientic
practice, taking paradigmatic examples specially into
account. For instance, if certain analysis of working within
a theory implies that the rst predictive failure amounts to
(epistemically) throwing the whole theory away, then it
enters into conict with paradigmatic examples of good scientic practice like working within Newtonian theory
when, before the discovery of Neptune, predictions about
Uranus trajectory didnt t the data.
As in the other cases, this endeavour of describing the
rules governing scientic practices has normative import
and allows for evaluating some performances as good, that
is, rule-following (e.g. Lavoisiers mercury experiment),
and some others as bad, that is, non-rule-following (e.g.
phlogiston theorists attempt to face Lavoisier experiment
postulating negative mass). This normativity, which
excludes certain courses of action, is, again, compatible
with leaving some others open, that is, it may not uniquely
determine what to do in a specic situation.
Scientic practices, though, have a peculiar character.
They are the paradigm (or at least, one prominent paradigm) of epistemic rationality, of coming to believe in a
rational way. And this does not simply follow from its
being constituted by rules, since there are many other practices that are constituted by rules which are not rational in
the relevant sense, such as exorcism, tarot, chiromancy,
astrology, and the like. Of course one can stipulate a sense
of rational equating it with performing a practice according to its constitutive rules, and maybe it could even be
useful for distinguishing, say, rational astrology practitioners from irrational ones. But this is not the relevant sense
at stake. Scientic practice is paradigmatically rational in a
sense in which other, also regulated, practices are not. And
this specic rationality must rest in the specicity of the
constitutive rules of scientic practice. It is then an additional task for philosophy of science to elucidate the sense
in which the constitutive rules of scientic practice make
such a peculiar practice a paradigm of rationality (although

This applies to any form of convention (cf. Lewis, 1969).

Of course not every description involving the rules of a practice has normative import. One can, for instance, describe what practitioners believe the
rules are, and this has no normative import (although may have psychological/sociological import). The point, to repeat, is that, given a practice
constituted by certain rules, making these constitutive rules explicit has normative import. Since making them explicit is also to describe them, we obtain
that description and prescription are in these cases two sides of the same coin. One could object the use of the word description for referring to the activity
of making rules explicit, precisely because this activity has normative import. But I want to keep it for stressing the analogy between logic or grammar and
(part of) philosophy of science. Needless to say, the distinction description/prescription is perfectly disjoint when making constitutive rules of a practice
explicit is not at stake.
3
A problem: practitioners could, in principle, suer of massive ill performance, therefore, evaluating the analysis is not merely checking whether it ts
the majority. Some complications arise here.
2

Author's personal copy

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

this does not mean that we have to constrain our task of

making the rules explicit with a preconceived notion of
rationality4).
These considerations apply to part of the PopperKuhn
debate about the ir/rationality rationality of normal science
(NS). The thesis I want to defend is that some aspects of
this controversy cannot be fully explained in terms of the
prescriptive/descriptive distinction, nor even plugging in
in addition the truth/problem-solving one, and that such
aspects cannot be fully understood without taking into
account something which apparently was not involved in
the controversy, namely the dierent views on the nature
and synchronic structure of scientic theories they had
implicitly in mind. That is, both Kuhn and Popper attempt
to make the rules of certain scientic practice explicit, both
are descriptive/prescriptive in the aforementioned sense;
but Kuhn does it better (that is, ts better paradigmatic
historical examples), and he does so because he is implicitly
running a better interpretation of the constructs implicitly
involved, namely, scientic theories.5
One could object here to the very idea of this project that,
since Kuhn rejects that scientic practice is constituted by
rules, one cannot see his work as making constitutive rules
explicit and, a fortiori, his controversy with Popper as a disagreement regarding such rules.6 For example:
The determination of shared paradigms is not, however,
the determination of shared rules . . . Anyone who has
attempted to describe or analyze the evolution of a particular scientic tradition will necessarily have sought
accepted principles and rules of this sort. Almost certainly, as the preceding section indicates, he will have
met with at least partial success. But, if his experience
has been at all like my own, he will have found the
search for rules both more dicult and less satisfying
than the search for paradigms . . . As a result, the search
for a body of rules competent to constitute a given normal research tradition becomes a source of continual
and deep frustration. (Kuhn, 1970d [1962], pp. 4344)
Normal science can be determined in part by the direct
inspection of paradigms, a process that is often aided
by but does not depend upon the formulation of rules
and assumptions. Indeed, the existence of a paradigm
need not even imply that any full set of rules exist. (Ibid.,
p. 44)
4

545

Scientists work from models acquired through education and through subsequent exposure to the literature
often without quite knowing or needing to know what
the characteristics have given these models the status
of community paradigms. And because they do so, they
need no full set of rules. Paradigms may be prior to,
more binding, and more complete than any set of rules
for research that could be unequivocally abstracted
from them. (Ibid., p. 46)
Contrary to appearances, I dont think these passages are
in tension with the above characterisation of the analysis of
scientic practice as making the rules of such practice explicit. There are two readings of the word rule involved here.
In the strong reading, it refers to a set of quasi-mechanised
procedures that one can explicitly be taught in practicelearning.7 This is the sense in which the late Wittgenstein
famously denied that to learn a language is to learn a set
of such rules and claimed, instead, that to learn a language
consist better in capturing similarities among groups of paradigmatic examples. This is the sense used by Kuhn in these
passages, who explicitly mentions Wittgensteins move at
the beginning of the chapter. In this sense, according to
Kuhn (and to me) normal science does not reduce to a set
of rules; in normal science the ability to capture similarities
among paradigms/exemplars plays a more important and
essential role. It is in this sense that normal science is not
guided by rules but by paradigms. But this is not the sense
in which I stated above that scientic practice, like any other
practice, is constituted by rules. In this weaker sense, for
instance, trying to apply the same symbolic generalisations
to cases which one nds similar to paradigms, is one of the
rules that constitute normal science, and not reducible to
any set of rules in the rst, stronger sense.8
2. Popper and Kuhn on the ir/rationality of normal science
Which are the aspects of the controversy that are better
understood as caused by their implicit disagreements on
the structure of theories? What might be called the possibility conditions for the rationality of NS. If Popper were
right, NS could not be rational, NS would be irrational in
principle. If Kuhn (and Lakatos) is (are) right, NS may
be perfectly rational (and as a matter of fact,9 it is almost
always rational). The parts of the controversy I deal with
are then conned to these, so to speak, transcendental

Of course one could always try to impose dierent rules, yet this is not to analyse a practice but to change it.
Of course this does not need to coincide with either Popper or Kuhn self-description of their task in general, and of their controversy in particular, since
they may be metaphilosophically misguided (as, if I am right, they partially are).
6
I owe this criticism to an anonymous referee.
7
For instance, to learn the syntax of a formal language, for example the rules of well formed formulae in, say, propositional logic. But there are very few
interesting real life examples, if any, of such rules and practices; maybe chess-moving or following driving regulations, which of course are only part of
the more complex practices playing chess and driving, really real life and not rule-guided in this rst sense.
8
Just as playing chess or driving are constituted by rules in this weak sense which are not reducible to sets of rules in the stronger sense. One might
now object that then the thesis that scientic practice is constituted buy rules is very weak. And it actually is, but not totally empty; it simply emphasises
that scientic practice is a practice and, like any other practice, analysing it consists in making explicit what the practice consists in. If someone objects that
in this weak sense it is better not to use the word rule, I would be happy with some other more adequate term.
9
But, importantly, not in virtue of the denition of NS (see the last section).
5

Author's personal copy

546

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

aspects of the rationality in NS. I wont deal with the conditions for NS to be actually rational, nor with any aspect
of inter-paradigmatic or trans-revolutionary rationality.10
Since this controversy is well known, I shall emphasise
only the most striking passages, beginning with the astonishing two which head this paper:
Though he is not a nave falsicationist, Sir Karl may, I
suggest, legitimately be treated as one. (Kuhn, 1970a, p.
14)
This passage is really astonishing. It is exactly like saying: Although he is not a murderer, Sir Karl may, I suggest, legitimately be treated as one. (Popper, 1983
[1956], p. xxxiv)
and continuing with Poppers complaints about Kuhns
(and Lakatos) incapacity for understanding his ideas:
[Kuhn] attended my William James Lectures at Harvard
in 1950 . . . But it seems clear that he does not fully
remember what happened during these lessons. (Ibid.,
p. xxxii)
Kuhn, early in his career, formed a theory of my views
which became his paradigm of Popper . . . Kuhn formed
this paradigm before he ever read any of my writings . . .
But, as we have learnt from Kuhn, paradigms are not
given up so easily. (Ibid., p. xxxiv)
Lakatos has read almost everything I have written. After
being my pupil, he was my colleague and after my successor at LSE. Yet I have to warn the reader that he
has not understood my theory of science. (Popper,
1974, p. 999)
There must be some noise in the controversy that
explains the diculties for understanding one another. It
is true that Popper said from the very beginning (together
with, and besides, his conventionalism on observational
statements) that theories cannot conclusively be proved
false, that there always are possible strategies for eluding
the refutation, etc. But at the same time he says things like:
scientic theories are inventions to be eliminated if they
clashed with observations. (Popper, 1963, p. 61)
the falsity of a theory can be inferred from empirical evidence, and this inference is a purely deductive one.
(Ibid., p. 72)
it must be possible for an empirical scientic system to
be refuted by experience. (Popper, 1958 [1934], p. 18)
I am going to propose that the empirical method shall be
characterized as a method that excludes precisely those
ways of evading falsication which, as my imaginary
critic rightly insists, are logically possible. (Ibid., p. 20)

the uncertainty of refutation (which I pointed out) must

not be taken too seriously (as I have pointed out too).
(Popper, 1983, Introduction to 1982 edition 1)
Etcetera. There is something that makes Poppers statements and qualications hardly coherent, or in need of further articulation, to say the least. He says he is not a nave
falsicationist, but what is he then?
Sir Karl is not, of course, a naive falsicationist. He
knows all that has just been said and has emphasised
it from the beginning of his career. Very early in his
LSD he says that . . . no conclusive disproof of a theory
can ever be produced . . .. Statements like this display
one more parallel between Sir Karls view of science
and my own, but what we make of them could scarcely
be more dierent. For my view they are fundamental . . .
For Sir Karls, in contrast, they are an essential qualication which threatens the integrity of his basic position.
Having barred conclusive disproof, he has provided no
substitute for it, and the relation he does employ
remains that of logical falsication. Though he is not a
naive falsicationist, Sir Karl may, I suggest, legitimately
be treated as one. (Kuhn, 1970a, p. 14)
Popper needs something to articulate (some of) his ideas
better. This something is what Kuhn (and Lakatos) has
(have), and what makes that the same statements sound
dierent, as Kuhn acknowledges:
our intentions are often quite dierent when we say the
same things. Though the lines are the same, the gures
which emerge from them are not. (Kuhn, 1970a, p. 3)
The conict clearly emerges in a topic they both explicitly
disagree about, namely the value of NS:
I admit the existence of what Kuhn calls NS, but I dont
admit the evaluation behind the term normal: It is not
only that I dislike it, but I also believe that it became
important just very recently and it is, in my view, a danger for science. (Popper, 1974, p. 1145)
the normal scientist, as Kuhn describes him, is a person
one ought to be sorry for . . . He has been taught in
a dogmatic spirit: he is a victim of indoctrination.
[. . . And his attitude] a danger to science and, indeed,
to our civilization. (Popper, 1970, pp. 5253)
Why is the normal scientist so dangerous? Because, according to Popper, Kuhns normal scientist has been indoctrinated, is dogmatic, acritic and, to sum up, such scientist
continues epistemically attached to the theory/paradigm
even when empirical tests show evidence against it. This
is the core of Poppers charge, an activity like this cannot
be rational in principle since rationality amounts to the dis-

10
Regarding the former, cf. the last paragraph of the last section; about the latter, although I think Kuhn has many insights (cf. Kuhn, 1977), their
elaboration is, to me, totally led astray by his ideas on incommensurability.

Author's personal copy

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

547

position to critically revise our beliefs when contra-evidence is found.

There are more textual references of a similar kind, but I
think that these suce for realising that something goes
wrong in this controversy. Something that, rstly, cannot
be explained in terms of the prescriptive/descriptive distinction since, as we have seen: (i) Popper attributes to Kuhn
an evaluation of the phenomenon; and (ii) although Kuhn
doesnt like to use normative terms, he talks of the dierent
signicance he and Popper attribute to NS, connoting his
positive evaluation of the phenomenon. And it could
hardly be otherwise, since we saw that, terminological preferences aside, both Popper and Kuhn are analysing a piece
of scientic practice constituted by certain rules and, in
attempting to make such rules explicit, both are performing
a normative/descriptive task. And secondly, something
which cannot be explained in terms of the truth/problemsolving distinction either, since their dierent evaluation
of NS does not follow from their theses on the goal of science. A friend of problem-solving may or may not believe
that refutations oblige us to (epistemically, see below)
abandon the paradigm/theory;11 and a friend of truth
may or may not believe that refutations are (epistemically)
compatible with the preservation of (the core of) theories/
paradigms (more on this below, Section 4). Therefore neither the prescriptive/descriptive distinction nor the truth/
problem-solving one suce here. There must be something
else involve and we have to look for it elsewhere.
As I said, I think we have to look for it in a eld apparently not involved in the controversy. As far as I can see,
the only way to fully understand this controversy is taking
into account the dierent conceptions of scientic theories
they implicitly hold. We will have now a brief look on these
conceptions and see later how they help with understanding the otherwise intractable aspects of this controversy.

same weight; this is a rocky view of theories with no dierence between more essential/central and less essential/central
constituents. Then, upon this view, every change is alike.
Kuhns (and Lakatoss) interest in the history of science
allows him (them) to emphasise a fact that, yet almost obvious once one pays attention to it, is extremely important for
a correct analysis of scientic practice, namely that scientic
theories are enduring entities: they change preserving their
identity for a while, passing through dierent stages/versions, until they die, that is, until they are replaced by a
new one. If theories, diachronically considered, are enduring entities then, synchronically considered, they must be
non-compact entities: they must have parts which are not
essential in the sense that their change is compatible with
the survival of the theory; and other parts which are essential, whose change amounts to the death of the theory.
When, after several (and sound) charges of imprecision,
Kuhn makes his original notion of paradigm more precise
by the new concept of disciplinary matrix, he says of the
rst component, the laws:

3. The structure of theories and normal science

The most general symbolic generalisations (like f = ma,

which are not proper laws but schemes of laws and thereby empirically empty/irrefutable taken in isolation) are supposed to apply, in dierent specic versions, to all
exemplars and are essential to the theory/matrix; whereas
the dierent specications of these general principles,
which apply only to specic applications/exemplars, are
not essential since they can change without putting the
identity of the theory into risk. Analogously, some of these
exemplars, the paradigmatic ones, are essential in the sense
that they cannot be questioned without putting the identity
of the theory into risk (for example planetary motions),
whereas others are not essential since they can go in (for
example ties) or out (for example light) with no danger
for the identity of the theory. (Something alike may happen

My thesis is that Popper cannot articulate his view

because he lacks a good reconstruction of essential aspects
of the synchronic (hence diachronic too) nature of theories.
And he thinks that Kuhns NS is a danger for (scientic)
rationality because he misunderstands Kuhn when Kuhn
says that there is no logical force involved in refutation,
that (epistemically) we can keep the same theory after refutations/anomalies and that to abandon a theory is not a
matter of logic.12
As is well known, Poppers view of (the synchronic nature of) scientic theories is the axiomatic traditional one: a
theory is basically a mere conjunction of statements, with
no ne grained structure, where every statement has the
11

. . .symbolic generalisations [like f = ma . . .] are not so

much generalisations as generalisation-sketches, schematic forms whose detailed symbolic expression varies
from one application to the next. For the problem of
free fall, f = ma becomes mg = md2s/dt2. For the simple
pendulum, it becomes mg sinh = md2s/dt2. For coupled harmonic oscillators it becomes two equations,
the rst of which may be written m1d2s1/dt2 + k1s1 =
k2(d + s2 s1). More interesting mechanical problems,
for example the motion of a gyroscope, would display
still greater disparity f = ma and the actual symbolic
generalisation to which logic and mathematics are
applied. (Kuhn, 1970c, p. 465; cf. also Kuhn, 1970d
[1962], Postscriptum, p. 155 and Kuhn, 1970b, p. 272)

Actually, from a problem-solving perspective, if the failure is not massive and the paradigm continues to work quite well in a vast number of other
cases, it might be more rational to try to solve the anomaly while preserving the (nucleus of) the paradigm than to abandon it building a new one, since the
later may imply a (perhaps temporal) lost of problem-solving capacity.
12
To be fair, Kuhn didnt help very much, perhaps because he was not conscious yet that part of the controversy was driven by their dierences about the
structure of scientic theories.

Author's personal copy

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

548

S-T11

f=m(s/t)"

s-dep. forces

dir. dep.

m(s/t)"=-kx

m(s/t)"=
-mgsin

(s/t)'-dep. forces

inv. dep.

sq. inv.

S-T1n

S-T21

S-T22

..

..

..

frict. forc.

m(s/t)"=
-(s/t)'mgsin

m(s/t)"=
G(mm'/s2)

Fig. 1.

with the other two components of the disciplinary matrices,

models and values, although we wont focus on them here).
The essential/accidental distinction applied to the components of scientic theories is obvious once one pays attention to and emphasises the fact that theories, like many
other natural and cultural entities, endure. What emerges
from these considerations is a picture of theories such that,
at a single moment, a theory can be represented as a (treelike) net with a nucleus/core (of laws, principles, and so on)
at the top from which dierent lines of specic laws/principles open out downwards in order to account for dierent
specic phenomena. For example, in the case of Classical
Mechanics (in a specic moment of its history, see below),
something like the tree-like net in Fig. 1.13
This net-like structure represents a single theory in a
specic moment/period of its life. Normal science, that is,
intratheoretic change, is the evolution of a theory through
dierent stages, a succession of such nets preserving the
nucleus (constituted, basically, by some very general
laws/principles and some paradigmatic exemplars). And
revolutionary science, that is, intertheoretic change, is the
substitution of (the last stage of) one theory-evolution by
(the rst stage of) another (which is compatible with some
things being preserved through the revolution,14 yet these
things cannot be the whole nucleus).
Ill use D-theory for referring to theories diachronically
along their whole history, and S-theory for referring to
theories synchronically as the components/stages of D-theories. We can then summarise the (idealised) kinematics of
scientic change in Fig. 2. A certain D-theory D-T1 (e.g.
Classical Mechanics) is constituted along its history by a
sequence of n S-theories S-T11 , S-T21 ; . . . ; S-Tn1 which share
the same nucleus. This sequence, D-T1s life, corresponds
to a period of normal science. When, after a crisis, changes
aect the nucleus, then we dont have normal science any
13

S-T12 ..

D-T1

D-T2

Normal science

Revolution

Normal science

Fig. 2.

more but a revolution, after which a new D-theory D-T2

(e.g. Relativistic Mechanics) comes out in its rst version
S-T12 , a new period of normal science starts with other Stheories S-T22 , S-T32 , . . . as improved versions all sharing
the new nucleus, until the new crisis and revolution.
Though some complexities arise here, what we have seen,
even so simplied, suces for showing how these dierent
conceptions of the structure of scientic theories may
account for the intriguing aspects of the controversy I have
pointed out.
4. Rationality, refutation and S-falsicationism
Poppers correct insight is that rationality implies that
when there is a predictive failure we cannot be epistemically15 indierent, we must do something. And this is correct irrespective of whether we believe that sciences goal
is truth or only problem-solving. But he stated this idea
many times (see the quotations above) in these simplistic
falsicationist terms:
(FALS) Theories must be (epistemically) abandoned
when refuted.
And he complained against Kuhns NS because he thought
that NS violates this rationality condition.
It is my claim that, properly understood, NS does not
violate this rationality condition. The key is to realise that
the occurrence of theories makes FALS ambiguous. It is
then normal that confusion and misunderstanding arise
in the debate: one, Popper, says he (roughly) endorses
FALS, and the other, Kuhn, says he doesnt; but at the
same time they both say they accept part of what the other
wants to mean, and so on. As we have seen Kuhn feels that
the same words sound dierent. Why? In brief: they were
equivocating over the term theory. If we disambiguate
it, we have

What follows is a Sneedianstructuralist presentation of these Kuhnian ideas, which Kuhn himself recognised as an accurate precisication of (this part
of) his thinking (cf. Kuhn, 1976, 2000). This graph shows only the net of laws, it should be completed with the associated applications, models, values, etc.
This is an informal presentation, but it can be made fully precise using the structuralist tools; for an early structuralist precisication of Kuhns ideas (cf.
Stegmuller, 1973, Ch. XIX, and 1979); for a later one with more standard structuralist apparatus (cf. Balzer, Moulines, & Sneed, 1987, Ch. 5, and Balzer &
Moulines, 1998).
14
For instance in the Classical to Relativistic revolution, the idea that for bodies to change their state of motion energy is needed.
15
Epistemically, since this is compatible with continuing to use the refuted theory for practical reasons as long as we dont have better alternatives, or
when we have an epistemically better one but the former is simpler and works quite well in certain domains (e.g. geocentrism for navigation, Classical
Mechanics in vast number of cases, etc.).

Author's personal copy

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

(D-FALS) D-theories must be (epistemically) abandoned when (their last stage is) refuted.
(S- FALS) S-theories must be (epistemically) abandoned
when refuted.
D-FALS is false, it has no rationality import (irrespective of whether sciences goal is truth or merely problemsolving). As Kuhn showed, there is no purely logical reason
for abandoning one D-theory and starting another (in
Lakatosian terms, for abandoning the core instead of keeping addressing the refutation to the protection belt). This
is not matter of logic but of condence failure; scientists
dont trust the theory any more as the frame for trying
to solve the anomalies. Just as, following Stegmullers analogy,16 there is no (formal) logic involved in the decision of
building a new house instead of continuing trying to x the
leaks. Yet according to Kuhn there is a lot of non-(formal)logic rationality: you can be rational or irrational in
doing it, but the criteria have nothing to do with (formal)
logic.17
S-FALS is true (irrespective of whether sciences goal is
problem-solving or truth), and keeps all of Poppers correct
insight. And Kuhn endorses it: in NS, when we have a problem/anomaly, we (epistemically) must do something, we
cannot be (epistemically) indierent, we must change/abandon the S-theory and build a new one. Either preserving
the core, that is, building a dierent S-theory stage of the
same D-theory (changing the protection belt, in Lakatosian
terms); or trying a dierent core, that is, building a dierent
S-theory with a new core opening thereby a new D-theory.
Nothing in logic obliges us to do the later better than the
former and this is perfectly compatible with any sensible
notion of rationality. Yet, logic does oblige us to do either
the former or the later, and Kuhn totally agrees with this
(remember that, among the former, there is room for the
redescription of facts/observations, changing some constants, changing some very specic laws, applying new
measurement methods, postulating new applications, and
so on18).
There is a curious passage that could seem a refutation
of my claim that Kuhn endorses S-FALS, a passage in
which Kuhn says that in NS it is not the theory but the scientist what is tested:19
16

549

when engaged with a normal research problem, the scientist must premise current theory as the rules of his
game. His object is to solve a puzzle . . . [If his conjecture] fails, only [his] own ability not the corpus of current science is impugned. In short, though tests occur
frequently in normal science, these tests are of a peculiar
sort, for in the nal analysis it is the individual scientist
rather than current theory which is tested. (Kuhn,
1970a, pp. 45)
This cannot mean, literally, that in NS the entity which is
the target of an empirical test is not a scientic construct
but a human being. They key is to realise that what scientists have to do in NS is to solve puzzles/anomalies. And an
anomaly is something that does not t with the current theory. Hence the current theory, the current D-theory, has a
problem; that is, its last current S-theory has a problem.
And the fact that Kuhn clearly states here (as everywhere
else) that normal scientists task is to try to x these problems shows uncontroversially that according to him the
normal scientist cannot remain (epistemically) indierent
when confronted to an anomaly. Normal scientists must
try to solve the problem. This is their task in NS, which
in our jargon is tantamount as saying that they have to
change the current S-theory by another one that solves
the anomaly.20 Thus the current S-theory must be (epistemically) abandoned, which is what S-FALS states. Since
we are in NS it must be replaced by another S-theory of
the same current D-theory; in NS the scientist must premise
the (nucleus of the) current (D-)theory. Of course in doing
so the normal scientist may fail, and in this sense it is him,
not the premised D-nucleus, that is challenged.
Hence, in both normal science and revolutionary science
(RS), when we proceed to an empirical test what is tested is
what we use to make the prediction, and this is the whole
theory. To be more precise, what is tested is at least one
branch of the theory-net including both top-nuclear and
bottom-specic parts which, remember, cannot be tested
independently of each other. What is dierent in NS and
in RS is what scientists are ready to do after a negative
outcome. As we have seen, NS is characterised by the
(implicit) rule of trying to solve negative tests addressing
the changes to the bottom-specic parts preserving the

Cf. Stegmuller (1973) (for a related, yet dierent ancestor of this analogy, see also Post, 1971).
This is not exactly inter-paradigmatic theory choice/evaluation (cf. Kuhn, 1977), since in theory choice the options already exist to be compared. The
rationality related to the choice between continuing to reforming the old paradigm or starting a new one has to do with plausibility conditions and what
Kuhn calls crisis phenomena. These plausibility conditions are vague and hard to systematise but, although Kuhn speaks often in quite psychological
terms, they provide the only possible notion of rationality applicable in this case, as we realise when we think in the building-a-new-house analogy: if
things start to go too badly, it is irrational to keep xing the house; as it would be irrational to build a new one after the rst leak (with lot of vagueness in
between, true, but vagueness is a dierent issue).
18
An aside: facts/observations may very well be theory-laden, but not by the same theory-laws that try to explain/predict them. Sometimes they are laden
through non-nuclear connections with other theories (e.g. optics in the case of the telescope and Venus phases), sometimes by a presupposed theory in
the nucleus (e.g. kinematics in mechanics) (cf. Lewis, 1970; Hempel, 1970 and 1973; Sneed, 1979 [1971]; Lakatos, 1970 and Dez, 2002) for the origin of the
distinction between the conceptual part of the theory that describes the facts (concepts whose determination does not presuppose the laws of the theory)
and the conceptual part that explains data so described as a dierent distinction from the traditional observable/theoretical one.
19
I thank Stephan Hartmann for pointing this out to me.
20
Of course this is an idealisation, since as Kuhn insists many times we can never x all the anomalies, paradigms/disciplinary matrices do have always
anomalies.
17

Author's personal copy

550

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

top-nuclear unchanged (the Lakatosian negative and positive heuristics), whereas RS consists in addressing changes
to the nucleus. The kind of activity scientists are ready to
do after a failure test is quite dierent. Normal scientists
have to nd out a way of xing the anomaly without
changing the nucleus, which amounts to dierent aims
and needs dierent abilities to the ones needed in RS. I take
Kuhns statement to be a peculiar way of emphasising this
dierence in attitudes and abilities, connoting that in NS
the (nucleus of the) D-theory is not put into question,
but premised. Yet of course there is one S-theory literally
tested in NS and, if unsuccessfully, scientists are (epistemically) obliged to change it. This is the only interpretation of
the passage compatible with other parts of Kuhns work.
Therefore, Kuhn is not denying S-FALS, quite the
contrary.
Another challenge to my interpretation21 comes from
the many places in which Kuhn says that theory replacement is exclusive of revolutionary science, or even that
[p]aradigms are not corrigible by normal science (Kuhn,
1970d [1962], p. 122). Moreover, Kuhn insists many times
that normal science, unlike revolutionary science, is cumulative, but how could it be so if, as I maintain, theories are
substituted during normal science? The key is to insist that
theory is ambiguous, it can refer to either S-theories or Dtheories. My whole point is just that when Kuhn says such
things, as when he explicitly argues against falsicationism,
he is referring to D-theories. Paradigms are not corrigible
by normal science in the sense that what constitutes the
paradigm/disciplinary matrix is the nucleus, and the
nucleus, as we have seen, is preserved during normal science; actually, the preservation of the nucleus is what
denes normal science. But if we take at a certain moment
the nucleus together with the whole set of complementary
assumptions, laws, and soon, what we obtain is (in our
terms) a complete S-theory, that is, a specic version of
the paradigm/disciplinary matrix in a specic moment.
And these things are replaced during normal science. Not
in their nuclear part, of course, but in their non-nuclear
components. Kuhn devotes the full Chapter III of The
structure to the dierent kinds of activities developed during normal science. The main ones are: improving the
experiments/measurements and the experimental/measurement apparatus; determining or making more precise some
physical constants; identifying new phenomena as similar
to familiar ones; improving the form of existing laws;
applying existing laws to new phenomena (for example
tides); developing new laws relating new phenomena to
old ones (for example hydrodynamics); and, we can add,
even sometimes throwing away some phenomena (for
21

example light). In our terminology, all this consists in modifying the non-nuclear parts of the S-theories, that is, in
substituting a specic S-theory by another which shares
the nucleus with the former. Of course Kuhn does not
express this in these terms;22 the whole point of this paper
is that reconstructing him in this way we understand him
and this part of his controversy with Popper better. This
is the sense in which Kuhn (rightly) endorses S-FALS,
yet (rightly too) not D-FALS. Intra-paradigmatic change
is S-theory replacement, which is compatible with leaving
the (essential part of the) paradigm unchanged. And this
is how normal science may be cumulative (and as a matter
of fact it isoftencumulative), because the common
nucleus is what theoretical content depends on. Since this
common core is what is not preserved in revolutionary/
inter-paradigmatic change, such change, according to
him, does not allow comparison by content and, as a consequence, revolutionary science is not cumulative (all this
depends on additional, and highly controversial theses on
content and meaning that I will not discuss here).
It is worth stressing that, as I have already said, nothing
in this interpretation depends on whether we believe that
the goal of science is truth instead of merely problem solving. Maybe one could say, along the lines of Feyerabends
proliferationism, that logic alone doesnt entitle us to
change the D-theory, but that logic plus the thesis that science aims at truth (not only at problem-solving) does, since
the more the proliferation of theories the higher the
chances for truth; hence this issue would not be independent of theses about sciences goal. But things are not so.
It is true that the greater the proliferation the higher the
chances for truth, but the key question is: the greater the
proliferation of what? Of D-theories or of S-theories? If science aims at truth, after a refutation the truth may lie in
principle in a dierent D-theory as well as in a dierent
S-theory of the same D-theory. That is, the falsehood of
an S-theory may lie in the top, core, D-essential, general
laws/principles, but it may lie in the bottom, more specic,
D-non essential ones as well, and since neither top nor bottom laws/principles can be tested in isolation, refutation does
not entitle us to modify the former rather than the latter,
even if the goal of science is truth. Maybe science aiming
at truth makes a case for proliferation, but the whole point
is that even regarding truth, proliferation of dierent S-versions of the current D-theory is as rational as proliferation
of (incipient) new D-theories.23
To summarise: Popper attributes to Kuhn the theses
that science often behaves non-FALS and that this is good
for science, whereas for him it is a danger for science and
even to our civilisation. Yet, because of the ambiguity of

I owe this criticism to an anonymous referee.

Although he has agreed with a very similar way of re-presenting his ideas (cf. Kuhn, 1976, 2000).
23
Feyerabend doesnt accept the distinction between S-theories and D-theories (cf. Feyerabend, 1970, 1975, 1977), since he considers that, given the
widespread theoretical interrelations, every change infects everywhere. But, again, if Kuhn is right, and we think he is, it is simply false that all parts of the
theories are equally interconnected with every other. The whole point of Kuhn we are emphasising here is that they are not, that we can distinguish degrees
of interaction. And again, Kuhns reconstruction ts historical data better than Feyerabends.
22

Author's personal copy

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

FALS, this charge is misguiding. I am not saying that Popper attributes to Kuhn the negation of both D-FALS and
S-FALS, since he didnt distinguish between them. But
the eect was the same, he attributed to Kuhn a general,
or global, anti-FALS thesis. If we distinguish them, I
think it is clear that Kuhn never defended an anti-S-FALS
thesis (a real danger for science, true, if not to our civilisation), though he did defend an anti-D-FALS thesis, which
is perfectly correct, completely coherent with any sensible
notion of rationality and not dangerous for science at all.
In brief: since Popper didnt have in mind the distinction
between S-theories and D-theories, he (i) could not articulate his non-nave falsicationism, and (ii) could not see
that NS persistence is compatible with FALS-rationality
properly understood. Since Kuhn did have this distinction
implicitly in mind, he could articulate better the same
facts and see that the persistence of NS is perfectly coherent with FALS-rationality properly understood. In our
terms, coherent not with D-FALS, which is false, but with
S-FALS, true and never questioned (at the epistemic level)
by Kuhn. If I am right, then, this implicit equivocation on
the notion of theory causes the puzzling aspects of the
controversy and looking at it we can remove the perplexity
Kuhn himself confesses to feeling:
our intentions are often quite dierent when we say the
same things. Though the lines are the same, the gures
which emerge from them are not. That is why I call what
separates us a gestalt switch rather than a disagreement
and also why I am at once perplexed and intrigued about
how to explore the separation. (Kuhn, 1970a, p. 3, my
emphasis)
All this has to do with possible NS-rationality, that is, it
shows just that, contrary to Popper, NS may be rational.
Which are the criteria for NS actually being rational is
another, related but dierent issue, an issue that has to
do with the connections (both lawful/formal and empirical/applicative) between successive S-theories stronger than
merely preserving the theory-core. It is worth emphasising
that NS is not actually rational merely by denition of
NS.24 What denes NS is, basically, the persistence of certain formal-and-applicative nucleus, plus certain models,
values and practices associated with it. But, although as a
matter of fact scientists do this kind of activity almost
always (and as a group, probably always) rationally, the
notion of NS does not exclude the possibility of an irrational normal scientist, nor even of an irrational normal scientic community. For instance, if they continue working
within what has become, in Lakatosian terms, a clearly
degenerate research programme (that is, post-Lavoisier
phlogiston chemistry, post-Keplerian geocentric astronomy, and the like). It is true that this irrationality is hard
to evaluate at present, since the conceptual possibility that
24

551

a degenerate research programme regenerates and

becomes progressive again is always open, and then it cannot command what to do in a certain moment. But why
should it? Maybe overall it is evaluable, as Lakatos
stressed, only ex post facto. But this is compatible with,
so to speak, local evaluations when the research programme is pretty degenerate at a certain moment. Even if
the posterior regeneration is conceptually possible, it seems
perfectly plausible to say that, at the moment of such a
degenerate stage, it is irrational to continue working in it
(again, the house-leaks analogy applies). Of course much
more has to be said regarding the criteria for progressive/
rational NS, and Kuhn and Lakatos didnt say very much,
but what they did suces for seeing that there is a genuine
sense in which NS is rational (when it is).25
5. Conclusion
If the hypothesis defended is sound, their implicit dierences on the notion of scientic theory is what explains the
noise in the PopperKuhn controversy, why the same
things sound dierent in both mouths, why they seemed
to misunderstand each other. In a sense, they actually misunderstood each other, since the meaning of their statements
about the relation between theory testing, theory abandonment and rationality depends on the implicit reading of
theory, which was dierent. Once this, apparently uninvolved element is made explicit, then Kuhn is right and
Popper is (partially) wrong in their descriptive/prescriptive
endeavour. And this is so independently of whether sciences goal is truth or merely problem-solving. The argument is transcendental in a sense: this interpretation
provides understanding of certain parts of the polemics
and I know no other way of providing such understanding.
Neither the prescriptive/descriptive dichotomy, nor the
truth/problem-solving one can. Nor can the statement/
non-statement distinction, contrary to what Stegmuller
suggests. Even someone like him, the rst who approached
Kuhns ideas in the vein I have presented here, misaddresses this controversy. Evaluating Poppers criticisms of
Kuhns normal science, and after correctly emphasising
that Kuhns normal science essentially presupposes the difference between very general central/core principles and
more specic non-central laws, regarding falsicationism
he says:
As a mathematical structure a theory is immune to falsication as a consequence of the fact that it is not the
kind of entity that make sense to say that it has been
refuted, as it wouldnt make sense either to say, for
instance, that a [algebraic] group has been refuted . . .
The common assumption of all Kuhns opponents is
the statement view, according to which theories are

Of course the rationality of NS does follow from the denition of good NS, but this is of course an additional denition (see below).
Maybe this sense is weaker than Poppers rationality, but as I said at the beginning, philosophy of science has to elucidate the kind of rationality
characteristic of the rules that actually constitute scientic practices, not to misinterpret the rules guided by a preconceived notion of rationality.
25

Author's personal copy

552

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

classes of statements . . . The notion of theory [presupposed by] Kuhn in his description of normal science
can be adequately precisied only within the framework
of the non-statement view of theories. Kuhn himself has
contributed to the confusion conceding too much to his
popperian opponents, admitting for instance, that a theory must be testable in principle. (Stegmuller, 1973, Ch.
IX, 2d)
According to our interpretation, this explanation is
totally misguiding. The statement/non-statement dierence
on the nature of theories has nothing to do with this controversy about falsicationism. One could present both the
Popper and Kuhn accounts in either way. Though Kuhns
ideas are better presented in a non-statement manner, we
can perfectly make sense of the distinction between S-theories and D-theories within the statement view. The reason
why the statement/non-statement dierence does not
account for the PopperKuhn controversy over falsicationism is that, though literally speaking theories reconstructed in a non-statement manner, that is, as sets of
models, are not the kind of entity of which truth and falsehood are predicable, nevertheless they carry a statement
associated with them, the so-called theorys empirical claim,
which asserts that certain empirical systems are embeddable into the models. And this statement is of course suitable
of truth/falsehood. This suces for referring to the refutation of the theory in the sense relevant to the controversy
analysed here.
What I propose is then, in Kuhns terms, like a gestalt
change: if you read things in the proposed way, paying
attention to their implicit disagreement on the nature and
structure of scientic theories, then you see what happened and you understand certain passages that, otherwise, are hardly understandable.26 The only author I
have found that, if I understand him well, is relatively close
to this interpretation is Worrall (2003). He correctly
emphasises the presence of a Duhemian ancestor in Kuhns
theses, namely the idea that the minimum testable unit in
science consists always of a central theory together with a
set of auxiliary assumptions, and that in some cases the
central theory breaks down into a core component together
with a series of more specic assumptions. This fact has the
consequence that [t]he falsity of the central theory does not

follow from the falsity of the empirical conclusion (ibid., p.

73), and therefore
if the empirical consequence entailed by some initially
accepted theoretical system turns out to be false, then
it would be just as dogmatic to arguein the way that
Poppers rhetoric seems to endorsethat it must be
the central theory or the core theory within the central
theory that is false, as it would be to argue that the fault
cannot be with the central theory but instead with some
auxiliary. (Ibid.)
Kuhns anomalies are, then, at least in the simplest case,
falsications of overall theoretical systems that [in NS]
scientists regard as likely to be resolved by replacing that
theoretical system with another one that shares the same
central theory and diers only over some auxiliary or
instrumental assumption. (Ibid., p. 77)
According to Worrall this shows the sense in which, in his
controversy with Popper about falsication and normal
science, reason lies on Kuhns side. Other dierences
aside,27 I nd this quite congenial with the general idea behind the interpretation I have defended here. In particular,
with the theses that (i) it is essential to pay attention to the
dierent essentiality of the dierent components of what
is tested, and (ii) Kuhn rightly rejected D-FALS
yet always accepted S-FALS.
To conclude, one might protest that all this interpretation is an unacceptable deationary view of the part of
the controversy at stake. If things are as I say, reason lies
too obviously on Kuhns side, and for too obvious reasons.
Well, these reasons may very well seem obvious to us,
familiar as we are with diachronic aspects of science and
their synchronic implications. Yet, it was not obvious at
that time at all; to me, one of Kuhns major contributions
was precisely to point out these almost obvious facts and
extract from them a more complex, more accurate and
more useful picture of what D-theories and S-theories
are28 (though often in a not very precise way, it is true).
I want to emphasise again that these conclusions are
explicitly conned to the possible rationality of NS which,
as we have seen, if Popper were right would be irrational
in principle. NS may be rational because it satises the minimum possibility condition for rationality, namely when
testing goes wrong we cannot acritically, dogmatically

26
One can then raise the following question: had Popper entertained and accepted the distinction between S-theories and D-theories, would he have
accepted that (*) the step from one S-theory to another S-theory of the same D-theory preserving the nucleus is compatible with his (non-nave)
falsicationism? Answer: if he accepted that the nuclear laws/principles cannot be tested in isolation (as he probably did), then he should have accepted (*)
(but, who can tell what Popper would actually had done?).
27
This diagnosis, Worrall says, is reinforced if we take into account two new features that Kuhn added to Duhem which are connected with Kuhns
insistence on the importance in science of commitment to tradition (p. 78): (i) often it is not clear for scientists which specic auxiliary assumptions are
the best for completing the central theory; (ii) the central theory is often associated with a heuristics. I think Worrall is right in emphasising these facts, to
which I have not paid enough attention; yet I also think that this does not undermine the core of the interpretation defended here.
28
The same could be said about Lakatos, upon a certain reading of him, since it is not clear to me that his research programmes always t the prole of
our D-theories (cf. 1977). Sometimes they do, but sometimes he seems to refer to more coarse grained entities, including more than one D-theory,
something in between Kuhns D-theories and Laudans research traditions (cf. 1977). But, as the many references to him in the text show, I think that his
essential point against Poppers nave falsicationism and his sophisticated alternative is along the same lines we have outlined here for Kuhn.

Author's personal copy

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

and dangerously remain epistemically indierent, we are

epistemically oblige to change the S-theory. Regarding
other parts of their controversy, like incommensurability,
comparability, intertheoretic progress, realism and sciences goal, I have not said anything and nothing in what
I have said implies that Kuhn does better than Popper
there.29 Indeed, I believe that in some of these topics Kuhn
is wrong, or at least less right than Popper. But this is a different issue. My only thesis here has been that, contrary to
appearances, an essential part of their otherwise intractable
disagreements in their controversy about the possible rationality of NS is explained and solved when we focus on their
dierent conceptions of what scientic theories are. If I am
right, though the goal of this paper is mainly historical, it
would also show that a better reconstruction/interpretation
of scientic constructs allows for a better analysis (descriptive and prescriptive at the same time) of scientic practices. And, as a consequence, that analysis of structural
aspects of scientic constructs is necessary for approaching
some epistemological issues.
Acknowledgements
Research for this work is part of the research projects
HUM2005-03469/FISO (Spanish Ministry of Science and
Education) and BFF2002-10164-E (European Science
Foundation). The main idea was rst presented in a Spanish version in Dez (1998). This paper is a substantial elaboration of this ancestor, thanks to a sabbatical stay, Spring
2005, at the Centre for the Philosophy of Natural and Social Sciences, London School of Economics; I want to
thank the CPNSS for the stimulating atmosphere and the
Spanish Ministry of Science and Education for the nancial support. I also want to thank Roman Frigg, Stephan
Hartmann, Carl Hoefer, Lefteris Farmakis, Salvador Lopez, Ulises Moulines, Ana Rosa Perez and Mauricio Suarez for their comments on previous versions of this paper.
References
Balzer, W., & Moulines, C. U. (1998). Structuralist theory of science:
Paradigmatic reconstructions. Amsterdam: Rodopi.
Balzer, W., Moulines, C. U., & Sneed, J. D. (1987). An architectonic for
science: The structuralist program. Dordrecht: Reidel.
Bird, A. (2000). Thomas Kuhn. Princeton, NJ: Princeton University Press.
Dez, J. A. (1998). Descripcion: prescripcion e interpretacion: La polemica
entre Popper y Kuhn sobre la falsacion y la ciencia normal. In C. Sols
(Ed.), Alta tension: Ensayos en memoria de Thomas Kuhn
(pp. 235258). Madrid: Paidos.
Dez, J. A. (2002). A program for the individuation of scientic concepts.
Synthese, 130, 1348.

553

Feyerabend, P. K. (1970). Consolations for the specialist. In I. Lakatos, &

R. Musgrave (Eds.), Criticism and the growth of knowledge
(pp. 385389). Cambridge: Cambridge University Press.
Feyerabend, P. K. (1975). Against method. London: New Left Books.
Feyerabend, P. K. (1977). Changing patterns of reconstruction. British
Journal for the Philosophy of Science, 28, 351382.
Fuller, S. (2003). Kuhn vs Popper. Cambridge: Icon Books.
Hempel, C. G. (1970). On the standard conception of scientic theories.
In M. Radner, & S. Winokur (Eds.), Analyses of theories and methods
of physics and psychology (pp. 142163). Minneapolis: University of
Minnesota Press.
Hempel, C. G. (1973). The meaning of theoretical terms: A critique of the
standard empiricist construal. In P. Suppes, L. Henkin, A. Joja, & G.
Moisil (Eds.). Logic, methodology and philosophy of science IV
(pp. 367378). Amsterdam: North-Holland.
Kuhn, T. S. (1970a). Logic of discovery or psychology of research? In I.
Lakatos, & R. Musgrave (Eds.), Criticism and the growth of knowledge
(pp. 123). Cambridge: Cambridge University Press.
Kuhn, T. S. (1970b). Reections on my critics. In I. Lakatos, & R.
Musgrave (Eds.), Criticism and the growth of knowledge (pp. 231278).
Cambridge: Cambridge University Press.
Kuhn, T. S. (1970c). Second thoughts on paradigms. In F. Suppe (Ed.),
The structure of scientic theories (pp. 459482). Urbana: University of
Illinois Press.
Kuhn, T. S. (1970d). The structure of scientic revolutions (2nd ed.).
Chicago: University of Chicago Press. (First published 1962)
Kuhn, T. S. (1976). Theory-change as structure-change. Erkenntnis, 10,
179199.
Kuhn, T. S. (1977). Objectivity, values and theory change. In idem, The
essential tension (pp. 320339). Chicago: University of Chicago Press.
Kuhn, T. S. (2000). A discussion with Thomas S. Kuhn [interview with A.
Baltas, K. Gavrogly, & V. Kindi, 1995]. In idem, The road since
Structure: Philosophical essays, 19701993, with an autobiographical
interview (J. Conant, & J. Haugeland, Eds.) (pp. 253324). Chicago:
University of Chicago Press.
Lakatos, I. (1970). Falsication and the methodology of scientic research
programs. In I. Lakatos, & R. Musgrave (Eds.), Criticism and the
growth of knowledge (pp. 91196). Cambridge: Cambridge University
Press.
Lakatos, I. (1977). The methodology of scientic research programmes.
Cambridge: Cambridge University Press.
Laudan, L. (1977). Progress and its problems. Berkeley: University of
California Press.
Lewis, D. (1969). Convention. Cambridge, MA: Harvard University Press.
Lewis, D. (1970). How to dene theoretical terms. In idem, Philosophical
Papers, I (pp. 7895). Oxford: Oxford University Press.
Perez Ransanz, A. R. (1999). Kuhn y el cambio cient co. Mexico: FCE.
Popper, K. R. (1958). The logic of scientic discovery. London: Hutchinson-(First published 1934).
Popper, K. R. (1963). Conjectures and refutations. London: Routledge.
Popper, K. R. (1970). Normal science and its dangers. In I. Lakatos, & R.
Musgrave (Eds.), Criticism and the growth of knowledge (pp. 5158).
Cambridge: Cambridge University Press.
Popper, K. R. (1974). Replies to my critics. In P. A. Schilpp (Ed.), The
philosophy of Karl Popper (pp. 9611199). La Salle: Open Court.
Popper, K. R. (1983). Realism and the aim of science. London: Routledge(First published 1956).
Post, H. R. (1971). Correspondence, invariance and heuristics: In praise of
conservative induction. Studies in History and Philosophy of Science, 2,
213255.

29
Of course this presupposes that it is possible to treat the dierent parts of the PopperKuhn controversy independently of one another, and at least to
treat this part centred on falsicationism independently of other parts related to comparison through revolutions, scientic progress, incommensurability,
etc. The strategy here is then opposite to those who treat the overall controversy holistically (for a moderate cases see Bird, 2000 and Perez Ransanz,
1999; for a radical case, cf. Fuller, 2003). My strategy does not deny that there may be certain aspects of the controversy that deserve a more holistic
approach, it simply states that the aspects presented in the second section can be elucidated prior to, and independently of, the rest in terms of the
distinction between D-theories and S-theories and the corresponding D- and S- falsicationism theses.

Author's personal copy

554

J. Dez / Stud. Hist. Phil. Sci. 38 (2007) 543554

Sneed, J. D. (1979). The logical structure of mathematical physics (2nd ed.).

Dordrecht: Reidel. (First published 1971).
Stegmuller, W. (1973). Theorienstrukturen und Theoriendynamik. Heidelberg: Springer.

Stegmuller, W. (1979). The structuralist view of theories. Berlin: Springer.

Worrall, J. (2003). Normal science and dogmatism, paradigms and
progress: Kuhn versus Popper and Lakatos. In T. Nickles (Ed.),
Thomas Kuhn (pp. 65100). Cambridge: Cambridge University Press.