LAUDAN, Larry - Science and Values - The Aims of Science and Their Role in Scientific Debate
LAUDAN, Larry - Science and Values - The Aims of Science and Their Role in Scientific Debate
LAUDAN, Larry - Science and Values - The Aims of Science and Their Role in Scientific Debate
Pittsburgh Series in
Philosophy and History
of Science
Series Editors:
Adolf Gnunbaurn
I a n - y Laudan
Nicholas Wescher
Wesley C. Salmon
SCIENCE
AND
VALUES
T h e Aims ofscience and
Their Role lliz Scientfic Debate
--
Larry Laudan
Laudan, Larry.
Science and values.
Acknowledgments
Preface
Epilogue
References
Index
ACKNOWLEDGMENTS
Blacksburg, Virginia
15 July 1983
PREFACE
many readers will probably expect the book to be about ethical rather
than cognitive values stands as eloquent testimony to the fact that, for
much too long, we have imagined that the only really deep value ques-
tions that arise in, and about, science have to do with ethical or moral
values. In this book I talk quite a lor about values, and to that extent
its title is apt; but I have nothing to say about ethical values as such,
for they are manifestly not the predominant values in the scientific
enterprise. Not that ethics plays no role in science; on the contrary,
ethical values are always present in scientific decision making and,
very occasionally, their influence is of great importance. But that
importance fades into insignificance when compared with the nbiqui-
tous role of cognitive values. One function of this book is to redress the
imbalance that has led so many recent writers on science to be pre-
occupied with scientific morality rather than with scientific rationality,
which is more centrally my focus.
In sum, this is a book about the role of cognitive values in the shap-
ing of scientific rationality. Among recent writers, no one has done
more to direct our attention to the role of cognitive standards and
values in science than Thomas Kuhn. Indeed, for more than two de-
cades, the views of Thomas Kuhn- and reactions to them- have occu-
pied center stage in accounts of scientific change and scientific ration-
ality. That is as it should be, for Kuhn's Structure of Scientific Revolu-
tions caused us all to rethink our image of what science is and how it
works. There can be no one active today in philosophy, history, or
sociology of science whose approach to the problem of scientific ration-
ality has not been shaped by the Gestalt switch Kuhn wrought on our
perspective on science. This debt is so broadly recognized that there is
no need to document it hereV3 Less frequently admitted is the fact that,
in the twenty-two years since the appearance of T h e Structure of Sci-
entific Revolutions, a great deal of historical scholarship and analytic
spadework has moved our understanding of the processes of scientific
rationality and scientific change considerably beyond the point where
Kuhn left it.
Indeed, we are now in a position to state pretty unequivocally that
-
this rough-and-ready characterization: an attribute will count as a cognitive value or
aim i f that attribute represents a property o f theories which we deem t d b e constitutive
o f "good science."
3 - o n e indication o f the magnitude o f that debt may be found by perusing Gutting,
1980.
Preface
tween science and everything else, might well dispute it. In any event,
it is not in my power, nor is it my intention, to detract from Kuhn's
seminal contribution when I assert that we have by now built upon it
and moved beyond it. It is the burden of the arguments in the chapters
that follow to show why that is the proper thing to say.
Chapter One
T W O PUZZLES
ABOUT SCIENCE:
REFLECTIONS ON SOME
CRISES IN PHILOSOPHY
AND SOCIOLOGY
OF SCIENCE
1. Recall, for instance, Karl Popper's remarkable claim dating from this period that
the most important task of epistemology was to demarcate between science and non-
science.
T w o Puzzles about Science 3
reasons why scientists might agree to differ, leave us largely in the dark
about how scientists could ever reasonably resolve their differences in
the definitive fashion in which they often do terminate controversies.
The theme of this essay, in its starkest form, is simply (a) that exist-
ing accounts lack the explanatory resources to tackle these two puzzles
in tandem; ( 6 ) that this is especially true of several recently fashionable
approaches to science, which turn out to be at least as flawed as those
they would replace; and ( c ) that we need a single, unified theory of sci-
entific rationality which promises to be able to explain both these strik-
ing features about science. My aim in this first chapter is to diagnose
how we landed in the mess of being able to explain one or the other of
these puzzles, but not both. The remainder of tlhe book delineates
some machinery that explains how both consensus and dissensus can
arise, and how each can sometimes give rise to the other.
point of view which may never even have been mooted a decade ear-
lier. Moreover, change occurs at a variety of levels. [Some of the central
problems of the discipline change; the basic explanatory hypotheses
shift; and even the rules of investigation slowly evolve. That a consen-
sus can be shaped and reshaped amid such flux is indeed remarkable.
However unsatisfactory the models of scientific consensus offered by
the last generation may now appear to be, it is surely easy enough to
understand why their framers believed that the explanation of scien-
tific agreement must be a central consideration for any theory about
how science works. For, when one takes into account the rapid-fire
manner in which new views emerge, the staggering t:hing about science
is not that consensus is generally reached so quickly and with such
unanimity; what is astounding is that consensus is ever reached at all.
Taking the high level of consensus in science as a datum, intellec-
tuals of the preceding generation constructed models of science, and
especially of scientific decision making, which were designed to explain
how science differed structurally and methodologically from such
ideology-laden fields as social and political theory or metaphysics. I
want to describe the salient features of some of those models, for an
appreciation of their strengths and weaknesses will be useful later on.
a ) Philosophers and consensus. -Philosophers of the 1930s and
1940s,turning anew to science after a generation of comparative philo-
sophical neglect by many idealists and Neo-Kantians in the first de-
cades of the twentieth century, already had some sophisticated ma-
chinery in their kits for explaining how science could be a consensual
activity. Indeed, for a very long time philosophers generally have been
inclined to accept what I call the Leibnizian ideal. In brief, the Leib-
nizian ideal holds that all disputes about matters of fact can be impar-
tially resolved by invoking appropriate rules of evidence. At least since
Bacon, most philosophers have believed there to be ;in algorithm or set
of algorithms which would permit any impartial observer to judge the
degree to which a certain body of data rendered different explanations
of those data true or false, probable or improbable. Philosophers have
expressed varying degrees of optimism about whether we now know
precisely what those evidentiary rules are. (Mill, for instance, believed
that we already had them in hand. Others, more pessimistic, believed
that we had yet to develop the full kit.) But whether optimist or pessi-
mist, rationalist or empiricist, most logicians and philosophers of sci-
ence from the 1930s through the 1950s believed, at least in principle,
6 Two Puzzles a bout Science
that this long-standing controversy was not the refutation of the reign-
ing consensual models and the Leibnizian ideal which it appeared to
be. Similar claims were made about the observational equivalence of
matrix and wave mechanics and about corpuscular and wave optics.
(As we now know, most of these arguments were bogus, for they de-
pended on showing that two theories were equivalent so long as their
formal structures-i.e., their mathematical representations- could be
shown to be homologous. Unfortunately, these proofs of "empirical
equivalence" work only if we divest these theories of most of their sub-
stantive claims. But more of that in chapter 5 below.) Thus the philo-
sophical advocates of consensus as the scientific norm could explain
away the apparent exceptions to that consensus by insisting that, when
consensus was not reached as quickly as one might expect, it was either
because the decisive evidence was not sought, or because the scientists
concerned did not realize that their rival theories really amounted to
the same thing, or (in the last resort) because scientists were not behav-
ing rationally.
Other prominent elements of the philosophy of science of logical
empiricism contributed to the impression that science should indeed
be a consensual activity. It was commonly asserted, for instance, that
one core rule of scientific method was that acceptable new theories
must be able to explain all the successes of their predecessors and some
new facts as well. Science, in short, was thought to be strictly cumula-
tive. With this strong constraint in place, it became possible to explain
how scientific change could be effected fairly quickly. After all, if a
new theory emerged which managed to account for everything its
predecessor could, and some other things besides, then it would seem
that no sensible person could resist the appeal of the new theory. So
long as theory change could be said to be strictly cumulative, the phi-
losopher had a ready explanation for the staggeringly swift changes of
loyalty which accompany many so-called scientific revolutions. And it
is for just this reason that the post-1960s discovery that theory change
in science is generally noncumulative and nonconvergent created such
acute difficulties for the logical empiricists and for Popperas
b) Sociologists and scientqic consensus. -If philosophers had a long
tradition of expecting and explaining the existence of agreement about
matters of fact, sociologists did not. Indeed, prior to the 1930s there
scarcely was a sociology of science worthy of the name. The following
two decades, however, saw a significant flowering of sociological stud-
ies of science. Central to much of the research of that era are our dual
problems of consensus and dissensus. As with the philosophers, sociolo-
gists tended to regard the former as the natural state of the physical
sciences, whereas the latter required special explanation as a deviation
from the expected norm.
Whereas philosophers located the source of the consensual character
of science in the scientist's adherence to the canons of a logic of scien-
tific inference, sociologists argued that science exhibited so high a
degree of agreement because scientists shared a set of norms or stan-
dards which governed the professional life of the scientific community.
Robert Merton, for instance, argued that scientific subcultures shared
the norms of "universalism, communism, disinterestedness, and orga-
nized ~cepticism."~ These norms, which are "held to be binding on the
man of science, . . . are expressed in the form of prescriptions, proscrip-
tions, preferences and perrni~sions."~ It is, in short, because scientists
share the same values or standards that they are able to form stable
patterns of consensus. Merton was later to find what he regarded as
strong support for the hypothesis of shared scientific norms and stan-
dards in research he did with Harriet Zuckerman. Specifically, he and
Zuckerman discovered that journals in the humanities and social sci-
ences have a consistently higher rejection rate for submitted articles
than do journals in the natural sciences. (In the Merton and Zucker-
man study, for instance, the physics journals sampled rejected only 24
percent of submissions, whereas sociology and philosophy journals
rejected more than 80 percent.) Merton took these divergences as evi-
dence that philosophers and sociologists could not agree about what
constituted significant or solid research, whereas natural scientists
could agree about the merits of specific contributions by virtue of their
shared norms and values. As Merton and Zuckerman wrote in 1971,
"This suggests that these fields of learning [i.e., sociology and philoso-
phy] are not greatly institutionalized in the reasonably precise sense
that editors and referees on the one side and would-be contributors on
6 . For a detailed account of these norms, see Merton's classic "The Normative
Structure of Science," reprinted in Merton, 1973.
7. Merton, 1973, pp. 268-269.
18 Two Puzzles about Science
13. Merton, 1973, p. 270. As further evidence that Merton believed that the norms
were ultimately grounded in the rules of method, consider that he defines one of the
core norms, organized skepticism, as "the detached scrutiny of beliefs in terms of
empirical and logical criteria" (Merton, 1968, chap. 8). Such c:riteria are surely those
provided by the rules of scientific methodology.
12 T w o Puzzles about Science
the consensual model and its implied Leibnizian ideal were sound, it is
very difficult to understand how mavericks or revolutionaries in the
scientific community could ever get their ideas off the ground. As
Kuhn cogently argued, "In short, if a new candidate for [reigning]
paradigm had to be judged from the start by hard-headed people who
examined only relative problem-solving ability, the sciences would
experience very few major revolution^."'^ Since revolutions do not
occur overnight, every scientific revolution must be inaugurated by a
period when some scientists are pursuing new ideas and others are
quite happy with the reigning theories. The consensual model was said
by its critics to make it very difficult to understand how reasonable
men could ever differ, in ways that seem to be required to permit the
exploration of new ideas. As Thomas Kuhn succinctly formulated this
objection to the consensual approach: the emergence of new scientific
ideas "requires a decision process which permits rational men to dis-
agree, and such disagreement would generally be barred by the shared
algorithm which philosophers have generally sought. If it [i.e., such an
algorithm] were at hand, all conforming [i.e., rational] scientists
would make the same decision at the same time."" Kuhn maintains
that it is only the existence of differential preferences and values
among scientists which allows new theories to flower. Otherwise, "no
one. . .would be inclined to try out the new theory, to articulate it in
ways which showed its fruitfulness or displayed its accuracy and
scope."ls It is telling that Kuhn in this passage, as in much of his work,
ignores the fact that scientists can distinguish between criteria for
acceptance of theories and criteria of pursuit worthiness.lg Such a dis-
tinction allows one to circumvent some of the problems Kuhn raises for
the consensual view. But to this extent Kuhn is surely right: the eon-
sensual view fails to make sense of the broad range and variety of cases
of scientific disagreement. Because it does, something more must be
going on than meets the consensual eye.
b ) T h e thesis of incommensurability. -Kuhn himself proposed to
fill in a part of the picture by claiming that the advocates of rival theo-
ries simply fail to communicate with one another. This failure is no
20. See Feyerabend, 1978; Mitroff, 1974. Mitroffs evidence for effective "counter-
normal" behavior is a good deal more compelling than Feyerabend's.
T w o Puzzles about Science
22. K u h n ( i b i d . , p. 17) writes: " W h a t is surprizing, and perhaps also unique i n its
degree t o t h e fields we call science, is that such initial divergences should ever largely
disappear."
23. Pbid., pp. 143 ff.
24. T h e Planck principle is summed u p i n Max Planck's famous q u i p ( 1 9 4 9 , pp. 33-
34): " a new scientific t r u t h does not t r i u m p h b y convincing its opponents and rnaking
t h e m see t h e light, b u t rather because its opponents eventually die and a new genera-
tion grows u p that is familiar w i t h it." For a n excellent critique o f t h e implications o f
Muhn's version o f t h e Planck principle, see Mull, 1978.
T w o Puzzles about Science 19
rival paradigms should have all the same difficulties their elders do in
reaching agreement about the respective merits of competing para-
digms. Identical objections apply to Kuhn's suggestion that hegemony
and normal science reassert themselves once the advocates of a particu-
lar paradigm get control of the major journals and the prestige
appointments in a discipline. Even if true, such a reduction of scien-
tific decision making to Realpolitik leaves unexplained the processes
whereby the scientific elite in science comes to rally around a single
new paradigm.
On other occasions, sounding rather more traditional, Kuhn says
that consensus eventually congeals around a new paradigm because it
can be seen to be objectively better than its predecessor by such criteria
as its degree of empirical support, its demonstrated fertility, and its
perceived problem-solving ability.25But if it is possible to compare
theories along these vectors so as to get all or most scientists to agree
about them, then it is unclear what all of Kuhn's earlier fuss about
incommensurability and the absence of shared standards amounted
to. He cannot have it both ways. Either there are shared and unambig-
uous standards which can be invoked by the proponents of rival para-
digms for deciding the issue between them (in which case Kuhn's talk
about incommensurability and the nonspecificity of shared cognitive
values comes to naught, thus undermining his explanation of dissen-
sus), or else there are no such standards (in which case Kuhn's account
of disagreement escapes unscathed but only at the apparent expense of
his being unable to explain consensus formation),
Kuhn is scarcely unique among contemporary philosophers and
sociologists of science in propounding an account of disagreement
which leaves little or no scope for explaining agreement. Imre Lakatos
and Paul Feyerabend, for instance, are in the same plight, if for rather
different reasons. Lakatos went to great lengths to stress the role of
various conventions in theory assessment. For him, the decision to treat
a prima facie falsifying instance as a genuine refutation was a matter
of "convention." Mindful of the Duhemian ambiguities of falsification,
Lakatos argued that rational scientists could completely ignore appar-
ent refutations for their research programs. If they do, it becomes
entirely conceivable that rival theorists might conduct a controversy
for years, even decades, without the disagreement issuing in any firm
25. For this side of Kuhn's work, see especially the last chapters of Kuhn, 1962.
20 Two Puzzles a bout Science
consensus. But what Lakatos always left opaque was how a community
of scientists might reasonably come to the conclusion that one research
program was genuinely superior to another, thereby reestablishing
consensus. On Lakatos's account, as on Kuhn's, it appears reasonable
to hang onto a theory-no matter what empirical anomalies confront
it -more or less indefinitely. But to say as much is, in effect, to say that
there are no rational mechanisms whereby consensus about the prefer-
ability of one line of research over another can be established. Since
that sort of consensus is commonplace in the sciences, Lakatos's
approach leaves us with no explanation of the fact that scientists come,
often quite rapidly, to regard most scientific controversies as defini-
tively resolved.
If Lakatos was an anarchist in spite of himself, Feyerabend set out
quite deliberately to elaborate a theory of knowledge which would
favor rampant theoretical pluralism. In Feyerabend's view of the mat-
ter, it is undesirable that scientists should ever reach consensus about
anything. His ideal of science is the sort of endless questioning of fun-
damentals which one associates with pre-Socratic natural philosophy:
nothing is taken as given, everything can reasonably be denied or
affirmed. Like Kuhn, Feyerabend believes in the radical incommen-
surability of theories. Far more than Kuhn, he denies that there are
any methodological principles or norms which it is reasonable to insist
that scientists follow in assessing theories ("anything goes"). Feyera-
bend does not deny that scientists sometimes do agree about which
theories are good and which are bad, but he deplores that state sf
affairs as unreasonable. 'If scientists only appreciated the finer points
of epistemology, he seems to say, they would see that no theories
should ever be regarded as having displaced or discredited their rivals
and predecessors.
Sociologists, too, have been quick to see that the existence of wide.
spread controversy in science fits ill with older models of science in
their field. Michael Mulkay seems to speak for many of the new-wave
theorists in regarding the phenomenon of scientific disagreement as a
refutation of older approaches: "'If, for example, the Mertonian norms
are effectively institutionalized in science, it becomes difficult to
account for the frequency of intellectual resistance [which] is recurrent
in science and is, indeed, an inescapable feature of the growth of scien-
tific k n ~ w l e d g e . " ~ ~
T H E HIERARCHICAL
STRUCTURE OF
SCIENTIFIC DEBATES
1. Among the influential philosophical advocates of the hierarchical model are Karl
Popper, C. G. Hempel, and Hans Reichenbach.
24 Hierarchical Structure of Scientific Debates
that ideal in its original forrn imagined that all factual disagreements
could be terminated by invoking the relevant rules, latter-day propo-
nents of methodological rules tend to be more modest. They continue
to believe that some disagreements can be immediately resolved by
utilizing the available evidence (and the shared rules). Failing that,
however, they go on to say that the rules are often sufficiently specific
to indicate procedures for the collection of such additional evidence as
will bring the issue to a definitive resolution. The rules themselves vary
from the highly general ("formulate testable and simple hypotheses")
to those of intermediate generality ("prefer the results of double-blind
to single-blind experiments"), to those specific to a particular disci-
pline or even subdiscipline ("make sure to calibrate instrument x
against standard y"). To the extent that these procedural or method-
ological rules are accepted by all parties to the dispute, and insofar as
they are sufficiently specific to determine a choice between the avail-
able rivals, they should indeed suffice for the mediation of factual con-
troversies. And a staggeringly large proportion of factual disputes have
evidently been ended simply by observing the relevant methodological
procedures.
Sometimes, however, scientists disagree about the appropriate rules
of evidence or procedure, or about how those rules are to be applied to
the case at hand.2 In such circumstances, the rules can no longer be
treated as an unproblematic instrument for resolviilg factual disagree-
ment. When this happens, it becomes clear that a particular factual
disagreement betokens a deeper methodological disagreement. In the
standard hierarchical view, such methodological controversies are to
be resolved by moving one step up the hierarchy, that is, by reference
to the shared aims or goals of science. This suggestion is a natural one
5. I have formulated the problem for a choice between two rival theories; it could
readily be generalized to apply to a larger number.
Hierarchical Structure of Scientqic Debates 29
6. Tough-minded readers may want to work with a more demanding sense of per-
missible belief, according to which a belief is regarded as warranted only if it enjoys
stronger support from the rules than all its extant rivals. (Otherwise, we might find
ourselves in a situation where contrary beliefs were all permissible.) I would myself
resist the stronger characterization of "permissible," but the argument I develop here
in the text will work, whichever formulation of permissible one inclines to.
7. Including Quine, Hesse, and Bloor.
30 Hierarchical Structure of Scientific Debates
have no rules that would unambiguously pick out a single theory to the
exclusion of all other possible theories about the relevant domain, it is
inevitable that the choice between any two theories could always go
one way or the other, given any set of values or norms about what we
expect our theories to achieve. Kuhn is not denying that rules play a
role in the choice of scientific theories, but he is insisting that their
intrinsic ambiguity precludes the possibility of decisive preferences
ever being justified on the basis of shared methodological rules. Kuhn
grants that "such canons [his examples are accuracy, simplicity, gen-
erality, etc.] do exist and should be discoverable," but he goes on to
insist that "they are not by themselves sufficient to determine the deci-
sions of individual scientists."* If all that Kuhn is saying here is that
general rules and values underdetermine choice (in the strict sense
defined above), one might not quarrel with his claim; but, as his sub-
sequent discussion makes clear, he is asserting-to use my earlier lan-
guage-that preferences are underdetermined as well. He puts it this
way: "every individual choice between competing theories depends on
a mixture of objective and subjective factors, of shared and individual
u rite ria."^ Kuhn thinks that this state of affairs must be so because,
according to his analysis, the "objective" criteria shared by scientists
cannot justify one preference to the exclusion of another. Since, as
Kuhn knows perfectly well, scientists do in fact voice theory prefer-
ences, he takes this as evidence that they must be working with various
individual and idiosyncratic criteria that go well beyond the shared
ones. Without the latter, he seems to say, how could scientists ever
have preferences? But neither Kuhn nor anyone else has shown, either
in fact or in principle, that such rules and evaluative criteria as are
shared among scientists are generally or invariably insufficient to indi-
cate unambiguous grounds for preference of certain theories over
others.
Indeed, using some of Kuhn's own criteria, we can show precisely
how a rationally based preference can arise. Suppose that a scientist is
confronted with a choice between specific versions of Aristotle's physics
and Newton's physics. Suppose moreover that the scientist is com-
mitted to observational accuracy as a primary value. Even granting
with Kuhn that "accuracy" is usually not precisely defined, and even
typically be many theories ruled out by the operant rules; and if one
party to a scientific debate happens to be pushing for a theory that can
be shown to violate those rules, then the rules will eliminate that theory
from contention.
It follows from this analysis that the hierarchical model is not (as
many of its critics claim) rendered toothless by the argument from
underdetermination. To the contrary, it seems entirely reasonable to
say that many disputes about matters of fact have been terminated by
invoking shared procedural rules. But it is important to realize what
"hedges" are built into this reformulation of the hierarchical view. Not
all factual disagreements can be resolved in this Leibnizian fashion
because two or more extant rivals may be equally well supported by the
existing rules and evidence. Similarly, it may happen (a case that 1 will
explore shortly) that scientists differ about which evidential rules
should be applied to the case at hand. The fact that this version of the
hierarchical model leads us to expect consensus only in some instances
will satisfy neither the archrationalist advocates of the Leibnizian ideal
(who want an in-principle, instant termination for every factual dis-
agreement) nor the proponents of radical underdetermination (who
believe in the potential prolongation of every disagreement indefi-
nitely). But the strength of this version of the hierarchical model is that
it can (unlike the underdeterminationists) specify circumstances in
which we would expect a factual disagreement to dissolve into consen-
sus and it can (unlike the Leibnizian idealists) also specify a broad
range of circumstances in which we would expect factual dissensus to
endure.
11. Idiscuss the early history of the rule of predesignation in Laudan, 1981.
12. A second, probably insuperable, obstacle to the quest for the scientific method
is the absence of a full consensus among scientists about what the cognitive goals of sci-
ence should be. Without that agreement we should scarcely be surprised if some dis-
sensus about methods persists. (But it should be stressed, as noted below, pp. 43 ff.,
that a lack of consensus about cognitive aims does not necessarily entail the existence
of any dissensus about the appropriate methods.)
Hierarchical Structure of Scientific Debates 37
atic is the situation where the disagreement about rules derives from
an even deeper disagreement about cognitive aims (discussed in detail
in chap. 3).
But before we move on to deal with the very tricky case of disagree-
ments about basic goals, we should pause a moment to reflect on some
of the implications of our account of the resolution of methodo'logical
disagreements. If the analysis sketched here is correct, we have to real-
ize that the vertical hierarchy of facts-rules-aims has a rather more
complex structure than figure 1 and our earlier discussion might sug-
gest. Although we appraise methodological rules by asking whether
they conduce to cognitive ends (suggesting movement up the justifica-
tory hierarchy), the factors that settle the question are often drawn
from a lower level in the hierarchy, specifically from the level of factual
inquiry. Factual information comes to play a role in the assessment of
methodological claims precisely because we are continuously learning
new things about the world and about ourselves as observers of that
world. Such knowledge comes to be formulated in theories concerning,
among other things, the various complexities of the process of collect-
ing evidence. To take an elementary example, we have learned that
nature does not offer information to us in a random or statistically rep-
resentative way. As medium-sized objects in a world replete with the
very small and the very large, the entities and processes we are most
likely to encounter in our everyday scrutiny of the world are highly un-
representative of that world in many crucial respects. Once we learned
that fact, it became necessary to develop elaborate sampling tech-
niques in order to make our evidence more representative than it
would have been had we simply collected whatever information casu-
ally came our way. Contrary to popular conception, the superiority of
randomly collected data over nonrandomly collected data is not a dis-
covery made by mathematics or formal logic as a feature of inquiry in
all possible worlds. It is because the particular world we inhabit turns
out to be so uncooperative and, as we have learned that only through
hard-won experience, we find it appropriate to insist on a variety of
stringent rules about sampling in order to achieve our cognitive ends.
Consider a different example. Within the past fifty years we have
learned a great deal about the so-called placebo effect. In brief, many
patients report an improvement upon being given any apparent medi-
cation, even if (unbeknownst to the patient) it is pharmacologically
inert. Until we learned this, simple controlled experiments were re-
Hierarchical Structure ofscientzyic Debates 39
13. There is thus a central, but monvicious, circularity about our evaluative proce-
dures: we use certain methods for studying the world and those very methods may
serve initially to authenticate discoveries that expose the weaknesses of those selfsame
methods.
14. This suggestion should not be confused with Quine's familiar argument about
"naturalized epistemology." Whereas Quine sees close affinities between psychology
and epistemology, I am not committed to that particular program of reduction. My
insistence, rather, is that the appraisal of proposed cognitive methods and aims re-
quires extensive empirical research. That research will often have nothing sperifically
40 Hierarchical Structure of Scientfic Debates
reveal how deeply they have misunderstood the fact that an empirical
approach to epistemology requires attention to precisely those norma-
tive linkages between cognitive ends and means which constitute scien-
tific rationality.
To conclude, we find ourselves with mixed results. There do seem to
be circumstances under which both factual and methodological dis-
agreements can be brought to a rational resolution by seeking shared
assumptions at a higher level. The familiar idea that agreement about
factual matters is possible only among those who already accept the
same methodoilogical rules is clearly seen to be too restrictive, for it
fails to reckon with the fact that, by moving a step up the hierarchy to
aims and goals, dissensus about rules may dissolve into consensus. And
that consensus about rules, once in place, may be sufficient to resolve
the disagreement about the contested factual matters. But we have
also discovered a number of problem cases. When the shared rules fail
to dictate a factual preference, when the shared goals fail to specify a
methodological preference, when values are shared but not weighted
equally, and when values are not fully shared, we seem to be con-
fronted by an irresolvable disagreement-irresolvable, that is, if we
stick to the limited resources of the classical hierarchical model. Most
crucial of all is the situation where scientists subscribe to different
goals. Such circumstances occur sufficiently often to pose a fundamen-
tal challenge to the very idea that science is a rational and progressive
undertaking. But, as we shall see in chapter 3, that challenge is not
always so stark as it may seem.
Chapter Three
CLOSING T H E
EVALUATIVE CIRCLE:
RESOLVING
DISAGREEMENTS ABOUT
COGNITIVE VALUES
same ontology and the same axiology. So far as I know, Kuhn never
imagines that there might be fundamental ontological or theoretical
differences between scientists who share the same cognitive goals.
Because he works constantly with a presumed covariance between
these two levels, Kuhn normally takes differences at the level of basic
theories as evidence for fundamental differences of aim or value. Of
course, many others besides Kuhn tend to assume that divergences
about matters of fact warrant an assumption of axiological differences.
In chapter 4 I show in detail how badly flawed this thesis of covariance
is; it is sufficient here simply to note the fact that the claim of covari-
ance is a non sequitur. Precisely because (as Kuhn himself stresses in
other contexts) cognitive values underdetermine methodological rules,
and because those rules in turn sometimes underdetermine theory pref-
erence, it is entirely conceivable that two scientists may subscribe to
precisely the same cognitive goals and yet advocate fundamentally dif-
ferent views about the furniture of the universe. Because they may, one
must resist the tendency to read off a divergence about aims from every
long-term disagreement about factual or methodological matters.
The second form of the covariance fallacy is the mirror image of the
first. Here, one tends to assume that when scientists agree about fac-
tual and methodological matters, such agreement must have resulted
from shared cognitive goals. So tight are these particular linkages often
regarded that it is frequently assumed that an enterprise like science
(in which agreement about theories and methods is rather common-
place) could exhibit so high a degree of factual and methodological
consensus only if there was consensus at the level of aims and goals.
Indeed, as pointed out in chapter 1, a key stimulus for classical sociol-
ogy and philosophy of science was the conviction that scientists must
operate with common goals, since they can reach agreement so often
about the "facts." Many sociologists of the last generation saw their
major role as that of identifying the norms that underwrote such fac-
tual consensus; and philosophers of that era assumed that, because
scientists were often able to agree at the factual level, such agreement
must issue from a prior agreement about cognitive aims or epistemic
utilities. Yet a bit of reflection makes clear that the connections be-
tween consensus about values and agreement at other levels are much
less tight than the covariance fallacy might lead one to imagine. It is
entirely conceivable, for instance, that thinkers who subscribe to genu-
inely different cognitive aims might well subscribe to similar (or even
Closing the Evaluative Circle 45
1. See, for instance, the discussion of realism and instrumentalism in Nagel, 1961.
46 Closing the Evaluative Circle
2. It is more than a little ironic that many of those philosophers who have been most
vocal in their condemnation of relativism (e.g., Popper, Lakatos) have subscribed to
the view that the selection of goals and methods is a matter of largely unreasoned con-
ventions. A relativism with respect to aims is at least as debilitating as (and probably
itself entails) a relativism with respect to knowledge.
48 Closing the Evaluative Circle
3. See esp. Popper's discussion of the status of methodological rules and aims in the
early sections of his 1959 book.
4. Lakatos, 1978, p. 144.
5. Reichenbach, 1938, pp. 10-13.
50 Closing the Evaluative Circle
wise accept the Kuhnian point that scientists of different schools per-
sistently endorse different cognitive goals, then we are forced to say
that the various shifts in the predominant goals in science are just part
of the history of taste and fashion, rather than part of the reasoned
and rational history of human thought. Still worse, given the centrality
of cognitive goals in the justificatory structure of every science, any
arbitrariness that infects the choice of cognitive goals will raise real
doubts about the credentials of the factual claims of the sciences para-
sitic on those goals. If no set of (consistent) cognitive goals can legiti-
mately be held to be rationally preferable to any other, we seem forced
to face the prospect that there may be indefinitely many alternative
of "science," each tailored to meet different ends and each
forms
entirely legitimate in its own right. In short, radical relativism about
science seems to be an inevitable corollary of accepting (a) that differ-
ent scientists have different goals, ( b ) that there is no rational delibera-
tion possible about the suitability of different goals, and (c) that goals,
methods, and factual claims invariably come in covariant clusters.
But here a crucial flaw appears, for what is being assumed is that a
rational choice between alternative sets of internally consistent sets of
cognitive goals is always impossible. This assumption, I believe, is
false, not always, but in a sufficiently large range of cases so as to make
its general espousal highly misleading. It is false because, to make a
long story unconscionably short, there is a wide array of critical tools
which we can utilize for the rational assessment of a group of cognitive
aims or goals. Once one realizes how extensive that tool kit is, it will
become clear that we are driven to none of the consequences outlined
above, provided only
that we are prepared to make some major
changes in the hierarchical model.
age-old parental advice like "do as I say, not as I do," drive home the
fact that the aims that an agent claims to endorse are often at odds
with those that apparently guide his actions. When we find ourselves in
a situation where there is a tension between our explicit aims and those
implicit in our actions and judgments, we are naturally under signifi-
cant pressure to change one or the other, or both. On pain of being
charged with inconsistency (not to mention hypocrisy, dishonesty,
etc.), the rational person, confronted with a conflict between the goals
he professes and the goals that appear to inform his actions, will
attempt to bring the two into line with each other.
Precisely the same sort of thing happens in science. Often a scientist
will find himself explicitly advocating certain cognitive aims, yet seem-
ingly running counter to those aims in terms of the actual theory
choices he makes in his daily scientific work. Still worse, as we shall see
below, it sometimes happens that the dominant goals or an entire com-
munity of scientists, as voiced in the explicit accounts they give of these
matters, are discovered to be at odds with the goals that actually seem
to inform that community's choices and actions as scientists. Whenever
a case can be made that a group of scientists is not practicing what it
preaches, there are prima facie grounds for a change of either explicit
or implicit values. The change may come, of course, in either area, or
in both. One may retain one's professed goals and force them to shape
one's practical judgments and actions; or one may adopt a new set of
explicit values that accord more nearly with the one's actions and prac-
tical judgments. Whichever way it goes, the engine driving axiological
change is grounded in a theory of rationality, acting to overcome a
state of disequilibrium.
Because this mode of critically evaluating cognitive values is so
important, I want to illustrate its workings in some detail by a signifi-
cant historical example: the decision by many working scientists in the
late 1700s and early 1800s to give up the view that we should seek to
restrict our theories entirely to claims about observable entities and
processes. This important shift in cognitive orientation was absolutely
essential to the development of such theories as atomism, uniformi-
tarianism, and natural selection. So long as people insisted, as they
had through most of the eighteenth century, that science must avoid
postulating entities that cannot be directly observed, theories that
referred to objects too small to be observed (e.g., atoms), or to pro-
Closing the Evaluative Circle
7 . From Boscovich's De Solis a Lunae Defectibus (1760). Quoted from, and trans-
lated by, Dugald Stewart, 1854-, 2:212.
8. See esp. Hartley, 1749, 1:341-352.
9. From a letter published in Prevost, 1805, p. 390.
10. Prevost, 1805, p. 300.
11. Ibid., p. 265.
12. Ibid., pp. 464-4165,
Closing the Evaluative Circle 59
quent verification can rarely establish the truth of any general conclu-
sions. But then, as he pointed out, induction and analogy-the meth-
ods favored by traditional empiricists -are also inconclusive. What we
must aim at in these matters, says Lesage, is high probability, and he
indicated circumstances under which we are entitled to assert well-
confirmed hypotheses with confidence. He went on to point out that
the great Isaac Newton, for all his professed inductivism, extensively
utilized the method of hypothesis. It is, he says, to hypothesis "without
any element of [induction or] analogy that we. . . owe the great discov-
ery of the three laws which govern the celestial bodies."l3 Generalizing
this point, Lesage argued that there is an element of conjecture or
hypothesis in every inductive inference that goes beyond its premises,
which all except so-called perfect inductions do. Since no interesting
scientific claim can restrict itself to what has been observed, concludes
Lesage, it is unreasonable to make such a restriction into an epistemic
aim.
Within a half century after Lesage's death, the "official" method-
ology of the scientific community had come to acknowledge the legiti-
macy of hypotheses about unobservable entities. What forced the
change was a growing recognition that the explicit axiology of empiri-
cism was fundamentally at odds with the axiology implicit in scientists'
theory preferences. This recognition becomes fully explicit in the writ-
ings
- of Herschel and Whewell.
I have discussed this case at such length because it vividly illustrates
the manner in which implicit and explicit axiological commitments
can be played off against one another so as to bring theory and prac-
tice into closer agreement. The older explicit aim of a science free of
unobservable entities became a casualty of the striking success of theo-
ries that postulated such entities. But although this case is more vivid
than most, it represents a common mechanism for the rational adjudi-
cation of rival scientific goals.
Moreover, this episode illustrates how the existence of broad agree-
ment about which scientific theories are the best can play a crucial role
in resolving differences between thinkers with respect to the goals they
explicitly profess. So long as two warring scientific factions can agree
about certain instances of exemplary science (and is there a situation
13. This quotation is from Lesage's "Premier MCmoire sur la Mkthode dlHypo-
th;se," published posthumously in Prevost, 1804, par. 23.
60 Closing the Evaluative Circle
where scientists are completely unable to agree about some such exam-
ples?), those examples can be brought to bear in examining the con-
flicting goals to which the factions explicitly subscribe. If one of the
two parties happens to be insisting on a certain goal that the shared
exemplars fail to exhibit, then there is a prima facie case for rejecting
the goal in question, precisely because it is not realized by (what all
parties to the dispute agree is) a good scientific theory. Of course, it is
always possible that, confronted with this conflict between implicit
and explicit goals, certain scientists will choose to retain their explicit
aims and to reject as illegitimate what they had previously regarded as
an ideal example of sound scientific practice. (In the historical exam-
ple cited above, several late eighteenth-century thinkers were prepared
to do exactly that, even to the point of rejecting large parts of Newton-
ian physics because it did not exemplify their explicit values.) But un-
less scientists are prepared to go that far (and in the overwhelming
number of cases they are not prepared to renounce the major scientific
achievements of their predecessors when they have nothing with which
to replace them), it remains a compelling argument against a proposed
cognitive aim if the primary theories of a discipline fail to exemplify it.
There is yet another way in which scientific standards come to be
reasonably abandoned, even if the process is as much by default as by
design. Such abandonment occurs when scientists discover that, de-
spite persistent and arduous efforts, they can produce no theories that
manage to exemplify those standards or ideals. Consider, for instance,
the fate of the demand for intelligibility or cogency of conception, in
the wake of Newton's Principia. Until Newton's day, natural philoso-
phers had generally insisted that our explanatory principles must have
a conceptual accessibility about them. The whole point about scien-
tific explanation, it was said, was to explain the less intelligible by the
more intelligible. The idea of explaining the world by postulating enti-
ties and processes which were even more obscurely understood was
anathema to the mainstream intellectual traditions of the West.
Shortly before Newton's time, this explanatory ideal had been de-
scribed at length by Descartes, who insisted that our explanatory con-
cepts must be both clear and distinct. Descartes, of course, had singled
out the idea of action at a distance as an archetypal example of an
obscure concept, which had no place in the explanatory repertoire of
the working scientist. Barely had this dismissal of action at a distance
become the orthodoxy, when Newton came along, arguing that gravi-
Closing the Evaluative Circle 61
Methods
. -
Theories 4 c Aims
Must harmonize
It may seem to the skeptical reader that the constraints on goals pro-
posed in the reticulational model are very weak, so much so that many
different sets of cognitive goals may well satisfy them. Doubtless a wide
range of cognitive goals or values can satisfy the demands laid down
here. In one sense, that is good news, for if only one set complies with
these constraints, and if these principles are taken to define a minimal
form of rationality, then there can never be legitimate grounds for
scientists to disagree about standards. And that, in turn, would mean
that all those occasions when scientists have disagreed about standards
in the past reflected massive amounts of irrationality in science. The
bad news, or so it will be regarded by some, is that several different,
even mutually incompatible, goals may satisfy these constraints. Those
who want a highly ambitious theory of scientific rationality will prob-
ably ask: "But how does the reticulational analysis tell us which among
the surviving goals is the right one?" I have no answer to give to that
question, but I hasten to add that the question itself rests on illicit pre-
suppositions. There is no single "right" goal for inquiry because it is
64 Closing the Evaluative Circle
14. We could, of course, utilize goals to which neither the actors nor we as observers
subscribe in order to make judgments of progress, although it is not clear what use
such an analysis would serve.
66 Closing the Evaluative Circle
DISSECTING THE
HOLIST PICTURE OF
SCIENTIFIC CHANGE
tory one, to such an extent that some passages in his later writings
make him sound like a closet positivist. More than one commentator
has accused the later Kuhn of taking back much of what made his mes-
sage interesting and provocative in the first place.'
But that is not entirely fair, for if many of Kuhn's clarifications have
indeed taken the sting out of what we once thought Kuhn's position
was, there are several issues about which the later Kuhn is both clear
and controversial. Significantly, several of those are central to the
themes of this essay. Because they are, I want to use Kuhn's work as a
stalking-horse to show how the features of the reticulational model,
proposed in the two preceding chapters, can be used to produce a
more satisfactory account than Kuhn offers of scientific debate in par-
ticular and scientific change in general.
Kuhn, then, will be my immediate target, but I would be less than
candid if I did not quickly add that the views I discuss here have spread
considerably beyond the Kuhnian corpus. To some degree, almost all
of us who wrote about scientific change in the 1970s (present company
included) fell prey to some of the confusions I describe. In trying to
characterize the mechanisms of theory change, we have tended to lapse
into sloppy language for describing change. However, because Kuhn's
is the best-known account of scientific change, and because Kuhn most
overtly makes several of the mistakes I want to discuss, this chapter
focuses chiefly on his views. Similar criticisms can be raised with vary-
ing degrees of severity against authors as diverse as Foucault, Lakatos,
Toulmin, Holton, and Laudan.
1. Alan Musgrave spoke for many of Kuhn's readers when he noted, apropos of the
second edition of The Structure of Scientific Revolutions, that in "his recent writings,
then, Kuhn disowns most of the challenging ideas ascribed to him by his critics. . . the
new, more real Kuhn who emerges. . . [is] but a pale reflection of the old, revolution-
ary Kuhn" (Musgrave, 1980, p. 51).
Dissecting the Holist Picture of Scientific Change 69
interact. In short, every paradigm will make certain claims about what
populates the world. Such ontological claims mark that paradigm off
from others, since each paradigm is thought to postulate entities and
modes of interaction which differentiate it from other paradigms. Sec-
ond, a paradigm will specify the appropriate methods, techniques,
and tools of inquiry for studying the objects in the relevant domain of
application. Just as different paradigms have different ontologies, so
they involve substantially different methodologies. (Consider, for in-
stance, the very different methods of research and theory evaluation
associated with behaviorism and cognitive psychology respectively.)
These methodological commitments are persistent ones, and they
characterize the paradigm throughout its history. Finally, the propo-
nents of different paradigms will, according to Kuhn, espouse differ-
ent sets of cognitive goals or ideals. Although the partisans of two para-
digms may (and usually do) share some aims in common, Kuhn insists
that the goals are not fully overlapping between followers of rival para-
digms. Indeed, to accept a paradigm is, for Kuhn, to subscribe to a
complex of cognitive values which the proponents of no other para-
digm accept fully.
Paradigm change, on this account, clearly represents a break of
great magnitude. To trade in one paradigm for another is to involve
oneself in changes at each of the three levels defined in chapter 2
above. We give up one ontology for another, one methodology for an-
other, and one set of cognitive goals for another. Moreover, according
to Kuhn, this change is simultaneous rather than sequential. It is
worth observing in passing that, for all Kuhn's vitriol about the impov-
erishment of older models of scientific rationality, there are several
quite striking similarities between the classical version of the hierarchi-
cal model and liuhn's alternative to it. Both lay central stress on the
justificatory interactions between claims at the factual, methodologi-
cal, and axiological levels. Both emphasize the centrality of values and
standards as providing criteria of choice between rival views lower in
the hierarchy. Where Kuhn breaks, and breaks radically, with the tra-
dition is in his insistence that rationality must be relativized to choices
within a paradigm rather than choices between paradigms. Whereas
the older account of the hierarchical model had generally supposed
that core axiological and methodological commitments would typi-
cally be common property across the sciences of an epoch, Kuhn
asserts that there are methodological and axiological discrepancies
70 Dissecting the Holist Picture of Scientific Change
between any two paradigms. Indeed (as we shall see below), one of the
core failings of Kuhn's position is that it so fully internalizes the classi-
cal hierarchical approach that, whenever the latter breaks down (as it
certainly does in grappling with interparadigmatic debate, or any
other sort of disagreement involving conflicting goals), Kuhn's ap-
proach has nothing more to offer concerning the possibility of rational
choices. 2
For now, however, the immediate point to stress is that Kuhn por-
trays paradigm changes in ways that make them seem to be abrupt and
global ruptures in the life of a scientific community. So great is this
supposed transition that several of Kuhn's critics have charged that,
despite Kuhn's proclaimed intentions to the contrary, his analysis
inevitably turns scientific change into a nonrational or irrational pro-
cess. In part, but only in part, it is Kuhn's infelicitous terminology that
produces this impression. Notoriously, he speaks of the acceptance of a
new paradigm as a "conversion e~perience,"~ conjuring up a picture of
the scientific revolutionary as a born-again Christian, long on zeal and
short on argument. At other times he likens paradigm change to an
"irreversible Gestalt-~hift."~ Less metaphorically, he claims that there
is never a point at which it is "unreasonable" to hold onto an old para-
digm rather than to accept a new o n e 5 Such language does not en-
courage one to imagine that paradigm change is exactly the result of a
careful and deliberate weighing-up of the respective strengths of rival
contenders. But impressions based on some of Kuhn's more lurid lan-
guage can probably be rectified by cleaning up some of the vocabulary
of The Structure of Scientific Revolutions, a task on which Kuhn has
been embarked more or less since the book first a ~ p e a r e dNo. ~ changes
of terminology, however, will alter the fact that some central features
of Kuhn's model of science raise serious roadblocks to a rational analy-
2. It has been insufficiently noted just how partial Kuhn's break with positivism is,
so far as cognitive goals and values are concerned. As I show in detail below, most of
his problems about the alleged incomparability of theories arise because Kuhn accepts
without argument the positivist claim that cognitive values or standards at the top of
the hierarchy are fundamentally immune to rational negotiation.
3. Kuhn, 1962.
4. Ibid.
5. Ibid.., D. 159.
L
phlogiston theory, just as it was fully rational for most of his contem-
poraries to be converting to the oxygen theory of Lavoisier. According
to Kuhn, Priestley's continued adherence to phlogiston was reasonable
because -given Priestley's cognitive aims and the methods he regarded
as appropriate-his own theory continued to look good. Priestley lost
the battle with Eavoisier, not because Priestley's paradigm was objec-
tively inferior to its rivals, but rather because most of the chemists of
the day came to share Lavoisier's and Dalton's views about what was
important and how it should be investigated.
The clear implication of such passages in Kuhn's writings is that
interparadigmatic debate is necessarily inconclusive and thus can
never be brought to rational closure. When closure does occur, it must
therefore be imposed on the situation by such external factors as the
demise of some of the participants or the manipulation of the levers of
power and reward within the institutional structure of the scientific
community. Philosophers of science, almost without exception, have
found such implications troubling, for they directly confute what phi-
losophers have been at pains for two millennia to establish: to wit, that
scientific disputes, and more generally all disagreements about matters
of fact, are in principle open to rational clarification and resolution. It
is on the strength of passages such as those I have mentioned that
Kuhn has been charged with relativism, subjectivism, irrationalism,
and a host of other sins high on the philosopher's hit list.
There is some justice in these criticisms of Kuhn's work, for (as I sug-
gest in chap. 1) Kuhn has failed over the past twenty years to elaborate
any coherent account of consensus formation, that is, of the manner in
which scientists could ever agree to support one world view rather than
another. But that flaw, serious though it is, can probably be remedied,
for I want to suggest that the problem of consensus formation can be
solved if we will make two fundamental amendments in Kuhn's posi-
tion. First (as argued in chap. 3), we must replace the hierarchical
view of justification with the reticulated picture, thereby making cog-
nitive values "negotiable." Second, we must simply drop Kuhn's insis-
tence on the integral character of world views or paradigms. More spe-
cifically, we solve the problem of consensus once we realize that the
various components of a world view are individually negotiable and
individually replaceable in apiecemeal fashion (that is, in such a man-
ner that replacement of one element need not require wholesale repu-
diation of all the other components), Kuhn himself grants, of course,
74 Dissecting the Holist Picture of Scientific Change
and
9. Ibid., p. 149.
Dissecting the Holist Picture of Scientqic Change 75
As shown in chapter 2, choices like those between (1) and (2), or be-
tween (1)and (3), are subject to strong normative constraints. And we
saw in chapter 3 that choices of the sort represented between (1) and
(4) are also, under certain circumstances, equally amenable to rational
analysis.
In all these examples there is enough common ground between the
rivals to engender hope of finding an "Archimedean standpoint"
which can rationally mediate the choice. When such commonality
exists, there is no reason to regard the choice as just a matter of taste or
whim; nor is there any reason to say of such choices, as Kuhn does
(recall his characterization of the Priestley-Lavoisier exchange), that
there can be no compelling grounds for one preference over another.
Provided theory change occurs one level at a time, there is ample scope
for regarding it as a thoroughly reasoned process.
But the crucial question is whether change actually does occur in
this manner. If one thinks quickly of the great transitions in the history
of science, they seem to preclude such a stepwise analysis. The shift
from (say) an Aristotelian to a Newtonian world view clearly involved
changes on all three levels. So, too, did the emergence of psychoanaly-
sis from nineteenth-century mechanistic psychology. But before we
accept this wholesale picture of scientific change too quickly, we
should ask whether it might not acquire what plausibility it enjoys only
because our characterizations of such historical revolutions make us
compress or telescope a number of gradual changes (one level at a
time, as it were) into what, at our distance in time, can easily appear as
an abrupt and monumental shift.
By way of laying out the core features of a more gradualist (and, I
argue, historically more faithful) picture of scientific change, I will
sketch a highly idealized version of theory change. Once it is in front of
us, I will show in detail how it makes sense of some real cases of scien-
76 Dissecting the Holist Picture ofScientific Change
------ Methodology
tific change. Eventually, we will want a model that can show how one
might move from an initial state of disagreement between rival tradi-
tions or paradigms to consensus about which one is better. But, for
purposes of exposition, I want to begin with a rather simpler situation,
namely, one in which consensus in favor of one world view or tradition
gives way eventually to consensus in favor of another, without scientists
ever being faced with a choice as stark as that between two well-devel-
oped, and totally divergent, rival paradigms. My "tall tale," repre-
sented schematically in figure 4, might go like this: at any given time,
there will be at least one set of values, methods, and theories which one
Dissecting the Holist Picture of Scientqic Change
11. See Whewell's remarkably insightful essay of 1851, where he remarks, apropos
the transition from one global theory to another: "the change. . . is effected by a trans-
formation, or series of transformations, of the earlier hypothesis, by means of which it
is brought nearer and nearer to the second [i.e., later]" (1851, p. 139).
Dissecting the Holist Picture of Scientijic Change 81
12. Some amplification of this point is required. Kuhn evidently believes that there
are some values that transcend specific paradigms. He mentions such examples as the
demand for accuracy, consistency, and simplicity. The fortunes of these values are not
linked to specific paradigms. Thus, if they were to change, such change would pre-
sumably be independent of shifts in paradigms. In Kuhn's view, however, these values
have persisted unchanged since the seventeenth century. Or, rather, scientists have
invoked these values persistently since that time; strictly speaking, on Muhn's analysis,
these values are changing constantly, since each scientist interprets them slightly dif-
ferently. For a detailed discussion of Kuhn's handling of these quasi-shared values, see
the final section of this chapter.
82 Dissecting the Holist Picture of Scientqic Change
tant to see that this deep shift in axiological sensibilities was indepen-
dent of any specific change in scientific world view or paradigm. No
new scientific tradition or paradigm in the nineteenth century was
associated with a specifically fallibilist axiology. Quite the reverse, fal-
libilism came to be associated with virtually every major program of
scientific research by the mid- to late nineteenth century. Atomists and
antiatomists, wave theorists and particle theorists, Darwinians and
Lamarckians, uniformitarians and catastrophists-all subscribed to
the new consensus about the corrigibility and indemonstrability of sci-
entific theories. A similar story could be told about other cognitive
values which have gone the way of all flesh. The abandonment of intel-
ligibility, of the requirement of picturable or mechanically construct-
ible models of natural processes, of the insistence on "complete" de-
scriptions of nature- all reveal a similar pattern. The abandonment of
each of these cognitive ideals was largely independent of shifts in basic
theories about nature.
Once again, the holistic approach leads to expectations that are con-
founded by the historical record. Changes in values and changes in
substantive ontologies or methodologies show no neat isomorphism.
Change certainly occurs at all levels, and sometimes changes are con-
current, but there is no striking covariance between the timing of
changes at one level and the timing of those at any other. I conclude
from such examples that scientific change is substantially more piece-
meal than the holistic model would suggest. Value changes do not
always accompany, nor are they always accompanied by, changes in
scientific paradigm. Shifts in methodological rules may, but need not,
be associated with shifts in either values or ontologies. The three levels,
although unquestionably interrelated, do not come as an inseparable
package on a take-it-or-leave-itbasis.
This result is of absolutely decisive importance for understanding
the processes of scientific change. Because these changes are not always
concomitant, we are often in a position to hold one or two of the three
levels fixed while we decide whether to make modifications at the dis-
puted level. The existence of these (temporarily) fixed and thus shared
points of perspective provides a crucial form of triangulation. Since
theories, methodologies, and axiologies stand together in a kind of jus-
eificatory triad, we can use those doctrines about which there is agree-
ment to resolve the remaining areas where we disagree. The uncon-
tested levels will not always resolve the controversy, for underdetermi-
Dissecting the Holist Picture of Scientqic Change 85
nation is an ever present possibility. But the fact that the levels of
agreement are sometimes insufficient to terminate the controversy pro-
vides no comfort for Kuhn's subjectivist thesis that those levels of agree-
ment are never sufficient to resolve the debate. As logicians say, we
need to be very careful about our quantifiers here. Some writers have
not always exercised the care they should. Kuhn, for instance, confus-
edly slides from ( a )the correct claim that the shared values of scientists
are, in certain situations, incapable of yielding unambiguously a pref-
erence between two rival theories to ( b ) the surely mistaken claim that
the shared values of scientists are never sufficient to warrant a prefer-
ence between rival paradigms. Manifestly in some instances, the shared
rules and standards of methodology are unavailing. But neither Kuhn
nor anyone else has established that the rules, evaluative criteria, and
values to which scientists subscribe are generally so ambiguous in ap-
plication that virtually any theory or paradigm can be shown to satisfy
them. And we must constantly bear in mind the point that, even when
theories are underdetermined by a set of rules or standards, many the-
ories will typically be ruled out by the relevant rules; and if one party
to a scientific debate happens to be pushing for a theory that can be
shown to violate those rules, then the rules will eliminate that theory
from contention.
What has led holistic theorists to misdescribe so badly the relations
among these various sorts of changes? As one who was himself once an
advocate of such an account, I can explain specifically what led me
into thinking that change on the various levels was virtually simultane-
ous. If one focuses, as most philosophers of science have, on the pro-
cesses of justification in science, one begins to see systemic linkages
among what I earlier called factual, methodological, and axiological
ideas. One notices further that beliefs at all three levels shift through
time. Under the circumstances it is quite natural to conjecture that
these various changes may be interconnected. Specifically, one can
imagine that the changes might well be simultaneous, or at least
closely dependent on one another. The suggestion is further borne
out - at least to a first approximation -by an analysis of some familiar
scientific episodes. It is clear, for instance, that the scientific revolution
of the seventeenth century brought with it changes in theories, ontolo-
gies, rules, and values. Equally, the twentieth-century revolution in
relativity theory and quantum mechanics brought in its wake a shift in
both methodological and axiological orientations among theoretical
86 Dissecting the Holist Picture of Scientific Change
badly flawed. Some authors, for instance, tend to confuse the logical
underdetermination of theories by data with the underdetermination
of theory choice by methodological rules. Others (e.g., Hesse and
Bloor) have mistakenly taken the logical underdetermination of theo-
ries to be a license for asserting the causal underdetermination of our
theoretical beliefs by the sensory evidence to which we are exposed.lg
But there is a weaker, and much more interesting, version of the thesis
of underdetermination, which has been developed most fully in Kuhn's
recent writings. Indeed, it is one of the strengths of Kuhn's challenge
to traditional philosophy of science that he has "localized" and given
flesh to the case for underdetermination, in ways that make it prima
facie much more telling. In brief, Kuhn's view is this: if we examine
situations where scientists are required to make a choice among the
handful of paradigms that confront them at any time, we discover that
the relevant evidence and appropriate methodological standards fail
to pick out one contender as unequivocally superior to its extant
rival(s). I call such situations cases of "local" underdetermination, by
way of contrasting them with the more global forms of underdeterrni-
nation (which say, in effect, that the rules are insufficient to pick out
any theory as being uniquely supported by the data). Kuhn offers four
distinct arguments for local underdetermination. Each is designed to
show that, although methodological rules and standards do constrain
and delimit a scientist's choices or options, those rules and standards
are never sufficient to compel or unequivocally to warrant the choice
of one paradigm over another.
1) T h e "anabiguity of shared standards" argument. -Kuhn's first
argument for methodological underdetermination rests on the pur-
ported ambiguity of the methodological rules or standards that are
shared by advocates of rival paradigms. The argument first appeared
in T h e Structure of Scientzyic Revolutions (1962) and has been ex-
tended considerably in his later T h e Essential Tensiori (1977). As he
put it in the earlier work, "lifelong resistance [to a new theory]. . . is
not a violation of scientific standards. . . though the historian can
always find men- Priestley, for instance - who were unreasonable to
resist for as long as they did, he will not find a point at which resistance
becomes illogical or un~cientific."2~ Many sf Kuhn's readers were per-
19. See ibid. for a lengthy treatment of some issues surrounding underdetermina-
tion of theories.
25. Muhn, 1962, p. 159.
Dissecting the Holist Picture of Scientific Change 89
them in the same way. Simplicity, scope, fruitfulness, and even accu-
racy can be judged differently (which is not to say they may be judged
arbitrarily) by different people."z2Because, then, the shared standards
are ambiguous, two scientists may subscribe to "exactly the same stan-
dard" (say, the rule of simplicity) and yet endorse opposing viewpoints.
Kuhn draws some quite large inferences from the presumed ambi-
guity of the shared standards or criteria. Specifically, he concludes
that every case of theory choice must involve an admixture of objective
and subjective factors, since (in Kuhn's view) the shared, and presum-
ably objective, criteria are too amorphous and ambiguous to warrant a
particular preference. He puts the point this way: "I continue to hold
that the algorithms of individuals are all ultimately different by virtue
21. Kuhn, 1977, p. 322.
22. Ibid., p. 262.
90 Dissecting the Holist Picture of Scientific Change
are shared-are so often able to agree about which theories to bet on.
But we can leave it to Kuhn to sort out how he reconciles his commit-
ment to the social psychology of science with his views about the indi-
vidual vagaries of theory preference. What must concern us is the
question whether Kuhn has made a plausible case for thinking that the
shared or collective criteria must be supplemented by individual and
subjective criteria.
The first point to stress is that Kuhn's thesis purports to apply to all
scientific rules or values that are shared by the partisans of rival para-
digms, not just to a selected few, notoriously ambiguous ones. We can
grant straightaway that some of the rules, standards, and values used
by scientists ("simplicity" would be an obvious candidate) exhibit pre-
cisely that high degree of ambiguity which Kuhn ascribes to them. But
Kuhn's general argument for the impotence of shared rules to settle
disagreements between scientists working in different paradigms can-
not be established by citing the occasional example. Kuhn must show
us, for he claims as much, that there is something in the very nature of
those methodological rules that come to be shared among scientists
which makes the application of those rules or standards invariably
inconclusive. He has not established this result, and there is a good rea-
son why he has not: it is false. To see that it is, one need only produce a
methodological rule widely accepted by scientists which can be applied
to concrete cases without substantial imprecision or ambiguity. Con-
sider, for instance, one of Kuhn's own examples of a widely shared sci-
entific standard, namely, the requirement that an acceptable theory
must be internally consistent and logically consistent with accepted
theories in other fields. (One may or may not favor this methodological
rule. I refer to it here only because it is commonly regarded, including
by Kuhn, as a methodological rule that frequently plays a role in
theory evaluation.)
I submit that we have a very clear notion of what it is for a theory to
be internally consistent, just as we understand perfectly well what it
means for a theory to be consistent with accepted beliefs. Moreover, on
at least some occasions we can tell whether a particular theory has vio-
lated the standard of (internal or external) consistency. Kuhn himself,
in a revealing passage, grants as much; for instance, when comparing
the relative merits of geocentric and heliocentric astronomy, Kuhn
says that "the consistency criterion, by itself, therefore, spoke unequiv-
Dissecting the Holist Picture of Scientqic Change
ocally for the geocentric tradition."Z6 (What he has in mind is the fact
that heliocentric astronomy, when introduced, was inconsistent with
the then reigning terrestrial physics, whereas the assumptions of geo-
centric astronomy were consistent with that physics.) Note that in this
case we have a scientific rule or criterion "speaking unequivocally" in
favor of one theory and against its rival. Where are the inevitable
imprecision and ambiguity which are supposed by Kuhn to afflict all
the shared values of the scientific community? What is ambiguous
about the notion of consistency? The point of these rhetorical ques-
tions is to drive home the fact that, even by Kuhn's lights, some of the
rules or criteria widely accepted in the scientific community do not
exhibit that multiplicity of meanings which Kuhn has described as
being entirely characteristic of methodological standards.
One could, incidentally, cite several other examples of reasonably
clear and unambiguous methodological rules. For instance, the re-
quirements that theories should be deductively closed or that theories
should be subjected to controlled experiments have not generated a
great deal of confusion or disagreement among scientists about what
does and does not constitute closure or a control. Or, consider the rule
that theories should lead successfully to the prediction of results un-
known to their discoverer; so far as I am aware, scientists have not dif-
fered widely in their construal of the meaning of this rule. The signifi-
cance of the nonambiguity of many methodological concepts and rules
is to be found in the fact that such nonambiguity refutes one of Kuhn's
central arguments for the incomparability of paradigms and for its
corollary, the impotence of methodology as a guide to scientific ration-
ality. There are at least some rules that are sufficiently determinate
that one can show that many theories clearly fail to satisfy them. We
need not supplement the shared content of these objective concepts
with any private notions of our own in order to decide whether a theory
satisfies them.
2) The "collective inconsistency of rules" argument.-As if the
ambiguity of standards was not bad enough, Kuhn goes on to argue
that the shared rules and standards, when taken as a collective, "re-
peatedly prove to conflict with one another."Z7For instance, two scien-
tists may each believe that empirical accuracy and generality are desir-
able traits in a theory. But, when confronted with a pair of rival (and
thus incompatible) theories, one of which is more accurate and the
other more general, the judgments of those scientists may well differ
about which theory to accept. One scientist may opt for the more gen-
eral theory; the other, for the more accurate. They evidently share the
same standards, says Kuhn, but they end up with conflicting apprais-
als. Kuhn puts it this way: . . in many concrete situations, different
I ' .
"tension." Since these two claims were the linchpins in Kuhn's argu-
ment to the effect that shared criteria "are not by themselves sufficient
to determine the decisions of individual scientist^,"^^ we are entitled to
say that Kuhn's effort to establish a general form of local underdeter-
mination falls flat.
3 ) The shifting standards argument. -Equally important to Kuhn's
critique of methodology is a set of arguments having to do with the
manner in which standards are supposed to vary from one scientist to
another. In treating Kuhn's views on this matter, I follow Gerald Dop-
pelt's excellent and sympathetic explication of Kuhn's position.g1 In
general, Kuhn's model of science envisages two quite distinct ways in
which disagreements about standards might render scientific debate
indeterminate or inconclusive. In the first place, the advocates of dif-
ferent paradigms may subscribe to different methodological rules or
evaluative criteria. Indeed, "may" is too weak a term here, for, as we
have seen, Kuhn evidently believes that associated with each paradigm
is a set of methodological orientations that are (at least partly) at odds
with the methodologies of all rival paradigms. Thus, he insists that
whenever a "paradigm shift" occurs, this process produces "changes in
the standards governing permissible problems, concepts and explana-
t i o n ~ . "This
~ ~ is quite a strong claim. It implies, among other things,
that the advocates of different paradigms invariably have different
views about what constitutes a scientific explanation and even about
what constitutes the relevant facts to be explained (viz., the "permis-
sible problems"). If Kuhn is right about these matters, then debate
between the proponents of two rival paradigms will involve appeal to
different sets of rules and standards associated respectively with the
two paradigms. One party to the dispute may be able to show that his
theory is best by his standards, while his opponent may be able to
claim superiority by his.
As I have shown in detail earlier in this chapter, Kuhn is right to say
that scientists sometimes subscribe to different methodologies (includ-
ing different standards for explanation and facticity). But he has never
shown, and I believe him to be chronically wrong in claiming, that dis-
agreements about matters of standards and rules neatly coincide with
disagreements about substantive matters of scientific ontology. Rival
scientists advocating fundamentally different theories or paradigms
often have the same standards of assessment (and interpret them iden-
tically); on the other hand, adherents to the same paradigm will fre-
quently espouse different standards. In short, methodological dis-
agreements and factual disagreements about basic theories show no
striking covariances of the kind required to sustain Kuhn's argument
about the intrinsic irresolvability of interparadigmatic debate. It was
the thrust of my earlier account of "piecemeal change" to show why
Kuhn's claims about irresolvability will not work.
But, of course, a serious issue raised by Kuhn still remains before us.
If different scientists sometimes subscribe to different standards of
appraisal (and that much is surely correct), then how is it possible for
us to speak of the resolution of such disagreements as anything other
than an arbitrary closure? To raise that question presupposes a picture
of science which I sought to demolish in chapters 2 and 3. Provided
there are mechanisms for rationally resolving disagreements about
methodological rules and cognitive values (and I describe several of
those mechanisms in chap. 3), the fact that scientists often disagree
about such rules and values need not, indeed should not, be taken to
show that there must be anything arbitrary about the resolution of
such disagreements.
4) the problem-weighting argument. -As I have said earlier, Kuhn
has another argument up his sleeve which he and others think is ger-
mane to the issue of the rationality of comparative theory assessment.
Specifically, he insists that the advocates of rival paradigms assign dif-
ferential degrees of importance to the solution of different sorts of
problems. Because they do, he says that they will often disagree about
which theory is better supported, since one side will argue that it is
most important to solve a certain problem, while the other will insist
on the centrality of solving a different problem. Kuhn poses the diffi-
culty in these terms: "if there were but one set of scientific problems,
one world within which to work on them, and one set of standards for
their solution, paradigm competition might be settledmore or less rou-
tinely by some process like counting the number of problems solved by
each. But, in fact, these conditions are never met completely. The
Dissecting the Holist Picture of Scientqic Change
A RETICULATIONAL
CRITIQUE OF REALIST
AXIOLOGY AND
METHODOLOGY
This essay was originally conceived as a book about how scientific theo-
ries are debated and evaluated. It is that, to be sure, but it must have
become increasingly evident that it is also a book about how to judge
philosophical doctrines about science. Every time a scientist seeks to
justify a theory choice by citing a methodological rule, or seeks to
make sense of a methodological rule by invoking a cognitive aim, he is
inevitably engaging in the philosophical tasks traditionally associated
with epistemology. This intimate involvement of epistemology in the
workaday life of the scientist decisively gives the lie to those who imag-
ine that scientists, especially natural scientists, have freed themselves
from the age-old moorings of their disciplines in philosophy. But it is
not my purpose here to argue for the blurring of any sharp distinction
between science and philosophy; I have already tried to state the case
for the merging of the two on numerous occasions.' What I do want to
do in this last chapter is to flesh out the model of scientific change
sketched out in earlier chapters by showing in practice and in detail
how one can go about critically evaluating proposals concerning epis-
temic aims and methods. I claim in chapter 3 that one can criticize a
cognitive goal or value in a variety of ways. The forms of criticism pro-
posed above include (a) showing that we are not in possession of means
whereby we can warrantedly ascertain whether the value has been real-
realist. I next seek to show that many of the realist's proposals about
the aims of science are too murky for us to implement them as scien-
tific values, precisely because we cannot ascertain when one theory is
truer than another. Leaving temporarily to one side the charge that
the realist's values are unsatisfiable, I go on to show that, if we were to
take seriously the proposals the realist makes about the goals and
values of scientific inquiry, then we should be forced to say that some
of the most impressive and successful pieces of science were (by the
realist's lights) unscientific. I conclude by arguing that, quite apart
from the dubious merits of realism itself, the realist's charge that the
only possible values for science are those he himself proposes is a non
sequitur.
But the targets of this chapter go beyond the proposed aims of scien-
tific realism. Realists have advocated a variety of methods and rules
for achieving their cognitive aims. Chief among these are two: (a) the
method of "inference to the best explanation," and ( b ) the method-
ological requirement that acceptable new theories must preserve sig-
nificant portions of the theoretical content (or extension) of their suc-
cessful predecessors. The former rule is designed to link pragmatic suc-
cess to epistemic warrant by virtue of claiming that a theory that ex-
hibits certain explanatory and predictive virtues can warrantably be
presumed to be true (or nearly true). The latter, if sound, is an impor-
tant principle governing the dynamics of intertheory relations. Utiliz-
ing techniques described above, I establish that the methods typically
advocated by realists have not been shown to stand in a relation to the
realist's aims which would justify those methodological rules as suitable
for realist axiology. In short, two principal rules of realist methodology
can be shown to fail to satisfy the demands we can legitimately make of
any adequate methodology of science.
1) Convergent realism. --Like other philosophical isms, the term
"realism" covers a variety of sins. Many of these are not at issue here.
For instance, semantic realism (in brief, the claim that all theories are
either true or false and that some theories -we know not which - are
true) is not in dispute. Nor do I discuss what one might call intentional
realism (i.e., the view that theories are generally intended by their pro-
ponents to assert the existence of entities corresponding to the terms in
those theories). What I do focus on instead are certain forms of episte-
mological realism. Put in its most general form, epistemic realism
amounts to the claim that certain forms of evidence or empirical sup-
port are so epistemically probative that any theory that exhibits them
106 A Reticulational Critique
2. Nor will I claim to do justice to the complex epistemologies of those whose work I
am about to criticize.
A Reticulational Critique 107
4. Putnam (1975, p. 69) insists, for instance, that if the realist is wrong about theo-
ries being referential, then "the success of science is a miracle."
5. Putnam, 1973, p. 21. Boyd remarks that "scientific realism offers an explanation
for the legitimacy of ontological commitment to theoretical entities" (Putnam, 1973,
q. 2, n. 10). It allegedly does so by explaining why theories containing theoretical enti-
tles work so well: because such entities genuinely exist.
A Reticulational Critique
6 . Whether one utilizes Putnam's earlier or later versions of realism is irrelevant for
the central arguments of this essay.
7 . Putnam, 1978, pp. 20-22.
A Reticulational Critique 111
8. Ibid., p. 22.
9. Ibid., p. 20.
A Reticulational Critique 115
10. Boyd, 1913, p. 1. Boyd also says (p. 3) that "experimental evidence for a theory
is evidence for the truth of even its non-observational laws." See also Sellars, 1963,
p. 97.
11. Certain latter-day realists (e.g., Glymour) want to break out of this holist web
and argue that certain components of theories can be directly tested. This approach,
whatever its merits, runs the very grave risk of undercutting what the realist desires
most: a rationale for taking our deepest-structure theories seriously, and a justifica-
tion for linking reference and success. After all, if the tests to which we subject our
theories only test portions of those theories, then even highly successful theories may
well have central terms that are nonreferring and central tenets that, because un-
tested, we have no legitimate grounds for believing to be approximately true. Under
those circumstances, a theory might be highly successful and yet contain important
constituents that were patently false. Such a state of affairs would wreak havoc with
the realist's presumption ( R I ) that success betokens approximate truth. In short, to be
less than a holist about theory testing is to put at risk precisely that predilection for
deep-structure claims which motivates much of the realist enterprise.
116 A Reticulational Critique
12. A caveat is in order here. Even if all the central terms in some theory refer, it is
not obvious that every rational successor to that theory must preserve all the referring
terms of its predecessor. One can easily imagine circumstances in which the new theory
is preferable to the old one even though the range of application of the new theory is
less broad than that of the old. When the ranee " is so restricted. it mav well be entirelv
appropriate to drop reference to some of the entities that figured in the earlier theory.
13. For Putnam and Boyd both "it will be a constraint on T ' [i.e., any new theory in
a domain]. . . that T ' must have this property, the property that f r o m its standpoint
one can assign referents to the terms of T [i.e., an earlier theory in the same domain]"
(Putnam, 1978, p. 22). Boyd says (1973, p. 8) that "new theories should, primafacie,
resemble current theories with respect to their accounts of causal relations among
theoretical entities."
A Reticulatonal Critique 117
ence for the prima facie referring terms in earlier theories is without
foundation. l4
More generally, we seem forced to say that such linkages as there are
between reference and success are rather murkier than Putnam's and
Boyd's discussions would lead us to believe. If the realist is going to
make his case for CER, it seems that it will have to hinge on approxi-
mate truth, R1, rather than reference, R2.
3) Approximate truth and success: the downward path. --Ignoring
the referential turn among certain recent realists, most realists con-
tinue to argue that, at bottom, epistemic realism is committed to the
view that successful scientific theories, even if strictly false, are none-
theless "approximately true" or "close to the truth" or "~erisimilar."'~
The claim usually amounts to this pair:
T1) if a theory is approximately true, then it will be explanatorily
successful; and
T2) if a theory is explanatorily successful, then it is probably ap-
proximately true.
What the realist would like to be able to say, of course, is:
Tl') if a theory is true, then it will be successful.
T I ' is attractive because self-evident. But most realists balk at invoking
T I ' because they are (rightly) reluctant to believe that we can reason-
ably presume that any given scientific theory is true. If all the realist
could explain was the success of theories that were true simpliciter, his
explanatory repertoire would be acutely limited. As an attractive move
in the direction of broader explanatory scope, T1 is rather more
appealing. After all, presumably many theories we believe to be false
(e.g., Newtonian mechanics, classical thermodynamics, wave optics)
were - and still are -highly successful across a broad range of appli-
cations.
Perhaps, the realist evidently conjectures, we can find an epistemic
16. Although Popper is usually careful not to assert that actual historical theories
exhibit ever-increasing truth content (for an exception, see Popper, 1963, p. 220),
other writers have been more bold. Thus, Newton-Smith writes that "the historically
generated sequence of theories of a mature science is a sequence in which succeeding
theories are increasing in truth content without increasing in falsity content" (forth-
coming, p. 2).
17. On the more technical side, Niiniluoto has shown that a theory's degree of cor-
roboration covaries with its "estimated verisimilitude" (1977; forthcoming). Roughly
A Reticulational Critique 119
22. A nonrealist might argue that a theory is approximately true just in case all ies
observable consequences are true or within a specific interval from the true value.
Theories that were "approximately true" in this sense would indeed be demonstrably
successful. But, the realist's (otherwise commendable) commitment to taking seriously
the theoretical claims of a theory precludes him from utilizing any such construal of
approximate truth, since he wants to say that the theoretical as well as the observa-
tional consequences are approximately true.
A Reticulational Critique 121
26. If this argument, which I attribute to the realists, seems a bit murky, I challenge
any reader to find a more clear-cut one in the literature. Overt formulations of this
position may be found in Putnam, Boyd, and Newton-Smith.
29. Putnam, 1998, p. 21.
A Reticulational Critique 125
like-minded convergentist, puts the point this way: "It typically hap-
pens in the history of science that when some hitherto dominant theory
T is superceded by T ' , T ' is in the relation of correspondence to T
[i.e., T is a limiting case of T']."Z8
Numerous recent philosophers of science have subscribed to a simi-
lar view, including Popper, Post, Krajewski, and Koertge.Zg This form
of retention is not the only one to have been widely discussed. Indeed,
realists have espoused a wide variety of claims about what is or should
be retained in the transition from a once successful predecessor (To) to
a successor (Tn) theory. Among the more important forms of realist
retention are the following: (1)T n entails To (Whewell); (2) T n retains
the true consequences or truth content of To (Popper); (3) T n retains
the "confirmed" portions of To (Post, Koertge); (4) T n preserves the
"theoretical laws and mechanisms" of To (Boyd, McMullin, Putnam);
(5) Tn preserves To as a limiting case (Watkins, Putnam, Krajewski);
(6) Tn explains why To succeeded insofar as To succeeded (Selfars);
(7) Tn retains reference for the central terms of To (Putnam, Boyd).
The question before us is whether, when retention is understood in any
of these senses, the realist's theses about convergence and retention are
correct.
5.1) Do scientists adopt the retentionist strategy of CER?-One part
of the convergent realist's argument is a claim to the effect that scien-
tists generally adopt the strategy of seeking to preserve earlier theories
in later ones. As Putnam puts it, "preserving the mechanisms of the
earlier theory as often as possible, which is what scientists try to do. . . .
That scientists try to do this. . . is a fact, and that this strategy has led
to important discoveries. . . is also a fact."g0In a similar vein, Szumi-
28. Watkins, 1978, pp. 376-377.
29. Popper (1959, p. 276) writes: "a theory which has been well corroborated can
only be superseded by one. . . [which] contains the old well-corroborated theory-or at
least a good approximation to it." Post (1971, p. 229) says: "I shall even claim that, as
a matter of empirical historical fact, [successor] theories [have] always explained the
whole of [the well-confirmed part of their predecessors]." And Koertge (1973, pp. 176-
177) writes: "nearly all pairs of successive theories in the history of science stand in a
correspondence relation a n d . . .where there is no correspondence to begin with, the
new theory will be developed in such a way that it comes more nearly into correspon-
dence with the old." Among other authors who have defended a similar view, one
should mention Fine (1967), Kordig (1971), Margenau (1950), and Sklar (1967).
30. Putnam, 1978, p. 20. Putnam fails to point out that it is also a fact that many
scientists do not seek to preserve earlier mechanisms and that theories that have not
preserved earlier theoretical mechanisms (whether the germ theory of disease, plate
tectonics, or wave optics) have led to important discoveries.
126 A Reticulational Critique
lewicz (although not stressing realism) insists that many eminent scien-
tists made it a main heuristic requirement of their research programs
that a new theory stand in a relation of correspondence with the theory
it supersede^.^^ If Putnam and the other retentionists are right about
the cognitive goals that most scientists have adopted, we should expect
to find the historical literature of science abundantly provided with
( a ) proofs that later theories do indeed contain earlier theories as lim-
iting cases, or ( b ) outright rejections of later theories that fail to con-
tain earlier theories. Except on rare occasions (coming primarily from
the history of mechanics), one finds neither of these concerns prorni-
nent in the literature of science. For instance, to the best of my knowl-
edge, literally no one criticized the wave theory of light because it did
not preserve the theoretical mechanisms of the earlier corpuscular
theory; no one faulted Eyell's uniformitarian geology on the grounds
that it dispensed with several causal processes prominent in catastroph-
ist geology; Darwin's theory was not criticized by most geologists for its
failure to retain many of the mechanisms of Lamarckian evolutionary
theory.
For all the realist's confident claims about the prevalence of a reten-
tionist value in the sciences, 1 am aware of no historical studies that
would sustain as a general thesis his hypothesis about the evaluative
strategies utilized in science. Moreover, insofar as Putnam and Boyd
claim to be offering "an explanation of the [retentionist] behavior of
scientist^,"^^ they have the wrong explanandum, for, if there is any
widespread strategy in science, it is one that says, "accept an empiri-
cally successful theory, regardless of whether it contains the theoretical
laws and mechanisms of its predecessor^."^^ Indeed, one could take a
leaf from the realist's C2 and claim that the success of the strategy of
assuming that earlier theories do not generally refer shows that it is
true that earlier theories generally do not.
(One might note in passing how often, and on what evidence, real-
ists imagine that they are speaking for the scientific majority. Putnam,
for instance, claims that "realism is, so to speak, 'science's philosophy
of science,' " and that "science taken at 'face value' implies
31. Szumilewicz, 1977.
32. Putnam, 1978, p. 21.
33. I have written a book about this strategy (Laudan, 1977).
34. After the epistemological and methodological battles about science during the
past three hundred years, it should be fairly clear that science, taken at its face value,
implies no particular epistemology.
A Reticulational Critique
38. This matter of limiting conditions consistent with the "reducing" theory is curi-
ous. Some of the best-known expositions of limiting-case relations depend (as Mrajew-
ski has observed) upon showing an earlier theory to be a limiting case of a later theory
only by adopting limiting assumptions explicitly denied by the later theory. For in-
stance, several standard textbook discussions present (a portion of) classical mechanics
as a limiting case of special relativity, provided c approaches infinity. But special rela-
tivity is committed to the claim that c is a constant. Is there not something suspicious
about a derivation of one theory from another which essentially involves a n assun~ption
inconsistent with the deriving theory? If the deriving theory is correct, then it specifi-
cally denies the adoption of a premise commonly used to prove the derived theory as a
limiting case. (It should be noted that most such proofs can be reformulated unobjec-
tionably, e.g., in the relativity case, by letting v--+ 0 rather than c-4-.)
A Reticulational Critique
Readers who have followed the argument this far may well be expect-
ing me finally to deliver the goods by stating what the central values,
aims, and methods of science are, or at least what they should be. Any
such expectations will, I am afraid, be dashed. To lay out a set of cog-
nitive aims and methods and to say "those are what science is about"
would be to undermine much of the foregoing analysis, for we have
seen time and again that the aims of science vary, and quite appropri-
ately so, from one epoch to another, from one scientific field to an-
other, and sometimes among researchers in the same field. We have
also seen that, even among researchers who share the same aims or val-
ues, methods may legitimately differ.
What I do claim to have exhibited here is some of the analytic
machinery we can bring to bear in assessing proposed aims and meth-
ods. More than that, I have shown by example how that machinery
can be applied to the critique of some popular and initially plausible,
but ultimately unsatisfactory, axiologies of science (e.g., scientific real-
ism and epistemic relativism).
One is tempted to speculate whether the model of methodological
and axiological critique sketched out here could be adapted to deal
with extrascientific axiologies, such as those of moral theory. For my
part, I have resisted the temptation to draw out several of the apparent
parallels between debates about cognitive values and debates about
moral or political values. The reason is straightforward: one fights one
battle at a time. If the analysis of cognitive values developed here
Epilogue
Gutting, Gary, ed. (1980). Paradigms and Revolutions. Notre Dame: Univer-
sity of Notre Dame Press.
Wartley, David (1749). Observations o n M a n . London.
Hooker, Clifford (1974). "Systematic Realism," Synthese 26:409-497.
Iiull, David, et al. (1978). 'TPlanck's Principle," Science 202:717-723.
Koertge, Noretta (1973). '"Theory Change in Science." I n G. Peance and
P. Maynard, eds., Conceptual Change. Dordrecht: Reidel. Pp. 167-198.
Rordig, Carl (1971). "Scientific Transitions, Meaning Invariance, and Deriv-
ability," Southern Journal of Philosophy 1 : 119-125.
Krajewski, W. (1977). Correspondence Principle and Growth of Science. Dor-
drecht: Reidel.
Ruhn, Thomas (1962). T h e Structure of Scientific Revolutions. Chicago:
University of Chicago Press.
-- (1970). "Reflections on My Critics." I n it. Lakatos and A. Musgrave,
Criticism and t h e Growth of Knowledge. Cambridge: Cambridge Univer-
sity Press.
(1977). T h e Essential Tension. Chicago: University of Chicago Press.
Eakatos, Imre (1978). T h e Methodology of Scientific Research Programmes.
Cambridge: Cambridge University Press.
Laudan, Larry (1968). "Theories of Scientific Method from Plato to Mach,"
History of Science 7: 1-63.
(1976). "Two Dogmas of Methodology," Phiiosophy of Science 43:
467-472.
(1977). Progress and Its Problems. Berkeley, Los Angeles, London:
University of California Press.
-- (1978). "Ex-Huming Hacking," Erkenntnis 13:417-435.
(1981). Science and Hypothesis. Dordrecht: Reidel.
-- (1982). "The Demise of Demarcation." I n R. S. Cohen and L. Lau-
dan, eds., Physics, Philosophy and Psychiatry. Dordrecht: Reidel. Pp. 111-
128.
-- (forthcoming). Science and Method.
Eugg, Andrew (1978). "Disagreement in Science," Zeitschrift ffiir allgemeine
Wissenschaftstheorie 9:276-292.
McMullin, Ernan (1970). "The History and Philosophy of Science: A Taxon-
omy." I n R. Stuewer, ed., Minnesota Studies in the Philosophy of Science
V :12-67. Minneapolis: University of Minnesota Press.
Margenau, Henry (1950). T h e Nature of Physical Reality. New York:
McGraw-Hill.
Merton, Robert (1968). Social Theory and Social Structure. New York: Free
Press.
(1973). Sociology of Science. Chicago: University of Chicago Press.
Mitroff, Ian (1974). T h e Subjective Side of Science. New York: Elsevier.
References 143
58, 59; intelligible, 60-61, 84; on/of science, 1-2, 3-4, 9-12, 16,
limiting cases in, 107, 108, 121, 20, 21-22, 44
124-125, 126, 127, 128, 130, 132, Success: and approximate truth,
136; mature, 81, 106, 107, 108, 117-118, 119, 120-124, 133, 136;
109, 114, 121-122, 124, 130-131, in science, 107-108, 109-117,
132; normal, 17-18, 19; observ- 120-121, 122, 123, 124, 132-133,
able v. unobservable entities in, 134, 135, 137
23, 55-59, 81; philosophy of/on, Szumilewicz, I., 125-126
1-2, 3-4, 5-8, 9, 11, 12, 16, 21-22,
44, 47, 48, 67, 103; progress in, Testing, 25, 39
64-66, 107-108; as quest for cer- Toulmin, Stephen, 68
tainty, 83; retention in, 105, 107,
116, 124-125, 126-127, 131-132; Underdetermination, 13, 15-16, 84-
sociology nf/on, 1-2, 3-4, 9-12, 85, 86; of factual choices, 27, 28-
16, 20, 21-22, 44; success in, 107- 30, 31, 32, 33; Kuhn on, 88-102;
108, 109-117, 120-121, 122, 123, local, 88-102; in methodology,
124, 132-133, 134, 135, 137; veri- 35-36, 39-40, 44; of theories, 27,
similitude in, 117-120, 134. See 87-88
also Axiology; Consensus; Dissen- 1Jniversalism, 9
sus; Methodology Utopianism, 50, 51-53, 61
Scientific method. See Methodol-
ogy/methodological rules Values. See Axiology
Scientific rationality, xii-xiii, 3 , 41, Verisimilitude, 117-120, 134
47, 48, 63, 64
Scientific realism, 104-137 Watkins, John, 124-125, 127, 131
Sellars, Wilfrid, 104, 109, 125, 132 Whewell, William, 59, 80, 81, 125
Skepticism, organized, 9, 11 n. 3
Sociology: on axiology, 44; dissen- Ziman, John, 4
sus in, 4, 9-10; norms of, 9-19; Zuckerman, Harriet, 9