Progress and its Problems

Larry Laudan

ISBN: 0520037219
To Rachel, Heather & Kevin—
fellow pilgrims
 Contents

Preface ix
Prologue 1
Part One: A Model of Scientific Progress
1. THE ROLE OF EMPIRICAL PROBLEMS 11
The Nature of Scientific Problems—Empirical Problems—Types of
Empirical Problems—The Status of Unsolved Problems—The Na-
ture of Solved Problems—The Special Role of Anomalous Problems
—Converting Anomalies to Solved Problems—The Weighting of
Empirical Problems—Theory Complexes and Scientific Problems

2. CONCEPTUAL PROBLEMS 45
The Nature of Conceptual Problems—The Sources of Conceptual
Problems—The Relative Weighting of Conceptual Problems—Sum-
mary and Overview

3. FROM THEORIES TO RESEARCH TRADITIONS 70

Kuhn's Theory of Scientific ‘‘Paradigms’’—Lakatos’ Theory of “Re-
search Programmes’’—The Nature of Research Traditions— Theories
and Research Traditions—-The Separability of Theories from Re-
search Traditions—The Evolution of Research Traditions— Research
Traditions and Changes in Worldviews—The Integration of Re-
search Traditions—‘‘Nonstandard”’ Research Traditions—The Eval-
uation of Research Traditions—Adhocness and the Evolution of
Research Traditions—Anomalies Revisited—Summary: A General
Characterization of Scientific Change

4. PROGRESS AND REVOLUTION 121

Progress and Scientific Rationality—Scientific Revolutions—Revolu-
tion, Continuity, and Commensurability—Non-Cumulative Progress
—In Defense of ‘“‘Immature”’ Science
Part Two: Applications
5. HISTORY AND PHILOSOPHY OF SCIENCE 155
The Role of History in the Philosophy of Science—The Role of
Norms in the History of Science—Rational Appraisal and ‘‘Rational
Reconstruction”

6. THE HISTORY OF IDEAS 171

Disciplinary Autonomy and the History of Ideas—Ideas and their
Problem Contexts—The Aims and Tools of Intellectual History—
Problem Solving and Nonscientific Research Traditions—The In-
dispensability of History for Theory Appraisal

7. RATIONALITY AND THE SOCIOLOGY OF 196

KNOWLEDGE
The Domain of Cognitive Sociology—The Theoretical Foundations
of Cognitive Sociology—Conclusion

Epilogue: Beyond Veritas and Praxis 223

Notes 227
Bibliography 247
Index of Names 255
 Preface

It has been my good fortune to have been student or colleague

to many of the scholars whose work has done much to shape
the character of contemporary history and philosophy of
science: C. G. Hempel, T. S. Kuhn, Gerd Buchdahl, Paul
Feyerabend, Karl Popper, Imre Lakatos and Adolf Griinbaum
have all left their mark on the eclectic doctrines that make up
this essay. If the book is persistently critical of some of their
ideas, it is because healthy disagreement (unlike imitation) is
the deepest sign of an abiding admiration. Unfortunately, it is
no longer possible for me to acknowledgespecifically how much
my approachto science owes to each of these thinkers; but the
collective debt is enormous. What originality there may be in
this essay derives almost entirely from the insights (and in some
cases the pregnant confusions) to be found in their writings.
Other forms of indebtedness are, however, easier to localize.
Research grants from the National Science Foundation, from
the German Fulbright-Kommission and from the University of
Pittsburgh provided the release time necessary to undertake this
project. The hospitality of the University of Konstanz provided
a congenial atmosphere for putting to paper ideas which had
been percolating in my seminars since 1970. Cindy Brennan and
Karla Goldman provided yeoman service in preparing the
manuscript. Specific portions of earlier drafts of this essay have
been profitably discussed with A. Griinbaum, D. Hull, J. E.
McGuire, K. Schaffner, M. J. S. Hodge, M. and R. Nye, I.

ix
xX PREFACE

Mitroff, P. Machamer, N. Rescher, R. Creath, A. G. Molland,

S. Wykstra, F. Kambartel, J. Mittelstrass, P. Janich, and J. M.
Nicholas. The book would be far more flawed than it is without
their criticism and suggestions. My greatest debt, however, is to
Rachel, whose patience, critical sense, and unflagging encour-
agement sustained this project through its difficult incubation
period.

June, 1976
 Prologue
We must explain why science—
our surest example of sound knowledge—
progresses as it does, and wefirst must find out how,
in fact, it does progress. T. S. KUHN (1970), p. 20

Epistemology is an old subject; until about 1920, it was also a

great one. What produced the change was a confluenceof three
quite separate developments, each of which effected a profound
transformation in the study of knowledge. There was, first of
all, the crisis produced by the realization that knowledge was
neither as certain nor as incorrigible as thinkers since Plato and
Aristotle had presumed it to be. There was, secondly, the
increasing professional insularity of academic philosophers, and
their related conviction that disciplines such as psychology and
sociology, which had played a major role in earlier epistemo-
logical theories, had no interesting insights to offer. (This
insularity was further promoted by the guileless duplicity of
scholars in other fields, who were all too prepared to bequeath
“the problem of knowledge’”’ to the professional philosophers.)
There was, finally and catastrophically, a growing tendency
(especially in the English-speaking world) to imagine that one
could grapple with the nature of knowledge while remaining
blissfully ignorant of its best extant example—the natural
sciences.
Despite the attempted appropriation of epistemological issues
by the professional philosophers, many of the classical questions

1
2 PROLOGUE
about the nature of scientific knowledge still remain of broad,
general interest: Does science progress? Are our ideas about
nature really worthy of credence? Are some beliefs about the
world more rational than others? Issues such as these go well
beyond the boundsofspecialized disciplinary monopolies. They
do so in large part because most people in the West draw the
bulk of their beliefs about nature, and even about themselves,
from the corpus of science. Without Newton, Darwin, Freud,
and Marx (to mention only the more obvious), our picture of the
world would be vastly different from what it is. If science is a
rationally well-founded system of inquiry, then it is only right
and proper that we should emulate its methods, accept its
conclusions, and adopt its presuppositions. If, however, science
is predominately irrational, then there is no reason to take its
knowledge claims any more (or less) seriously than we take
those of the seer, the religious prophet, the guru, or the local
fortuneteller.
For a long time, many have taken the rationality and
progressiveness of science as an obvious fact or a foregone
conclusion, and somereaders will probably still think it bizarre
to believe that there is any important problem to be solved here.
Although this confident attitude has been almost inescapable
given the cultural biases in favor of science in modern culture,
there have been a number of recent developments which bring
it into serious question:
1. Philosophers of science, whose primary aim is to define
what rationality is, have generally found that their models of
rationality find few, if any, exemplifications in the actual
process of scientific activity.’ If we accept the claim made on
behalf of these models to the effect that they define rationality
itself, then we seem forced to view virtually all science as
irrational.
2. Attempts to show that the methodsof science guaranteeit
is true, probable, progressive, or highly confirmed knowledge—
attempts which have an almost continuous ancestry from
Aristotle to our own time—have generally failed,? raising a
distinct presumption that scientific theories are neither true,
nor probable, nor progressive, nor highly confirmed.
 3
PROLOGUE

3. Sociologists of science have been able to point to several

episodes in the recent (and distant) past of science which seem
to reveal many nonrational, or irrational, factors decisively
involved in scientific decision making.’
4. Some historians and philosophers of science (e.g., Kuhn
and Feyerabend) have argued, not merely that certain decisions
between theories in science have been irrational, but that
choices between competing scientific theories, in the nature of
the case, must be irrational.‘ They (especially Kuhn) have also
suggested that every gain in our knowledge is accompanied by
attendant losses, so that it is impossible to ascertain when, or
even whether, we are progressing. °
The skepticism to which such conclusions point has been
reinforced by the general arguments of cultural relativism to the
effect that science is just one set of beliefs among manypossible
ones, and that we in the West venerate science, not becauseit is
morerational than its alternatives, but simply because we are a
product of a culture that has traditionally set great store by
science. All systems of belief, including science, are seen as
dogmas and ideologies, between which objective, rational
preference is impossible.
Confronted by the acknowledged failure of the traditional
analysis to shed much light on the rationality of knowledge,
three alternatives seem to be open to us:
1. We might continue to hope that some as yet undiscovered
minor variation in the traditional analysis will eventually clarify
and justify our intuitions about the cognitive weil-foundedness
of science and thus prove to be a worthy model of rationality.
2. We might, alternatively, abandon the search for an
adequate model of rationality as a lost cause, thereby accepting
the thesis that science is, so far as we know, blatantly
irrational.
3. Finally, we might begin afresh to analyze the rationality of
science, deliberately trying to avoid some of the key presupposi-
tions which have produced the breakdown of the traditional
analysis.
Enormousefforts have been devoted, particularly in the last
decade, to the pursuit of strategies (1) and (2). Philosophers of
4 PROLOGUE

science, by and large, have taken the first option. Thus,

Lakatos asks, ‘“‘What are the minimum changes needed in the
Popperian analysis of science to enable it to solve the problem of
rationality?’’® Salmon asks, ‘“‘What are the minimum adjust-
ments needed in Reichenbach’s theory to square it with
scientific practice?’ Hintikka poses the question, ‘““What kind
of tinkering with Carnap’s inductive logic will make it relevant
to scientific testing?”” While one admires the tenacity and
ingenuity illustrated by proponents of this approach, the results
are not, on the whole, very encouraging. Most of the difficulties
which stood in the way of a Popper, a Carnap, or a Reichen-
bachstill remain obstacles for their latter-day disciples.’
The second option has proved more popular with historically
oriented thinkers. Thus, both Kuhn and Feyerabend conclude
that scientific decision making is basically a political and
propagandistic affair, in which prestige, power, age, and
polemic decisively determine the outcome of the struggle
between competing theories and theorists. Their mistake seems
to be one of jumping to a premature conclusion. They start
from the premise that rationality is exhaustively defined by a
certain model of rationality (each of them takes Popper's
model offalsifiability as the archetype). Having observed, quite
correctly, that the Popperian model of rationality will do scant
justice to actual science, they precipitately conclude that science
must have large irrational elements, without stopping to consider
whether some richer and more subtle model of rationality
might do the job.
Because the one option seems unpromising and the other
premature, I am inclined to think that we should consider
pursuing the third strategy. Let us drop someofthe traditional
language and concepts (degree of confirmation, explanatory
content, corroboration and the like), and see if a potentially
more adequate model of scientific rationality begins to emerge.
Let us see whether, by asking anew some of the elementary
questions about science, we cannot get a slightly different
perspective on scientific knowledge.
In whatfollows, I shall attempt to trace out the consequences
of the view that science fundamentally aims at the solution of
PROLOGUE 5
problems. Although the view itself is commonplace, very little
attention has been given to exploring it in detail. What the
different types of problems are, what makes one problem more
important than another, the criteria for counting something as
an adequate solution, the relation of nonscientific problems to
scientific ones; none of these issues has been addressed in the
detail it demands. To anticipate some of my conclusions, I
propose that the rationality and progressiveness of a theory are
most closely linked—not with its confirmation or its falsifica-
tion—but rather with its problem solving effectiveness. | shall
be arguing that there are important nonempirical, even ‘‘non-
scientific’’ (in the usual sense), factors which have—and which
should have—played a role in the rational development of
science. I shall suggest, further, that most philosophers of
science have mistakenly identified the nature of scientific
appraisal, and thereby the primary unit of rational analysis, by
focussing on the individual theory, rather than on what I call
the research tradition. This study will show, moreover, that we
need to distinguish between the rationality of acceptance and
the rationality of pursuit if we are to make any progress at
reconstructing the cognitive dimensions of scientific activity.
Mybasic strategy in what follows will involve the blurring,
and perhaps the obliteration, of the classical distinction be-
tween scientific progress and scientific rationality. These two
notions, both central to any discussion of science, have often
seemed at cross purposes. Progress is an unavoidably temporal
concept; to speak about scientific progress necessarily involves
the idea of a process occurring through time. Rationality, on the
other hand, has tended to be viewed as an atemporal concept;
it has been claimed that we can determine whether a statement
or theory is rationally credible independently of any knowledge
of its historical career. Insofar as rationality and progressiveness
have been linked at all, the former has taken priority over the
latter—to such a degree that most writers see progress as
nothing more than the temporal projection of a series of
individual rational choices. To be progressive, on the usual
view, is to adhere to a series of increasingly rational beliefs. I
am deeply troubled by the unanimity with which philosophers
6 PROLOGUE

have made progress parasitic upon rationality. In part, my

worry arises from a concern that it involves explaining some-
thing which can be readily understood (progress) in terms of
something else (rationality) which may be far more obscure.
More serious, however, is the absence of any convincing
argument as to why we should explicate our concept of progress
in terms of rationality. The two concepts are doubtless related,
but not necessarily in the manner usually supposed.
It will be the assumption here that we may be able to learn
something by inverting the presumed dependencyof progress on
rationality. I shall try to show that we have a clearer model for
scientific progress than we do for scientific rationality; that,
moreover, we can define rational acceptance in terms of
scientific progress. In a phrase, my proposal will be that
rationality consists in making the most progressive theory
choices, not that progress consists in accepting successively the
most rational theories. This inversion of the usual hierarchy
offers some insights into the nature of science which tend to
elude us if we preserve the traditional relation between progress
and rationality.
Anotherof the chief obstacles to the development of a theory
of scientific progress has been the universal assumption that
progress can occur only if it is cumulative, that is, if knowledge
grows entirely by accretion. Because there are grave difficulties,
both historically and conceptually, with the progress-by-accre-
tion view, I propose a definition of scientific progress which
does not demand cumulative development.
In order for the ambitions of this enterprise to be brought to
fruition, and to prevent its being misconstrued, two key points
must be stressed. First, the term ‘‘progress’’ has many emotive
overtones deeply rooted in the subjective intuitions of both
friends and critics of science. The object of this work is not to
exploit that emotiveness, but rather to offer objective criteria for
determining when progress has occurred. In too many discus-
sions of progress, insufficient attention has been given to
separating out the question of what progress is from the
question of its moral and cognitive desirability. Any adequate
theory of progress must make such a distinction as sharply as
possible. There is a second crucial ambiguity in normal usages
PROLOGUE 7

of “progress’”” which must also be noted. Specifically, it is

commonplace to speak of progress, meaning an improvement in
the material or the ‘‘spiritual”’ conditions of life. Although that
sense of progress is unquestionably important, I shall say
virtually nothing about it in this essay. My exclusive preoccupa-
tion will be with what I-call “cognitive progress,"’ which is
nothing more nor less than progress with respect to the
intellectual aspirations of science. Cognitive progress neither
entails, nor is it entailed by, material, social, or spiritual
progress. These notions are surely not altogether disconnected,
but they do refer to very different processes, and, at least for
purposes of the present discussion, should be sharply
distinguished.
Onefinal point is in order. Previously, too many discussions
of scientific rationality and progress have been both uninformed
by, and inapplicable to, the actual course of the evolution of
science, The various well-known philosophical models of ration-
ality have been shown to be inapplicable to most of those cases
in the history of science where, at least intuitively, we are
convinced that sensible, rational choices were being made.
Without assuming that whatever science does is, by definition,
rational, we should nonetheless be able to demand of any model
of science that it substantially ‘“‘fit’’ the actual course of
scientific change. Accordingly, historical cases and episodes
will be used extensively in this essay; these are intended not
merely to dlustrate my philosophical claims, but also to test
them. If the model under discussion here fails to illustrate the
mannerin which scientific decision making has actually worked
(at least some of the time), then it will have failed entirely in
its ambitions.
Because of the unusually heavy weight attached in_ this
approach to historical material—material which some philoso-
phers deem to be absolutely irrelevant to epistemology—I shall
also briefly discuss the general question of the bearing of
descriptive data (such as history) on a normative theory (such as
a model ofscientific rationality).
Part One of the following study articulates a model of
scientific progress and rationality, and exhibits how that model,
for all its evident incompleteness, avoids many of the paradoxes
8 PROLOGUE

which previous models have generated, and makes some sense

of the historical data. Part Two examines the ramifications of
that model for a variety of intellectual inquiries, ranging from
the history of ideas to the history and philosophy of science and
the sociology of knowledge.
It has not been possible for me to explore all the issues
concerned with scientific progress in the detail which they
deserve. For that failure, I can only ask the reader’s mercy. This
is not, nor is it intended to be, a finished piece of work. At
many points, argument sketches pass for arguments and plaus-
ible intuitions are invoked where, ideally, explicit doctrines
are called for. A great deal remains to be said on ail the
matters I address. But the study of rational knowledge and its
growth, like knowledge itself, is a cooperative venture of a
community of minds. My purpose is merely to offer a fresh
perspective on some problems which have preoccupied reflective
people for a very long time.
 Part One

A Modelof Scientific
Progress
The activity of understanding
Is, essentially, the same as
that of all problem solving. K. POPPER (1972), p. 166
 Chapter One
The Role of Empirical Problems
Problem formulation in science
is to be understood by looking
at the continuity of the whole
stream of scientific endeavor. H. SIMON (1966), p. 37

Science is essentially a problem-solving activity. This anodine

bromide, more a cliché than a philosophy of science, has been
espoused by generations of science textbook writers and self-
professed specialists on ‘‘the scientific method.” But for all the
lip service which has been paid to the view that science is
fundamentally the solving of problems, scant attention has been
paid, either by philosophers of science or historians of science,
to the ramifications of such an approach for understanding
science.' Philosophers of science, by and large, have imagined
that they can lay bare the rationality of science by ignoring, in
their analyses, the fact that scientific theories are usually
attempts to solve specific empirical problems about the natural
world.” Similarly, historians of science, for their part, have
usually imagined that the chronology of scientific theories
possesses an intrinsic intelligibility which requires little or no
cognizance of the particular problems which prominent theories
in the past were designedto solve.

il
12 ROLE OF EMPIRICAL PROBLEMS

It is the purpose of this short book to sketch what seem to be

the implications, for both the history of science and_ its
philosophy, of a view of scientific inquiry which perceives
science as being—aboveall else—a problem-solving activity.
The approach taken here is not meant to imply that science is
‘nothing but’ a problem-solving activity. Scfence has as wide a
variety of aims as individual scientists have a multitude of
motivations: science aims to explain and control the natural
world; scientists seek (among other things) truth, influence,
social utility, and prestige. Each of these goals could be (and
has been) used to provide a framework within which one might
try to explain the development and nature of science. My
approach, however, contends that a view of science as a
problem-solving system holds out more hope of capturing what
is most characteristic about science than any alternative
framework has.
As it becomes clear that many of the classic problems of
philosophy of science, and many of the standard issues of the
history of science, take on a very different perspective when we
look at science as a problem-solving and problem-oriented
activity, it will be argued that an attentive analysis of science
from this perspective generates new insights which run counter
to much of the ‘“‘conventional wisdom’’ which historians and
philosophersof science have taken for granted.
There is nothing modest about the claims this study makes.
In brief, I shall be suggesting that a sophisticated theory of
science qua problem-solving activity must alter the way we
perceive both the central issues in the historiography of science
and the central problems in the philosophy or methodology of
science. I shall argue that if we take seriously the doctrine that
the aim of science (and of all intellectual inquiry, for that
matter) is the resolution or clarification of problems, then we
shall have a very different picture of the historical evolution and
the cognitive evaluation of science.
Before I contrast the problem-solving view of science with
certain better known philosophies and histories of science, I
must indicate specifically what I mean by a ‘“‘problem-oriented
theory of science.” It is this preliminary goal which this chapter
and the next aim to achieve.
ROLE OF EMPIRICAL PROBLEMS 13

The Nature of Scientific Problems

Throughout this essay, I shall be talking about what I call
scientific problems. 1 should stress at the outset that I do not
believe that “‘scientific’’ problems are fundamentally different
from other kinds of problems (though they often are different
in degree). Indeed, I shall show in chapter six that the view I
am espousing can be applied, with only a few qualifications, to
all intellectual disciplines. But, if we wish to study problem
solving, we ought to begin with its most successful instances; so
I shail limit my remarks in these preliminary sections largely to
scienceitself.
If problems are the focal point of scientific thought, theories
are its end result. Theories matter, they are cognitively
important, insofar as—and only insofar as—they provide
adequate solutions to problems. If problems constitute the
questions of science, it is theories which constitute the answers.
The function of a theory is to resolve ambiguity, to reduce
irregularity to uniformity, to show that what happens is
somehow intelligible and predictable; it is this complex of
functions to which I refer when I speak of theories as solutions
to problems.

Thesis 1: The first and essential acid test for any theory is
whether it provides acceptable answers to interesting
questions: whether, in other words, it provides satis-
factory solutions to important problems.

At one level, this might appear completely noncontroversial.

Most writers who have dealt with the nature of science would
probably claim to subscribe to such a view. Unfortunately, as
we shall see, most philosophies of science manifestly fail to go so
far as to justify even that seemingly harmless and obvious
sentiment, let alone to explore its many ramifications.
The literature of the methodology of science offers us neither
a taxonomy of the types of scientific problems, nor any
acceptable method of grading their relative importance. It is
noticeably silent about what the criteria are for an adequate
solution to a problem. It does not recognize there are degrees of
adequacy in problem solution, some solutions being better and
 14 ROLE OF EMPIRICAL PROBLEMS

richer than others. Insofar as contemporary philosophy of

science says anything at all about these matters, it tends to
regard all solutions on a par, and to assign all problems equal
weight. In assessing the adequacy of any theory, the philoso-
pher of science will usually ask how many facts confirm it,
not how important those facts are. He will ask how many
problems the theory solves, not about the significance of those
problems. To this extent, contemporary philosophy of science
has not captured the sense of thesis (1) above. It is for reasons
such as these that I propose:
Thesis 2: In appraising the merits of theories, it is more
important to ask whether they constitute adequate
solutions to significant problems than it is to ask
whether they are ‘‘true,"’ ‘‘corroborated,"' ‘‘well-con-
firmed” or otherwise justifiable within the framework
of contemporary epistemology.
Butif it is plausible to think that the counterpoint between
challenging problems and adequate theories is the basic dialec-
tic of science, we must get a great deal clearer than we now are
about what problems are and how they work, about how
problems are weighted, and about the nature of theories and
their precise relation to the problems which generate them (and
which, as we shall see, they sometimesgenerate).

Empirica] Problems
There are two very different kinds of problems which
scientific theories are designed to solve. For now, I want to
focus on the first, more familiar and archetypal, sense of the
concept, which I shall call an empirical problem. Empirical
problemsare easierto illustrate than to define. We observe that
heavy bodies fall toward the earth with amazing regularity. To
ask how and why theyso fall is to pose such a problem. We
observe that alcohol left standing in a glass soon disappears. To
seek an explanation for that phenomenon is, again, to raise an
empirical problem. We may observe that the offspring of plants
and animals bear striking resemblances to their parents. To
inquire into the mechanism of trait transmission is also to raise

rm
 ROLE OF EMPIRICAL PROBLEMS
1S
an empirical problem. More generally, anything about the
natural world which strikes us as odd, or otherwise in need
of
explanation, constitutes an empirical problem.
In calling such inquiry situations “empirical” problems, [ do
not mean to suggest they are directly given by the world as
veridical bits of unambiguous data. Both historical examples
and recent philosophical analysis have made it clear that the
world is always perceived through the “‘lenses’” of some
conceptual network or other and that such networks and the
languages in which they are embedded may, for all we know,
provide an ineliminable ‘tint’ to what we perceive. More to the
point, problems of all sorts (including empirical ones) arise
within a certain context of inquiry and are partly defined by
that context. Our theoretical presuppositions about the natural
order tell us what to expect and what seems peculiar or
“problematic” or questionable (in the literal sense of that term).
Situations which pose problems within one inquiry context will
not necessarily do so within others. Hence, whether something
is regarded as an empirical problem will depend, in part, on the
theories we possess.
Why, then, call them ‘empirical’ problems at all? I do so
because, even granting that they arise only in certain contexts
of theoretical inquiry, even granting that their formulation will
be influenced by our theoretical commitments, it is nonetheless
the case that we treat empirical problems as if they were
problems about the world. If we ask, “How fast do bodies fall
near the earth?”’, we are assumingthere are objects akin to our
conceptions of body and earth which move towards one another
according to some regular rule. That assumption, of course, is
a theory-laden one, but we nonethelessassert it to be about the
physical world. Empirical problems are thus first order prob-
lems; they are substantive questions about the objects which
constitute the domain of any given science. Unlike other, higher
order problems (to be discussed in chapter two), we judge the
adequacy of solutions to empirical problems by studying the
objects in the domain.
We have already noted that there is an apparent functional
similarity between talk of problems and problem solving and the
more familiar rhetoric about facts and the explanation of facts.
16 ROLE OF EMPIRICAL PROBLEMS

Given that similarity, one might be inclined to translate the

claims ] shall make about the nature and logic of problem
solving into assertions about the logic of explanation. To do so,
however, would be to misconstrue the enterprise, for problems
are very different from ‘‘facts” (even “‘theory-laden facts’) and
solving a problem can not be reduced to “explaining a fact.”
Full discussion of the disanalogies must wait until later, but
some of the discrepancies can be seen by considering a few of
the differences between facts or states of affairs on the one
hand, and empirical problems on the other.
Certain presumed states of affairs regarded as posing empiri-
cal problems are actually counterfactual. A problem need not
accurately describe a real state of affairs to be a problem: all
that is required is that it be thought to be an actual state of
affairs by some agent. For instance, early members of the Royal
Society of London, convinced by mariners’ tales of the existence
of sea serpents, regarded the properties and behavior of such
serpents as an empirical problem to be solved. Medieval natural
philosophers such as Oresme, took it to be the case that hot
goat’s blood could split diamonds and developed theories to
explain this counterfactual empirical ‘‘occurrence.”’* Similarly,
early nineteenth century biologists, convinced of the existence of
spontaneous generation, took it to be an empirical problem to
show how meatleft in the sun could transmute into maggots or
how stomach juices could turn into tapeworms. For centuries,
medical theory sought to explain the ‘“‘fact’”’ that bloodletting
cured certain diseases. If factuality were a necessary condition
for something to count as an empirical problem, then such
situations could not count as problems. So long as weinsist that
theories are designed only to explain ‘‘facts’’ (i.e., true state-
ments about the world), we shall find ourselves unable to
explain most of the theoretical activity which has taken place in
science.
There are many facts about the world which do not pose
empirical problems simply because they are unknown. It is,
for instance, presumably a fact (and always has been) that the
sun is composed chiefly of hydrogen; but until the fact was
discovered (or invented), it could not have generated a problem.
In sum, a fact only becomes a problem when it is treated and
 ROLE OF EMPIRICAL PROBLEMS
17
recognized as such; facts, on the other hand, are facts, whether
they are ever recognized. The only kind of facts which can
possibly count as problems are known facts.
But even many known facts do not necessarily constitute
empirical problems. To regard something as an empirical
problem, we must feel that there is a premium on solving it. At
any given moment in the history of science, many things will be
well-known phenomena, but will not be felt to be in need of
explanation or clarification. It was known since the earliest
times, for instance, that most trees have green leaves. But that
“fact” only became an “empirical problem’’ when someone
decided it was sufficiently interesting and important to deserve
explanation. Again, early societies knew certain drugs could
produce hallucinations, but that widely known fact only became
a recognized problem for physiological theories relatively
recently.
Finally, problems recognized as such at one time can, for
perfectly rational reasons, cease to be problems at later times.
Facts could never undergo that sort of transformation. Early
geological theorists, for instance, regarded one of the central
problems of their discipline to be that of explaining how the
earth took its shape within the last 6,000 to 8,000 years. With
the elongation of the geological time scale, that staggering issue
no longer remained a problem to be solved.

Types of Empirical Problems

Having seen some* of the differences between facts and
empirical problems and the need for clearly separating the two,
we can now turn to the role which such problems play in the
process of scientific analysis. Although a fuller taxonomy will
be developed later, we can roughly divide empirical problems
into three types, relative to the function they have in theory
evaluation: (1) unsolved problems—those empirical problems
which have not yet been adequately solved by any theory; 5 (2)
solved problems—those empirical problems which have been
adequately solved by a theory; (3) anomalous problems—those
empirical problems which a particular theory has not solved,
but which one or moreof its competitors have.°
18 ROLE OF EMPIRICAL PROBLEMS

Clearly, solved problems count in favor of a theory, anoma-

lous ones constitute evidence against a theory, and unsolved
ones simply indicate lines for future theoretical inquiry. Using
this terminology, we can argue that one of the hallmarks of
scientific progress is the transformation of anomalous and
unsolved empirical problems into solvéd ones. Of any and
every theory, we must ask how many problemsit has solved and
how many anomalies confront it. This question, in a slightly
more complex form, becomes one of the primary tools for the
comparative evaluation of scientific theories.

The Status of Unsolved Problems

It is part of the conventional wisdom that unsolved problems
provide the stimulus for scientific growth and progress; and
there can be no doubt that transforming unsolved into solved
problems is one (though by no means the only) way in which
progressive theories establish their scientific credentials. But it
is too often assumed that the body of unsolved problems at any
given time is clear cut and well defined, that scientists have a
definite sense of which unsolved problems should be solved by
their theories, and that a theory’s failure to digest its unsolved
problemsis a clear liability.
A careful examination of many historical cases reveals,
however, that the status of unsolved problems is a great deal
more ambiguous than is often imagined. Whether a given
‘‘phenomenon” is a genuine problem, how important it is, how
heavily it counts against a theory if it fails to solve it; these are
all very complex questions, but a good first approximation to an
answer is this: unsolved problems generally count as genuine
problems only when they are no longer unsolved. Until solved
by some theory in a domain they are generally only ‘potential’
problems rather than actual ones.’ There are two factors
chiefly responsible for this: one, which we have already dis-
cussed, arises when we are unsure an empirical effect is
genuine. Because many experimenta! results are difficult to
reproduce, because physical systems are impossible to isolate,
because measuring instruments are often unreliable, because
the theory of error even leads us to expect ‘‘freak”’ results, it
ROLE OF EMPIRICAL PROBLEMS 19
often takes a considerable time before a phenomenon is
sufficiently authenticated to be taken seriously as a well-estab-
lished effect. Second, it is often the case that even when an effect
has been well authenticated, it is very unclear to which domain
of science it belongs and, therefore, which theories should seek,
or be expected, to solve it. Is the fact that the moon seems
larger near the horizon a problem for astronomical theories, for
optical theories, or for psychological ones? Is the formation of
crystals and crystalline growth a problem for chemistry or
biology or geology? Are “shooting stars’’ a problem for
astronomy or for upper-atmosphere physics? Is the twitching of
an electrified frog leg a problem for biology, chemistry, or
electrical theory? We now have answers to all these questions
and feel confident about assigning these problems to one
domain or another. The chief reason for our confidence is that
we have solved these problems. But for long periods in the
history of science, these problems were unsolved and it was very
unclear within what domain they should fall. As a result of that
uncertainty, it did not count seriously against any theory in a
given domainif it failed to solve these unsolved problems; for no
one could show convincingly that theories in any particular
domain should be expected to solve such probiems.
The ambiguous status of unsolved problems is persuasively
illustrated by the history of the problem of Brownian motion.
First discussed at length by Robert Brown in 1828, it took the
greater part of a century before scientists could decide whether
it was a genuine problem, how important it was, and what sorts
of theories should be expected to solve it. During the 1830s and
1840s for instance, it was alternately viewed as a biological
problem (the Brownian particles perhaps being small ‘‘animal-
cules’), as a chemical problem, as a problem in the optics of
polarization (e.g., by Brewster), as a problem of electrical
conductivity (e.g., by Brongniart), as a problem in heat theory
(e.g., by Dujardin), as a completely uninteresting mechanical
effect which was too complicated and too insignificant to be
worthy of efforts at solution, and—by some—as a nonproblem
altogether.* So long as the problem remained unsolved, any
theorist could conveniently choose to ignore it simply by saying
that it was not a problem which theories in Ais field had to
20 ROLE OF EMPIRICAL PROBLEMS

address. Yet this selfsame phenonemon, which could find

neither home nor solution in the first half of the nineteenth
century, gradually emerged as one of the core anomalies for
classical thermodynamics and became, at the hands of Einstein
and Perrin (who solved the problem), one of the triumphal
successes of the kinetic-molecular theory of heat.
Consider, as another example, the famous case of Abraham
Trembley’s hydra-like polyp, first carefully observed in 1740. It
was a phenomenon which seemed to run counter to the
dominant biological ideas of the age; it could reproduce itself
without sexual coupling and, when cut up, each part would
quickly grow into a whole organism. These properties had been
commonly observed in plants, yet were specifically denied to
animals, suggesting that the polyp was a plant. On the other
hand, the polyp had powers of locomotion, a stomach and
patterns of food consumption usually associated with animals,
especially insects. Here, then, was a living organism—half-
plant, half-animal—whose very existence denied the long-
cherished biological principle of three separable kingdoms(ani-
mal, vegetable, and mineral). The reaction to Trembley’s dis-
covery was immediate—throughout the 1740s and 1750s, biolo-
gists and naturalists all over Europe speculated about it
and studied its behavior. This case would seem to be a
compelling example of the generation of a serious empirical
problem in the absence of any theory which couldsolve it.
But as Vartanian has convincingly shown,’ the above ac-
count, suggesting as it does the emergence of an acute
anomaly in the absence of any theoretical competition, is
deplorably incomplete. What it ignores is the fact that—
alongside of the dominant vitalistic biology—there existed a
minority of biologists committed to a more materialistic and
more mechanistic approach to biological processes. The regen-
erative powers of the polyp (along with its obvious animal
characteristics) suggested that perhaps the materialists were
correct. After all, if every part of the polyp, no matter how
small, could regenerate a fully developed animal, then the
materialists seemed to be right in denying there was an
indivisible, super-materialistic soul which belonged to the whole
organism only as an organized being.
 ROLE OF EMPIRICAL PROBLEMS 21

Practically from the first discovery of the polyp, proponents of

vitalistic biology recognized that the properties of the hydra
could give ‘“‘aid and comfort” to a rival research school.
Cramer, Lyonnet, and two anonymous writers (in the Mémoires
de l'Académie des Sciences and in the Journal de Trévaux) had,
through the early and mid-1740s, already remarked on the
susceptibility of the polyp to a materialistic interpretation (an
interpretation fully developed by La Mettrie in his L’homme
machine). In short, what transformed the polyp from an idle
curiosity into a threatening anomaly for vitalistic biology was
the presence of an alternative theory (or, as I shall later call it,
an alternative research tradition) which could count the polyp
as a solved problem.’°
Cases in which there is doubt about the appropriate domain
for some unsolved problem have frequently been of decisive
historical importance. Thevicissitudes of comets provide a neat
example. During antiquity and the Middle Ages, comets were
classified as sublunary phenomena and thus fell within the
domain of meteorology. Astronomers, whose concern was
exclusively with problemsin the celestial regions, felt no need to
offer theories about comets, nor even to plot their courses. By
the sixteenth century, however, it had become customary to
classify comets as celestial phenomena. This domain transi-
tion was crucial for the Copernican theory, since the motion of
comets came to constitute one of the decisive anomalies for
geocentric astronomy and one of the solved problems for the
heliocentric theory.
One ought not conclude from their ambiguity that unsolved
problems are unimportant for science, for transforming un-
solved problems into solved ones is one of the means by which
theories make empirical progress. But it must be stressed at the
same time that a theory’s failure to solve some unsolved
problem generally will not weigh heavily against that theory,
because we usually cannot know a priori that the problem in
question should be soluble by that sort of theory. The only
reliable guide to the problems relevant to a particular theory is
an examination of the problems which predecessor—and com-
peting—theories in that domain (including the theory itself)
have already solved. Hence, in appraising the relative merits of
22 ROLE OF EMPIRICAL PROBLEMS

theories, the class of unsolved problemsis altogether irrelevant.

What matters for purposes of theory evaluation are only those
problems which have been solved, not necessarily by the theory
in question, but by some known theory. (Here, as elsewhere,
the evaluation of a theory is closely linked to a knowledge of
its competitors.)

The Nature of Solved Problems

We have already indicated that “‘the solving of problems”
ought not be confused with “the explaining of facts,” and have
discussed at some length the disanalogies between facts and
empirical problems. What requires further elaboration is the
difference between the logic and pragmatics of problem solution,
and the logic and pragmaticsof scientific explanation.
Most of the major differences emerge clearly if we begin by
exploring the criteria for something to count as a solved
problem. In very rough form, we can say that an empirical
problem is solved when, within a particular context of inquiry,
scientists properly no longer regard it as an unanswered question,
i.e., when they believe they understand why the situation pro-
poundedbythe problem is the wayit is. Now clearly, it is theories
which are meant to provide such understanding and anyrefer-
ence to a solved problem presupposes the existence of a theory
which purportedly solves the problem in question. So when we
ask whether a problem hasbeen solved, weare really asking whe-
ther it stands in a certain relationship to some theoryor other.
What does that relationship amount to? If we ask a logician
of science the analogous question (to wit: what is the relation
between an explanans and its explanandum?), he will generally
tell us: the explaining theory must (along with certain initial
conditions) entail an exact statement of fact to be explained; the
theory must be either ¢rue or highly probable; whatever counts
as an adequate explanation of any fact must be regarded as
always having been such (so long as the epistemic appraisal of
the explanans does not change). By way of contrast, I shall
claim that: a theory may solve a problem so long as it entails
even an approximate statement of the problem; in determining
if a theory solves a problem, it is irrelevant whether the theory is
 ROLE OF EMPIRICAL PROBLEMS
23
true orfalse, well or poorly confirmed; what
counts as a solution
to a problem at one time will not necessaril
y be regarded as
such at all times. Each of these difference
s demands further
clarification.

The approximative character of problem soluti

on. Although
rare, it sometimes happens that a theory exactl
y predicts an
experimental outcome. Whenthat desirable
result is achieved,
there is cause for general rejoicing. It is far
more common for
the predictions deduced from a theory
to come close to
reproducing the data which constitute a specif
ic problem, but
with no exact coincidence of results. Newton
was not able to
explain exactly the motion of the planets; Einste
in's theory did
not exactly entail Eddington’s telescopic
observations; modern
chemical bonding theory does not predict
with exactitude the
orbital distance of electrons in a molecule;
thermodynamics
does not precisely fit heat transfer data
for any known steam
engine. There are many reasons one could
suggest (e.g., the use
of ‘ideal cases,’ the non-isolation of
real systems, imperfections
in our measuring instruments) to expla
in the frequent small
discrepancies between ‘‘theoretical result
s” and “laboratory
results,” but they are not of primary conce
rn here. What is
relevant is that facts are very rarely if ever
explained (if we take
our sense of explanation from the classi
cal deductive model),
because there is usually a discordance betwe
en what a theory
entails and our laboratory data. By contra
st, empirical prob-
lems are frequently solved because for probl
em solving purposes
we do not require an exact, but only
an approximate,
resemblance between theoretical results and
experimental ones.
Newton did solve, and was widely regarded
as having solved, the
problem of the curvature of the earth—even
though his results
were not identical with observational findin
gs. The thermo-
dynamic theories of Carnot and Clausius
were correctly per-
ceived in the nineteenth century as adequateso
lutions to various
problems of heat transfer, in spite of the
fact that they applied
exactly only to ideal (i.e., nonexistent) heat
engines.
As should be clear, the notion of soluti
on is highly relative
and comparative in a way that the notion
of explanation is not.
We can have two different theories which
solve the same
 ROLE OF EMPIRICAL PROBLEMS
24

problem, and yet say one is a better solution (i.e., a closer

approximation) than the other. Comparable locutions and
comparisons within the rhetoric of explanation are disallowed
by many philosophers of science; on the standard model of
explanation, something either is or definitely is not an explana-
tion—degrees of explanatory adequacy are not countenanced.
For instance, philosophers of science have been very troubled by
the relationship of Galileo’s and Newton’s theories of fall and
the data. Unable to say that both theories ‘‘explained” the
phenomena of fall (because the two are formally inconsistent),
they invented a variety of devices for excluding the title
“explanatory” from one or the other of the theories. Yet it is
surely more natural historically, and more sensible conceptually,
to say that both theories (Galileo’s and Newton’s) solved the
problem of free fall, one perhaps with more precision than the
other (although even that is dubious). It redounds to the credit
of both that, as Newton himself perceived, each provided an
adequate solution to the problem at hand. We are, however,
precluded from taking this natural way of describing the
situation if we accept many of the current doctrines about the
nature of explanation.

The irrelevance of truth and the falsity to solving a problem.

The suggestion that questions of truth and probability are irrele-
vant when determining whether a theory solves a particular prob-
lem probably seemsheretical, if only because one is so conditioned
to considering the search for true understanding as one of the
core aims of science. But whatever role questions of truth have
in the scientific enterprise (and this is a large question to which
we shall return''), one need not, and scientists generally do not,
consider matters of truth and falsity when determining whether
a theory does or does not solve a particular empirical! problem.
We can all agree, for instance, that Ptolemy’s theory of
epicycles solved the problem of retrograde motion of the
planets, regardless of whether we accept the truth of epicyclic
astronomy. Equally, everyone agrees that Thomas Young’s wave
theory of light—whether true or false—solved the problem of
the dispersion of light. Lavoisier’s theory of oxidation, whatever
its truth status, solved the problem of why iron is heavier after
ROLE OF EMPIRICAL PROBLEMS 25

being heated than before. Generally, any theory, T, can be

regarded as having solved an empirical problem, if T functions
(significantly) in any schema of inference whose conclusion is a
statement of the problem.

The frequent nonpermanenceof solutions. One of the richest

and healthiest dimensions of science is the growth through time
of the standards it demands for something to count as a solution
to a problem. What one generation of scientists will accept as a
perfectly adequate solution will often be viewed by the next
generation as a hopelessly inadequate one. The history of
science is replete with cases where solutions whose precision and
specificity were perfectly adequate for one epoch are totally
inadequate for another. Consider a few examples:
In his Physics, Aristotle cites the problemof fall as a central
phenomenon for any theory of terrestrial mechanics. Aristotle
himself sought to understand both why bodies fall downwards
and why they accelerate in fall. Aristotelian physics provided
answers to these questions which were taken seriously for over
two millenia. For Galileo, Descartes, Huygens, and Newton,
however, Aristotle’s views were not really solutions to the
problem of fall at all, for they failed utterly to explain the
“uniform difform”’ (i.e., uniformly accelerated) character of the
fall of a body. One might want to say that the later thinkers
were simply working with a very different problem than
Aristotle was; I would be more inclined to see this as a case
where, through the course of time, the criteria for what counts
as solving a problem have evolved so much that what was once
regarded as an adequate solution ceases to be, regarded as such.
A clearer case is provided by the history of the kinetic theory
of gases. By the 1740s, both Newton (using a central forces
model) and Daniell Bernoulli (using a collision model) had
shown that one could solve the problem of pressure-volume
relations of gases in terms of assumptions about the mechanical
interaction of their constituent particles. By the late nineteenth
century, however, enough data about gaseous behavior had
been accumulated to show that the simple kinetic theory
provided only very inexact approximations to gaseous behavior,
especially at low temperatures or high pressures. In short, given
26 ROLE OF EMPIRICAL PROBLEMS

eighteenth-century standards of experimental accuracy and

canons for adequacy of problem solution, the kinetic theory
was a far cry from an adequate problem solver, especially so
far as certain ranges of data were concerned. Accordingly, van
der Waals andothersset out to modify traditional kinetic theory
so as to enable it to solve the problem of pressure-volume
relations to meet their contemporary standards of problem
solution. What resulted, of course, was van der Waal’s
equation.
In the history of many disciplines, humanistic as weil as
scientific, one can perceive a gradual tightening and strengthen-
ing of the threshold at which a theory will be conceded to be a
solution to the relevant problem. Unless we acknowledge that
the criteria for acceptable problem solutions do themselves
evolve through time, the history of thought will seem enigmatic
indeed.

The Special Role of Anomalous Problems

Manyhistorians and philosophers of science have attached
special significance to the place of anomalies in science.
Thinkers from Bacon through Mill, Popper, Griinbaum and
Lakatos have stressed the importance of refuting or falsifying
experiments in the appraisal of scientific theories. Indeed,
certain philosophies of science (especially those of Bacon and
and Popper) make the search for, and resolution of, anomalies
the raison d’étre of the scientific enterprise, and the absence of
anomalies the hailmark of scientific virtue. While sharing the
view that anomalous instances have been, and should be,
among the most important components of scientific rationality,
I find myself seriously at odds with the conventional wisdom
about what anomalies are and about the interpretation of their
undoubted significance.
On the traditional view, anomalies have two chief charac-
teristics:
(a) the occurrence of even one anomaly for a theory should
force the rational scientist to abandonit;
(b) the only empirical data which can count as anomalies are
those which are logically inconsistent with the theory for
which they are anomalies.
ROLE OF EMPIRICAL PROBLEMS 27

I find these characteristics factually misleading as to actual

scientific practice (both past and present), and a conceptual
hindrance in understanding the role of anomalies in theory
appraisal. I want to claim, by contrast, that:
(a‘) the occurence of an anomaly raises doubts about, but
need not compel the abandonmentof, the theory exhibit-
ing the anomaly;
(b') anomalies need not be inconsistent with the theories to
which they are anomalies.
Thefirst of these two contentions (a’) is the less controversial,
if only because numerous critics of the classical view have
already offered cogent arguments for it; as a result, here I shall
only briefly rehearse the reasons for it. The second thesis (b’),
however, is not a familiar one, and I shall elaborate on it at
some length.
Taking (a’) first, several philosophers (particularly Pierre
Duhem, Otto Neurath, and W. Quine)’? have argued that we
cannot rationally decide whether a particular theory which
generates an anomaly should be abandoned because of certain
ineliminable ambiguities about the testing situation. The prin-
cipal ambiguities are two:
1. In any empirical test, it is an entire network of theories
which is required for deriving any experimental prediction. If
the prediction turns out to be erroneous, we do not know
where to locate the error within the network. The decision
that one particular theory within the network is false is, these
critics argue, completely arbitrary.
2. To abandon a theory because it is incompatible with the
data assumesthat our knowledge of the data is infallible and
veridical. Once we realize that the data themselves are only
probable, the occurrence of an anomaly does not necessarily
require the abandonment of a theory (we might rationally
choose, for instance, to ‘‘abandon”’ the data).
Still other critics’? of (a) have stressed not the ambiguity, but
the pragmatics, of theory testing and theory choice. They point
out that almost every theory in history has had some anomalities
or refuting instances; indeed, no one has ever been able to point
to a single major theory which did not exhibit some anomalies.
Accordingly, if we were to take (a) seriously, then we should
find ourselves abandoning our entire theoretical repertoire in
28 ROLE OF EMPIRICAL PROBLEMS

wholesale fashion, and thereby totally unable to say anything

whatever about most domains of nature. For these reasons,
there seem strong grounds for replacing (a) by the weaker, but
more realistic, view (a').
However, almost all writers on the subject of anomalies,
whether defenders or critics of the classical view (a), seem to
subscribe to (b), and to hold that an anomaly is only generated
when there is a /ogical inconsistency between our “‘theoretical”’
predictions and our “‘experimental’’ observations. They have
argued, in other words, that the only time data can be
epistemically threatening for a theory is when such data
contradict the claims of the theory. This strikes me as a far too
restrictive notion of an anomalous probiem. It is true, of course,
that a genuine inconsistency between theory and observation
may, under certain circumstances, constitute a particularly
vivid example of anomaly. But such inconsistencies are far from
being the only form of anomalous problem.
If, as I think we must, we take (a’) seriously, it is reasonable
to characterize an anomaly as an empirical situation which,
while perhaps not offering definitive grounds for abandoning a
theory, does raise rational doubts about the empirical creden-
tials of the theory. Proponents of (a’), in criticizing (a), are not
claiming that we should ignore anomalies; rather, they are
simply stressing that anomalies constitute important, but not
necessarily decisive, objections to any theory which exhibits
them. If we regard anomalies in this light (i.e., as empirical
problems which raise reasonable doubts about the empirical
adequacy of a theory), then we should abandon (b) for (b’),
since by parity of reasoning, there are many empirical problems
which, although consistent with a theory, can cast doubt upon
its empirical foundations. Putting the point another way, there
are occasions when scientists have rationally treated certain
problems (which were consistent with a theory) in the same
way that they would treat anomalies which were clearly
inconsistent with the theory. Such situations arise when a theory
in somefield or domain fails to say anything about a kind of
problem which other theories in the same domain havealready
solved.
Whether we treat such cases as anomalous depends, in part,
of course, on our views about the aims of science. If one takes
ROLE OF EMPIRICAL PROBLEMS 29

the narrow view that the object of science is simply to avoid

making mistakes (i.e., false statements), then unsolved prob-
lems will not necessarily count seriously against a theory. But
if one takes the broader view that science aims to maximize its
problem-solving capacity (or, in more conventional language, its
“explanatory content’’y then the failure of a theory to solve
some well-recognized problem, which has been solved by a
competitor theory, is a very serious mark against it. Ironically,
most philosophers of science have paid lip service to the broader
view, yet they have refused to recognize what that view entails—
the existence of a class of nonrefuting anomalies. \*
A careful look at the history of science makesit clear that a
number of situations generate behavior similar to the kind of
response which we have been led to expect when an inconsis-
tency between theory and observation arises. One of the most
important species of anomaly arises when a theory, although not
inconsistent with observational results, is nonetheless incapable
of explaining or solving those results (which have been solved by
a competitor theory).‘5 Thus, in Galileo’s classic study of
pendular motion, he criticizes the kinematical theories of his
predecessors because they cannot explain the mathematics of
pendular motion. His point is not that these earlier theories
give an incorrect prediction for the geometry of the moving
weight; rather his quarrel is that they give no prediction at all.
Similarly, many critics of Newtonian celestial mechanics in the
early eighteenth century argued that Newton’s system of the
world offered no explanation for the fact that all the planets
move in the same direction around the sun, a phenomenon
which had been solved by numerous previous astronomical
theories, particularly Keplerian and Cartesian astronomy.
Again, it is not that Newton's theory makes a false prediction
about the direction of planetary revolution; rather, the flaw is
that Newton's theory fails to address itself to the problem
altogether. (It would be compatible with the Newtonian system,
for instance, if adjacent planets moved in opposite directions.)
We can define this sort of anomaly more precisely by using
some of the terminology set out above: Whenever an empirical
problem, p, has been solved by any theory, then p thereafter
constitutes an anomaly for every theory in the relevant domain
which does not also solve p. Hence, the fact that some theory is
30 ROLE OF EMPIRICAL PROBLEMS

logically consistent with p does not render p nonanomalousfor

that theory, if p has been solved by any other known theory in
the domain.
The proposal, then, is that we should broaden our concept of
an anomalous instance so as to include this important class of
phenomena. Equally, in the spirit of (a’), we must weaken the
epistemic threat of a// anomalous instances by recognizing that,
although anomalies constitute good grounds for arguing against
a theory, they rarely, if ever, constitute final and decisive
arguments against a theory. They are important to the delicate
process of theory appraisal, but they remain but one of the
vectors which determine the scientific acceptability of a theory.
In stressing that a problem can only count as anomalous
for one theory if it is solved by another, the analysis seems to
run against the common view that one sort of anomaly, the
refuting instance, poses a direct cognitive threat to a theory,
even if it is unsolved by any competitor. If a theory predicts a
certain experimental outcome (say O) and experiment reveals
that ~O is the case, then surely, ~O constitutes an anomaly for
the theory even if no other theory can solve ~O? As paradoxical
as it may seem, this is generally unsound. An account of the
reasons why manyrefuting instances are not anomalousrequires
further analytic machinery which will be developed in chapter
three. We must here satisfy ourselves with the observation that
unsolved refuting instances are often of little cognitive
significance.

Converting Anomalies to Solved Problems

One of the most cognitively significant activities in which any
scientist can engage is the successful transformation of a
presumed empirical anomaly for a theory into a confirming
instance for that theory. Unlike the solution of some new
problem, the conversion of anomalies into problem-solving
successes does double service: it not only exhibits the problem
solving capacities of a theory (which the solution of any problem
will do) but it simultaneously eliminates one of the major
cognitive liabilities confronting the theory. This process of
converting anomalies (real or apparent) into solved problemsis
ROLE OF EMPIRICAL PROBLEMS 31

as old as science itself; the history of ancient astronomy is

replete with examples of it. Indeed, the basic idea is encap-
sulated in the classic aphorism exceptio probat regulam—which
originally meant that a rule or principle is tested by its ability
to deal with its apparent exceptions. Although numerous
examples of this conversion phenomenon could be cited, the
best known is probably the evolution of Prout’s hypothesis
concerning atomic composition. It was Prout’s view that all the
elements were composed of hydrogen and, consequently, the
atomic weights of all elements should be integral multiples of
the weight of hydrogen. Shortly after the appearance of this
doctrine in 1815, numerous chemists pointed to seeming
exceptions or anomalies. Berzelius and others found that several
elements had atomic weights incompatible with Prout’s theory
(e.g., weights of 103.5 for lead, 35.45 for chlorine, and 68.7 for
barium). These results constituted very serious anomalies for
Proutian chemists. By the beginning of the twentieth century,
however, the discovery of isotopes and the refinement of
techniques of isotopic separation enabled physical chemists to
separate out the isotopes of the same element; each isotope was
found to have an atomic weight which was an integra! multiple
of hydrogen. The previously anomalous results could now be
explained on Prout’s hypothesis by showing them to be isotopic
mixtures. Thus, the very phenomena which had earlier consti-
tuted anomalies for Prout's hypothesis became positive in-
stances for it. Almost every major theory in the history of
science has been able to produce comparable successes at
digesting someofits initial anomalies.

The Weighting of Empirical Problems

Up to this point in the discussion, we have been assuming
that all empirical problems are on essentially the same footing.
In fact, of course, some solved problems count for more than
others, and some anomalous problems are more threatening
than others. If the problem-solving approach is ever to become
a useful tool for appraisal, it must be able to show how, and
why, certain problems are more significant than others.
32 ROLE OF EMPIRICAL PROBLEMS

The Weight of Solved Problems

There are certain empirical problems which, at a given time
in a given scientific domain, are (and should be) given high
priority; so high that if a theory in that domain solves them it
will ipso facto be regarded as a serious contender for the
tational allegiance of the scientific community. On the other
hand, certain problemsare of marginal importance. It would be
nice to have a solution for them, but no theory is going to be
abandoned simply because it fails to solve them. Similarly,
anomalies range in importance from being decisive arguments
against a theory (usually called ‘‘crucial experiments’’) to being
rather minor exceptions which can often be completely ignored.
If a philosophy of science, or a model of scientific progress, is
going to be satisfactory, it must provide some guidelines not
only for counting, but also for weighting, scientific problems on
a scale of relative importance and cruciality.
In this section, I am going to be making some proposals
concerning ways in which problemscan be rationally weighted.
Before embarking on that task, however, two caveats should be
noted.
First, the criteria I am proposing are not meant to exhaust the
modesof rational weighting. A calculus of problem weights is a
major undertaking, well beyond the scope of this essay; hence,
my list is only partial, suggestive rather than exhaustive.
Second, what follows concerns only the cognitively rational
weighting of scientific problems. There are often occasions when
a problem becomes of major importance to a community of
scientists on nonrational or irrational grounds. Thus, certain
problems may assume a high importance because the National
Science Foundation will pay scientists to work on them or, as in
the case of cancer research, because there are moral, social, and
financial pressures which can “promote” such problems to a
higher place than they perhaps cognitively deserve. It is not my
purpose to discuss the nonrational dimensions of problem
weighting (although I shall have something to say about that in
chapter seven); we must first clarify what sorts of factors can
affect the weighting of problems within the context of the
rational appraisal of scientific theories.
ROLE OF EMPIRICAL PROBLEMS 33

In a new scientific domain, i.e., in a domain in which no

adequate, systematic theories have yet been developed, almost
all empirical problems are on a par. There is usually no good
reason for singling out one, or a group of them, as being more
important or crucial than another. Once we have one or more
theories in the domain, however, we immediately have certain
criteria for increasing the importance of certain empirical
problems.’* Three sorts of cases here are quite important:

Problem inflation by solution. If a particular problem has

been solved by any viable theory in the domain, then that
problem acquires considerable significance; to the extent that
any competitor theory in the domain will almost certainly be
expected either to solve it or to provide good grounds for
failing to solve it. Thus, once Galileo found a solution to the
problem of how fast bodies fall, every other subsequent theory
of mechanics was under strong constraints to provide an equally
adequate solution to the same problem.
Elaborating on anearlier point, it is tempting to formulate an
even stronger version of this thesis by claiming that, in many
(but not all) cases, an empirical situation does not even count as
a problem at all until it has been solved by some theory in the
domain. In such cases, solving a problem does not increase the
previous weight of the problem; rather it is the solution which
allows us to recognize the problem as a genuine problem at all.
The reason for this is that it is often unclear whether a seeming
problem really is an empirical problem, i.e., whether there is
any natural phenomenon there to explain at all. Experiments in
extrasensory perception are a case in point. Most scientists
today would claim to be unsure that there is any evidence of
ESP which is in need of theoretical explanation. The so-called
“‘pseudo-sciences”’ (as well as newly emerging sciences) generally
flourish on just such cases, where it is unclear whetherthereis,
at the outset, any problem which needs to besolved.

Problem inflation by anomaly solution. If a problem has

proved anomalous for, or resisted solution by, certain theories
in the domain, then any theory which can transform that
 34 ROLE OF EMPIRICAL PROBLEMS

anomalous problem into a solved one will have strong argu-

ments in its favor. The success of the special theoryofrelativity
in solving the results of the Michelson-Morley experiments
(which became anomalous problems for earlier aether theories)
is a widely known example of just such a process. Other
examples include: Newton’s explanation of the shape of the
earth and the elongation of the spectrum, Darwin’s explanation
of domestic breeding experiments, and Einstein’s explanation
of the photoelectric effect.

Problem inflation by archetype construction. At a more subtle

level, there are other ways in which theories may endowcertain
empirical problemswith greater significance than others. As we
shall see later in detail, many theories single out, from the range
of problems in the domain, certain empirical situations as
archetypal. I call them ‘‘archetypal’”’ because the theory indi-
cates that they are the primary orbasic natural process to which
other processes in the domain must be reduced. For instance,
before the time of Descartes, problems of the impact and
collision of bodies were at the periphery of the concerns of
writers on motion and mechanics, scarcely even recognized as
probiems which a theory of motion should resolve. But the
mechanical philosophy of Descartes, precisely because it con-
ceived collisions as the primary mode of interaction between
bodies, promoted problems about impact to the forefront of
mechanics, where they have remained ever since. In this case,
as in other similar ones, the inflation of the value of collision
problems was more than just a capricious shift of research
emphasis. As a Cartesian, one was committed to the thesis that
virtually the whole of natural science could be reduced to the
laws of collision. But those laws, on which so much hung, were
totally unknown early in the seventeenth century. It was thus
entirely reasonable for Cartesians, and those interested in the
Cartesian approach, to regard problems of impact andcollision
as among the most urgent in physics. Similarly, a century
later Franklin’s explanation of the Leyden jar, a primitive
condensor, managed doubly to increase the significance of the
problem of the Leyden jar, both by successfully solving what
ROLE OF EMPIRICAL PROBLEMS 35

had already been recognized as a puzzling phenomenon, and by

solving it utilizing a theory which made the Leyden jar an
archetypical case of electrification, rather than the mere bizarre
curiosity which it had generally been considered.’’
What is notable about all three of the modes of problem
weighting indicated above is the dependence of problem
importance on the available theories. Without an appropriate
type of theory, none of these three modes of problem weighting
would be possible. There is, however, one type of problem
weighting which is not always so dependent upon our existing
theories:

Problem weighting by generality. There are sometimes occa-

sions when one problem can be shown to be more general, and
thus more important, than another. For instance, Kepler’s
problem of finding the law for the motion of Mars is presum-
ably a special case of, and thereby less general than, his later
problem of finding the law for the motion of all the planets.
Mendel’s problem of trait transmission in pea plants is clearly
less general than the problem of trait transmission in all vege-
tables. But intuitions aside, the task of defining problem
generality is a difficult one. A certain type of case is relatively
straightforward: if we can show for any two problems p' and p,
that any solution to p' must also constitute a solution for p (but
not vice versa) then p' is more general, and thus of greater
weight, than p. Although this represents an important class of
cases, there are many other cases which do not permit one to
evaluate their comparative generality. In such cases, we must
fall back on thefirst three methodsof differential weighting.
Just as such circumstances can render some problems more
important than others, there are also circumstances which tend
to diminish the importance of empirical problems, whether
solved or unsolved.

Problem deflation by dissolution. As we have seen before,

problems represent presumed states of affairs, assumptions
about what we believe to be occurring in the world (or, more
usually, in the laboratory). Because we sometimes change our
36 ROLE OF EMPIRICAL PROBLEMS

beliefs about what is happening (if, for instance, certain

experimental results cannot be reproduced), many problems
simply vanish from a given domain. What was formerly
regarded as an important problem maypossibly cease to be one
altogether, and becomes instead a “‘pseudo-problem.” Even
when the problem does not vanish entirely, its importance
greatly diminishes as one’s doubts about its authenticity or its
relevance to the domain increase.

Problem deflation by domain modification. Another way in

which the importance of a problem within a domain is
signficantly diminished is by the expropriation of that problem
by another domain. Until the early seventeenth century, for
instance, writers on physical optics felt it important to explain
what was known about the physiology of the eye and the
psychology of vision. No “optical’’ theory was adequate unless
it addressed these problems. With the increasing specialization
of knowledge, however, problems in the physiology of vision and
in the psychology of perception were excised out of physical
optics, and thus had their prior importance within optics
radically devalued.

Problem deflation by archetype modification. As we saw

above, certain problems can be given prominence by the
emergence of a new theory which gives them special impor-
tance. The mirror image of this process occurs when a theory is
repudiated. Those problems which came into prominence
because they were archetypes of a now abandoned theory may
lose some of their importance as the theory with which they
were so closely allied wanes. After, for instance, Descartes and
other physicists in the seventeenth century had succeeded in
making collision processes into the archetypal mechanical
process, situations of work and energy expenditure—which had
been among Aristotle’s core examples—lost much of their
earlier prominence.

The Weight of Anomalous Problems

It has often been maintained, particularly by Karl Popper but
generally by all the logical empiricists, that any theory which
had anomalous empirical problems (in their language, a theory
ROLE OF EMPIRICAL PROBLEMS 37

which had been “refuted” or ‘‘disconfirmed’’) was no longer

worthy of serious scientific consideration. Any anomaly, any
“refuting instance,” was as important as any other. And one
empirical anomaly for a theory was as devastating as a hundred.
It has recently become clear, however, that such an approach
will not do; certainly not in practice and probably not even in
principle. As Kuhn and others have stressed, virtually every
theory ever devised, including those accepted by scientists
today, has anomalous instances. It simply is not true that, in
general, the discovery of an anomaly for a particular theory will
lead, in and of itself, to the abandonment of the theory which
exhibits the anomaly. At the same time, we must recognize that
there have been circumstances when theories were confronted
with sufficiently acute anomalous instances that they were
abandoned. If we are to capture whatever modicum of rational-
ity is implicit in such activity, we must be able, at least roughly,
to grade the anomalies which confront a theory in order to
indicate at least the differences between those anomalies which
are disastrous for a theory and those which are only a mild
embarrassment.
One possible way of dealing with this dilemma has been
offered by Thomas Kuhn, who proposes essentially that it is the
accumulation of a large number of anomalies which finally
induces scientists to abandon a theory.'* The difficulties with
Kuhn's solution to this problem are manifold: Kuhn offers no
reason why, for any number of anomalies, n, scientists should
be undisturbed by n-/ anomalies and suddenly ready to
abandon the theory altogether when it has n anomalies; Kuhn’s
account cannot be squaredwith the historical fact that scientists
have often abandoned a theory in the face of only a few
anomalies and have other times retained a theory in the face of
an ocean of empirical refutations.
I submit that if we are to find any rhyme or reason in the role
of anomalies within the history of science it can comeonly by a
recognition that it is not so much how many anomalies a theory
generates that count, but rather how cognitively important
those particular anomalies are.
How, then, can we begin to grade the importance of
empirical anomalies? The most natural approach here would
seem to involve grading anomalies in terms of the degree of
38 ROLE OF EMPIRICAL PROBLEMS

epistemic threat which they pose for a theory. A preliminary

first step in this direction comes from the recognition that the
importance of any particular anomaly for a theory depends
largely on the competitive state of play between that theory and
its competitors. If a theory happens to be the only known theory
in a particutar domain, then it may have dozens of “‘refuting”’
instances and probably none of them will be of decisive
importance. After all, when we ask about the importance of
anomaly, we are really asking the question: To what degree
should that anomaly dispose us to abandon the theory which
generated it? If there is no alternative theory in view to displace
it, all thought of abandoningit is probably academic, for, in the
absence of a successor, that would be a cognitive defeat of the
first order. So, assessing the importance of any seemingly
anomalous problem for a theory has to be done within the
context of the other competing theories in the domain. Given
that such theories exist, we can then ask whether a particular
unsolved problem, exhibited by T,, is also exhibited by 7,’s
competitors. If the answeris affirmative, that is, if all the extant
theories in the domain find themselves equally unable to solve
that particular phenomenon, then that problem cannot loom
very large in the assessment of T,—even if the problem is
logically inconsistent with T,. If, on the other hand, there is
some empirical problem which is unsolved by T,, but for which
some competing theory is able to provide a solution, then that
unsolved problem assumesa considerable significance for T,; in
short it becomes a genuine anomaly. Clearly, the importance of
an anomaly for a theory can vary enormously with time and
circumstances.
An example or two may make this clearer. Scientists since
antiquity have recognized that any astronomical and optical
theory ought to be able to explain the color of the sky. Until the
early twentieth century, however, no theory was able to provide
any adequate explanation why light, passing through empty
space and being refracted into the atmosphere, should produce
the familar blue color. It was only after Rayleigh worked out a
theory of atmospheric dispersion that the inability of an optical
theory to explain the blue of the sky counted as a major argu-
ment against such a theory. Similarly, the capacity of friction to
ROLE OF EMPIRICAL PROBLEMS 39

produce heat was a long known counterinstance to the view that

heat was a substance which inhered in bodies. But it was only
after the development of a kinetic theory of heat which could
successfully deal with the generation of heat by friction that
heat by friction became an important problem for heat-as-
substance theories. But the discussion thus far only tells us how
to identify an anomaly, not how to gradeits importance.
One important determinant of the importance of an anomaly
is the degree of discrepancy between the observed experimental
result and the theoretical prediction. Every theory is constantly
confronted by small order discrepancies between what it
predicts and what is observed. In the absence of a theory which
exhibits better fit with the data, few persons would attach much
importance to such quasi-anomalies. More serious, however, are
those discrepancies which are large, often representing several
orders of magnitude. Scientists are prepared to live with
theories which are approximate, but only to a certain degree.
Precisely where to draw the line depends very largely on the
conventional standards of accuracy, both theoretical and experi-
mental, within the domain. It is clear, for instance, that
cosmologists or geologists are often prepared to attach much
less significance to seemingly large discrepancies between
predicted and observed results than is, say, a physical chemist
or a spectroscopist. These differences in precision tolerance in
the various disciplines do not mean that these tolerance limits
are arbitrary. To the contrary, they usually reflect the subtle
instrumental and mathematical constraints on the field, as well
as the complexity of the process under investigation. What is
common to all the sciences is a conviction that certain experi-
mental results are so discordant as to constitute acutely
important anomalies, whereas other, only mildly disconcordant
results are relatively minor problems. Here again, the compara-
tive state of play between competing theories is decisive.
A second factor which influences the weight of an anomaly is
its age and its demonstrated resistance to solution by a
particular theory. No oneis very surprised if a newly discovered
phenomenon (perhaps anticipated predictively by one theory) is
anomalous for some other theory in the domain. Experience
teaches us that it sometimes takes a numberof intra-theoretic
40 ROLE OF EMPIRICAL PROBLEMS

adjustments before a problem can be convincingly solved. If,

however, after repeated efforts, a theory remains unable to
explain the anomaly, then it comes to loom ever larger as an
epistemic embarrassment. It is for this reason, incidentally, that
so-called crucial experiments—designed to choose between
competing theories—are rarely decisive immediately. It takes a
certain amount of time and effort at reconciliation before one
can reasonably cometo the conclusion that a theory is probably
going to be unable to solve any given anomalous problem.
I shall have more to say later about this general issue of the
weighting of empirical problems; but we can summarize the
discussion thus far by stressing two central claims:
1. the importanceof solving all empirical problems (whether
solved or anomalous) is not the same, some being of much
greater weight than others;
2. the assessment of the importance of a particular problem
or anomaly requires a knowledge of the various theories
within the domain and a knowledge of how successful or
unsuccessful those theories have been at offering solutions.

Theory Complexes and Scientific Problems

Up to this point in the discussion, I have been writing as if it
were single theories which solve, or fail to solve, empirical
problems. I have argued that individual theories can take credit
for the problems they solve and that they must bear the blame
for those anomalies they generate. It might be said, however,
that in taking such an approach I have ignored one of the most
striking and most significant aspects of the testing situation;
namely, the ambiguity of the epistemic threat posed by
anomalies. in order to determine whether my analysis founders
on this point, we must examine the arguments for the ambiguity
with some care.

The Alleged Ambiguity of Theory Testing

In the early years of this century, the French physicist-
philosopher Pierre Duhem argued that the testing of theories is
a great deal more complicated than the uncritical observer
might imagine.'? He pointed out that individual theories do not
ROLE OF EMPIRICAL PROBLEMS 4]

usually entail anything that can be directly observed in the

laboratory; rather, it is, he maintained, only a complex
conjunction of a variety of theories which can ever lead (given
certain statements of initial conditions) to any predictions about
the world. For instance, in order to test a theoretical statement
as simple as Boyle’s law, we must invoke (among other things)
theories about the behavior of our measuring instruments.
Boyle’s law byitself predicts nothing whatever about how those
instruments will behave. If, then, it is always (or even usually)
the case that theory complexes rather than individual theories
are subjected to empirical test, certain crucial ambiguities seem
to arise. Suppose, for instance, that a theory complex produces
an erroneous result (i.e., it leads to a prediction whichis refuted
by the evidence). What conclusion can we draw from that?
Duhem (and most of his recent commentators) wants to argue
that we can never deduce with certainty which theoretical
element(s) in the complex has been refuted or falsified by the
recalcitrant observation. All we learn from experience, he says,
is that we have gone astray somewhere, but thelogic of scientific
inference is too imprecise to allow us with certainty to pin the
blame on any particular component or components in the
theoretical complex. It follows that we can never legitimately
claim that any theory has ever been refuted.”°
A similar, but hitherto unnoted, ambiguity apparently affects
the confirmation as well as the refutation of individual scientific
theories of hypotheses. If it is true that theory complexes, and
only theory complexes, can confront experience, then the
successful prediction of an experimental outcome leaves us in as
much doubt about how to distribute credit, as an unsuccessful
prediction leaves us unclear where to locate blame. In the case
of a successful confirmation, should we assume that each
member of the theory complex is confirmed by the outcome?
And should we assume that each member gets the sameincre-
mental increase in its degree of confirmation as every other
member? These are difficult, and I think still unanswered,
questions.
But whatare we to make of these ambiguities of testing as far
as the model being discussed here is concerned? Is that model
open, in the ways that the received view is, to such an analysis
 42 ROLE OF EMPIRICAL PROBLEMS

and do these ambiguities make it meaningless to talk about the

appraisal of individual theories and hypotheses?

Problem Solving and Ambiguous Tests

I shall show below that the ambiguities of testing, while
genuine enough and worrying when directed against the
standard mode of discussing theory appraisal, are relatively
harmless when seen within the context of a problem-solving
model of theory appraisal. | shall show, further, that—within
the latter model—there is a natural way of handling the
Duhemian ambiguities which will still allow us to talk of the
rational appraisal of individual theories without having to
retreat to talking exclusively of theory complexes.
Let us deal with the ambiguities of refutation or falsification
first. The argument there, we recall, concluded that we cannot
legitimately deduce the falsity of any component of a theory
complex from the falsity of the theory complex as a whole. For
the sake of discussion, let us grant that this argument is
conclusive. Even if cogent, it implies nothing whatever about
the appropriateness of appraising the problem-solving effective-
ness of individual theories. We might, for instance, entirely
consistently with Duhemian worries, adopt the following prin-
ciple (A,),
Whenever any theory complex, C, encounters an
anomalous problem, a, then a counts as an anomaly
for each nonanalytic element, T,, T,,..., Ty, of C.”!
Why is the principle (A,) immune from criticism of a
Duhemian type? Simply because the whole thrust of the
Duhemian analysis has to do with assignments of truth or falsity
(or such weaker surrogates for them as probability or degree of
confirmation) to individual theories. The cogency of Duhem’s
position (as well as its recent elaborations) depends upon the
peculiar features of the assignment of truth-values within a
modus tollens argument. In that argument schema, we are
asked to imagine a situation where a theory complex, C, entails
some observation, O, which is false:
[C(consisting of T,, T,,..., T,) + initial conditions] > O
Not-O is observed

i a
ROLE OF EMPIRICAL PROBLEMS 43

The Duhemian points out that logic does not permit the
assertion of the falsity of any element, 7;, of the complex just
because the complex itself has been falsified.
Within the problem-solving model, however, we make no
assignments of truth or falsity; there is nothing in the structure
of deductive logic which precludesthe localization of properties
such as problem-solving effectiveness. When we say that a is an
anomaly for a theory 7,, we are not saying that a falsifies T, (to
claim that would open oneself to Duhemian objections); rather,
we are saying that a is the sort of problem which a theory such
as T, ought to be able to solve (albeit in conjunction with other
theories), but which it has failed as yet to solve. That, of course,
does not prove that T, is false; but it does clearly raise doubts
about the problem-solving effectiveness of 7, (and, for that
matter, about every other T; in the complex that failed to solve
the empirical problem a).
A similar sort of analysis applies to the apparent ambiguities
of confirmation. When westress those ambiguities, it is because
we are not clear how mucha successful confirmation of a theory
complex ought to increase our confidence in the truth (or the
likelihood) of its component elements. But if we shift from talk
about truth or probability to talk about problem solving, this
ambiguity dissolves as well, for there is a mirror image here of
the principle (A,) defined above for anomalies; namely (A,),
Whenever any theory complex, C, adequately solves
an empirical problem, b, then b counts as a solved
problem for each nonanalytic element, 7,, T,,...,
T, of C.
As principles (A,) and (A,) make clear, I am proposing we
turn the usual response to these Duhemian ambiguities on its
head. Where previous writers on this issue have tended to
imagine that the solution to the Duhemian ambiguity consists in
trying to find some way, contra Duhem’s analysis, for localizing
blame or credit, I want to try the opposite approach by
suggesting that a way out of the Duhemian conundrum may
emerge if, far from /ocalizing blameor credit in one place, we
simply spread it evenly among the members of the complex
(using a rational variant of the guilt-by-association doctrine).
 44 ROLE OF EMPIRICAL PROBLEMS

A full argument for principles (A,) and (A,) requires

lengthier treatment than I can give here. What I shall claim,
however, is that there is nothing in the usual arguments for the
ambiguity of testing which would undercut(A,) or (A,). To that
extent at least, we are entitled to claim that it seems to be
entirely appropriate to talk about the appraisal of individual
theories—with the proviso that such appraisals concern prob-
lem-solving effectiveness and not truth or falsity.
There is yet another important dimension of the Duhemian
problem which must be mentioned here, although a thorough
treatment of it will have to wait until we have developed further
machinery for theory appraisal in the next chapter. The dimen-
sion in question has to do with the nature of a rational response
to a so-called falsifying experiment. On my analysis, whenever a
complex of theories generates an anomaly, that anomaly counts
against each element within the complex. The fact that each of
those theories has this particular anomaly does not, of course,
require that they should each be abandoned; for, as we have
seen, the existence of an anomalous problem for a theory is not
ipso facto sufficient grounds for abandoning the theory. But
that is not an end on it. Precisely because the anomaly exists,
and because science seeks to minimize anomalies, thereis still
cognitive pressure on the scientific community to attempt to
resolve the anomaly. Resolving that anomaly will require,
presumably, the abandonment (though not by virture of its
“falsification”) of at least one of the theories that composed
the complex that was unable to deal with the anomaly. From
my point of view (and I suspect that from Duhem’s too), the
real challenge of the Duhemian analysis consists, not in showing
how wecan “‘localize’’ falsehood or truth, but rather in Showing
what rational strategies there are for selecting a better com-
plex.”? It is this point to which I shall return in chapter three,
where machinery for making the relevant assessments will be
described. "
 Chapter Two
Conceptual Problems
If a historian accepts the [customary]
analysis of confirmation, ... he may conclude
that the course of scientific development
is massively influenced by. . . nonevidential
considerations. WESLEY SALMON (1970), p. 80

Our discussion in chapter one focussed exclusively on empirical

problems and on the connections between such problems and
the theories which purport to solve them. It would be an
enormous mistake, however, to imagine that scientific progress
and rationality consist entirely of solving empirical problems.
There is a second type of problem-solving activity which has
been at least as important in the development of science as
empitical problem solving. This latter type of problem, which I
call a conceptual problem, has been largely ignored by
historians and philosophers of science (though rarely by scien-
tists), presumably because it does not comport well with those
empiricist epistemologies of science which have been the
reigning fashion for more than a century. The purposeof this
chapter is to state the case for a richer theory of problem
solving than empiricists have allowed, to explore the nature of
these nonempirical problems and to show whatrole they have in
theory appraisal.

45
46 CONCEPTUAL PROBLEMS

Even the briefest glance at the history of science makes it

clear that the key debates between scientists have centered as
much on nonempirical issues as on empirical ones. When, for
instance, the epicyclic astronomy of Ptolemy wascriticized (as
it often was in antiquity, the Middle Ages and the Renaissance),
the core criticisms did not deal with its adequacy to solve the
chief empirical problems of observational astronomy. It was
readily granted by most of Ptolemy’s critics that his system was
perfectly adequate for ‘‘saving the phenomena.” Rather, the
bulk of the criticism was directed against the conceptual
credentials of the mechanisms Ptolemy utilized (including
equants and eccentrics, as well as epicycles) for solving the
empirical problems of astronomy. Similarly, the later critics of
Copernican astronomydid not generally claim it was empirically
inadequate at predicting the motions ofcelestial bodies; indeed,
it could solve some empirical problems (such as the motion of
comets) far better than the available alternatives. What chiefly
troubled Copernicus’ critics were doubts about how heliocentric
astronomy could be integrated within a broader framework of
assumptions about the natural world—a framework which had
been systematically and progressively articulated since antiquity.
When, a century after Copernicus, Newton announced his
“system of the world,” it encountered almost universal applause
for its capacity to solve many crucial empirical problems. What
troubled many of Newton’s contemporaries (including Locke,
Berkeley, Huygens, and Leibniz) were several conceptual am-
biguities and confusions about its foundational assumptions.
What was absolute space and why wasit needed to do physics?
How could bodies conceivably act on one another at-a-distance?
What was the source of the new energy which, on Newton’s
theory, had to be continuously super-added to the world order?
How, Leibniz would ask, could Newton's theory be reconciled
with an intelligent deity who designed the world? In none of
these cases was a critic pointing to an unsolved or anomalous
empirical problem. They were, rather, raising acute difficulties
of a nonempirical kind. Noris it merely “early” science which
exhibits this phenomenon.
If we look at the reception of Darwin’s evolutionary biology,
Freud’s psychoanalytic theories, Skinner’s behaviorism, or
CONCEPTUAL PROBLEMS 47

modern quantum mechanics, the same pattern repeats itself.

Alongside of the rehearsal of empirical anomalies and solved
empirical problems, both critics and proponents of a theory
often invoke criteria of theoretical appraisal which have nothing
whatever to do with a theory’s capacity to solve the empirical
problemsof the relevant scientific domain.
Of course, this pattern has not gone unnoticed byhistorians,
philosophers and sociologists of science; it is too obvious and
too persistent to have been ignored altogether. But the usual
response, when confronted with cases in which theories are
being appraised along nonempirical vectors, has been to deplore
the intrusion of these ‘unscientific’? considerations and to
attribute them largely to prejudice, superstition, or a ‘“‘pre-
scientific temperament.’’ Some scholars (such as Kuhn) have
gone so far as to make the absence of such nonempirical
factors a token of the “maturity” of any specific science.'
Rather than seeking to learn something about the complex
nature of scientific rationality from such cases, philosophers
(with regret) and sociologists (with delight) have generally taken
them as tokens of the irrationality of science as actually
practiced.? As a result few scholars who study the nature of
science have found any room in their models for the role of such
conceptual problems in the rational appraisal of scientific
theories.’ Empiricist philosophies of science (including those of
Popper, Carnap and Reichenbach) and even less strident
empiricist methodologies (including those of Lakatos, Colling-
wood and Feyerabend)—all of which imagine that theory choice
in science should be governed exclusively by empirical consider-
ations—simply fail to come to terms with the role of conceptual
problems in science, and accordingly find themselves too
impoverished to explain or reconstruct much of the actual
course of science. Such empiricist theories of science exhibit
particularly awkward limitations in explaining those historical
situations in which the empirical problem-solving abilities of
competing theories have been virtually equivalent. Cases of this
kind are far more common in science than people generally
realize. The debates between Copernican and Ptolemianastron-
omers (1540-1600), between Newtonians and Cartesians (1720-
1750), between wave and particle optics (1810—1850), between
48 CONCEPTUAL PROBLEMS

atomists and anti-atomists (1815 to about 1880) are all

examples of important scientific controversies where the
empirical support for rival theories was essentially the same.
Positivistically inspired accounts of these historical encounters
have shed very little light on these important cases: this is
scarcely surprising since the positivist holds empirical support to
be the only legitimate arbiter of theoretical belief. These
controversies must, by the strict empiricist, be viewed as mere
querelles de mots, hollow and irrational debates about issues
which experience cannotsettle.
A broader view concerning the nature of problem solving—
one which recognizes the existence of conceptual problems—
puts us in a position to understand and to describe the kind of
intellectual interaction that can take place between defenders of
theories which are equally supported by the data. Because the
assessment of theories is a multi-factorial affair, parity with
respect to one factor in no way precludes a rational choice based
on disparities at other levels.

The Nature of Conceptual Problems

Thus far, we have defined conceptual problems by exclusion,
suggesting that they are nonempirical. Before we can under-
stand their role in theory appraisal, we must clarify precisely
what they are and how they arise. To begin with, we must stress
that a conceptual problem is a problem exhibited by some
theory or other. Conceptual problems are characteristics of
theories and have no existence independent of the theories
which exhibit them, not even that limited autonomy which
empirical problems sometimes possess. If empirical problems
are first order questions about the substantive entities in some
domain, conceptual problems are higher order questions about
the well-foundedness of the conceptual structures (e.g., theories)
which have been devised to answer the first order questions. (In
point of fact, there is a continuous shading of problems interme-
diate between straightforward empirical and conceptual prob-
lems; for heuristic reasons, however, [ shall concentrate on the
distant ends of the spectrum.)
Conceptual problems arise for a theory, T, in one of two ways:
CONCEPTUAL PROBLEMS 49

1. When T exhibits certain internal inconsistencies, or when

its basic categories of analysis are vague and unclear; these
are internal conceptual problems.
2. When is in conflict with another theory or doctrine, 7’,
which proponents of YT believe to be rationally well
founded; these are external conceptual problems.
Each of these forms of conceptual problems needs to be
analyzed in some detail.

Internal Conceptual Problems

The most vivid, though by no means the most frequent, type
of internal conceptual problem arises with the discovery that a
theory is logically inconsistent, and thus self-contradictory.
Probably most common in the history of mathematics, incon-
sistent theories have often been detected in almost all the other
branches of science.‘ Little need be said about the acuteness of
such problems. Unless the proponents of such theories are
prepared to abandon the rules of logical inference (which
provided the groundwork for recognizing the inconsistency), or
can somehow ‘“‘localize” the inconsistency, the only conceivable
response to a conceptual problem of this kind is to refuse to
accept the offending theory until the inconsistency is removed.*
More common, as well as more difficult to handle, are a
second class of internal conceptual problems; namely, those
arising from conceptual ambiguity or circularity within the
theory. Unlike inconsistency, the ambiguity of concepts is a
matter of degree rather than kind. Some degree of ambiguity is
probably ineliminable in any except the most vigorously axioma-
tized theories. It may even be true that some small measure of
ambiguity is a positive bonus, since less rigorously defined
theories can often be more readily applied to new domains of
investigation than more rigid ones. But granting that, it is
nonetheless true that systematic and chronic ambiguity or
circularity within a theory often has been, and should be,
viewed as highly disadvantageous.
Examples of such conceptual problems abound in the history
of science. For instance, Faraday’s early model of electrical
interaction was designedto eliminate the concept of action-at-a-
distance (itself a conceptual problem in earlier Newtonian
50 CONCEPTUAL PROBLEMS
physics). Unfortunately, as Robert Hare showed,° Faraday’s
own model required short range actions-at-a-distance. Faraday
had merely replaced one otiose concept by its virtual equivalent.
Even worse, Faraday’s model—as Hare was quick to point
out—postulated ‘‘contiguant” particles, which were notreally
contiguous at all. These kinds of criticisms led Faraday to
re-think his views on matter and force and were eventually
responsible for the emergence of Faraday’s field theory, which
avoided these conceptual problems. Taking another example
from nineteenth-century physics, it was often alleged by the
critics of the kinetic-molecular theory (e.g., Stallo and Mach)
that the kinetic theory was nonexplanatory because circular.
For instance, it explained the elasticity of gases by postulating
elastic constituents (i.e., molecules). But, observed the critics,
because we understand no more about the causesofelasticity in
solids than we do in fluids, the kinetic explanation is entirely
circular.’
The increase of the conceptual clarity of a theory through
careful clarifications and specifications of meaning is, as
William Whewell observed more than a century ago, one of the
most important ways in which science progresses. He called this
process “‘the explication of conceptions’”” and showed how a
numberof theories, in the course of their temporal careers, had
becomeincreasingly precise—largely as a result of the critics of
such theories emphasizing their conceptual unclarities.* Many
important scientific revolutions (e.g., the emergence of the
theory of special relativity, the development of behavioristic
psychology) have depended largely on the recognition, and
subsequent reduction, of the terminological ambiguity of
theories within a particular domain.
Although both these types of internal problems are doubt-
lessly important in the process of theory appraisal, neither have
played as decisive a historical role as the other kinds of
conceptual problemshave.

External Conceptual Problems

External conceptual problems are generated by a theory, T,
when T is in conflict with another theory or doctrine which the
CONCEPTUAL PROBLEMS 51

proponents of T believe to be rationally well founded. It is the

existence of this ‘‘tension’’ which constitutes a conceptual
problem. But whatprecisely do the ‘‘tension”’ and the “‘conflict”
amount to? The easiest form of ‘‘tension’’ to define, although by
no means the most frequent, is that of logical inconsistency or
incompatibility. When one theoryis logically inconsistent with
another accepted theory, then we have a vivid example of a
conceptual problem.
The development of astronomy in ancient Greece, to which
we have already referred, provides a useful case in point. The
unsolved empirical problem here (it was actually a host of
related problems) was summarized in tables of planetary
motion, tables which recorded the apparent positions of the
sun, moon, and planets at different times. This was the initial
empirical problem which had to be resolved. The succession of
planetary theories in antiquity, from the homocentric spheres of
Eudoxus and Aristotle to the complex epicycles, eccentrics, and
equants of Ptolemy, illustrates a series of attempts to solve the
problems of early astronomy. But as soon as the early
astronomical theories were developed each of them in turn
generated a plethora of other problems, some of them empiri-
cal, others conceptual. Thus, the homocentric spheres of
Eudoxus and Aristotle failed to explain accurately the retrogra-
dations of the planets and the seasonal inequalities exhibited by
the data. These phenomena wereclearly recognized as unsolved
problems. On the other hand, the later system of Ptolemy
managed to avoid most of the anomalous problems which
earlier Greek astronomy had encountered, but the price it paid
to do so wasthat of generating enormous conceptual problems.
Ever since the time of Plato, astronomers had worked on the
assumption that the heavenly motions were ‘‘perfect’’ (i.e., that
each planet moved in a perfect circle about the earth at
constant speed). This assumption put enormous constraints on
the kinds of hypotheses which were open to astronomers.
Ptolemy’s system, for all its empirical virtues, ran afoul of these
prohibitions by making assumptions about the behavior of
celestial bodies (e.g., the hypothesis that certain planets move
around empty points in space, that planets do not always move
$2 CONCEPTUAL PROBLEMS

at constant speed, and the like) which were in flagrant contra-

diction with the then universally accepted physical and cosmo-
logical theories about the nature and motion of the heavenly
bodies. In spite of ingenious efforts to reconcile these differ-
ences by Ptolemy and others, most of the crucial conceptual
of
problems remained, and were to plague the development
mathematical astronomy until the end of the seventeenth
century (and even beyond).
But there are other relations besides that of inconsistency
which also constitute conceptual problems for those theories
which exhibit them. One commonsituation arises when two
theories, although logically compatible, are jointly implausible,
i.e., when the acceptance of either one makesit less plausible
that the other is acceptable. For example, many late seven-
teenth-century theories of physiology were based on the (Carte-
sian) assumption that the various bodily processes were essen-
tially caused by the mechanical processes of collision, filtration,
and fluid flow. Once Newtonian physics was accepted, many
critics of mechanistic physiology pointed out that such mechan-
istic doctrines, although logically compatible with the physics of
Newton, were nonetheless rendered rather implausible by
Newtonian physics. The argument went something like this:
Newtonian physics, while certainly allowing for the existence of
collision phenomena, nonetheless shows that most physical
processes depend upon morethat the impacts between, and the
motions of, particles. To the extent that “‘mechanistic’’ (Carte-
sian inspired) theories of physiology postulate such processes as
the exclusive determinant of organic change, they rest on a huge
improbability. They are consistent with Newtonian physics (for
that physics does not deny that there can be some material
systems which are entirely mechanical); but it did seem highly
implausible, given Newtonian physics, that a system as complex
as a living organism could function with only a limited range of
the processes exhibited in the inorganic realm.
A second example may clarify the notion of conceptual
problem-generation by joint implausibility between theories.
Throughout the seventeenth and early eighteenth century, the
dominant theory of heat was a kinetic one; heat was conceived

een
CONCEPTUAL PROBLEMS $3

as the rapid agitation of the constituent parts of a body.

Throughout the eighteenth century, however, a number of
theories in a variety of fields began to suggest that many
natural processes depended upon the presence of one or more
highly elastic, highly rarefied fluids which could be absorbed
by, or released from, material bodies. Although electricity was
the best known example, such subtle fluids were postulated to
explain magnetism, neurological] functioning, perception, em-
bryology, and even gravity. As these theories became more
widely accepted, and as certain observable analogies between
heat, light and electricity began to be explored, kinetic theories
of heat came under sustained attack. While the acceptanceof,
for example, a fluid theory of electricity did not entail the denial
of the kinetic theory of heat, it was thought that kinetic theories
of heat became increasingly implausible as one domain after
another came to be dominated by highly successful ideas about
the substantial, as opposed to the kinetic, nature of physical
processes.
A third manner in which conceptual problems can be
generated occurs when a theory emerges which ought to
reinforce another theory, but fails to do so and is merely
compatible with it. To understand what is involved in such
cases, we must talk briefly about the interdisciplinary structure
of science, for compatibility between two systems or theories is
not, in common parlance, regarded as a sign of cognitive
weakness. The various scientific disciplines and domains are
never completely independent of one another. At any given
epoch, there are hierarchical systems of interconnection be-
tween the various sciences which condition the rational expecta-
tions which scientists have when they appraise theories. In our
own time, for instance, it is presumed that the chemist will look
to the physicist for ideas about atomic structure; that the
biologist should utilize chemical concepts when talking about
organic microstructures. The enunciation of a chemical theory
which was merely compatible with quantum mechanics, but
which utilized none of the concepts of quantum theory, would
be viewed askance by most modern scientists. Similarly, a
theory of heredity which was compatible with chemistry but
34 CONCEPTUAL PROBLEMS
failed to exploit any of its analytic machinery, would likewise be
suspect. Different epochs, of course, will have different expecta-
tions about which disciplines should borrow from, and rein-
force, others. (In the seventeenth century, for instance, it was
expected that any physical theory should be positively relevant
to, and not merely compatible with, Christian theology.)
As should be clear, mere compatibility between two theories
is not always a conceptual problem. No one thinks, for instance,
that a theory in micro-economics is flawed if it is merely
compatible with thermodynamics. But in many cases, compati-
bility, as opposed to positive relevance, between two theories is
quite rightly viewed as a major drawback to the acceptance of
the theories in question.
Our discussion thus far puts us in a position to outline a
taxonomy of the various cognitive relationships which can exist
between two (or more) theories:
1. Entailment—onetheory, 7, entails another theory, 7,.
2. Reinforcement—T provides a “‘rationale’’ for (a part of)
T,.°
3. Compatibility—T entails nothing about 7).
4. Implausibility—T entails that (a part of) T, is unlikely.
5. Inconsistency—T entails the negation of (a part of) T,.
In principle, any relation short of full entailment (1) could be
regarded as posing a conceptual problem for the theories
exhibiting it. It should be stressed, however, that although situa-
tions (2) to (5) can generate conceptual problems, they pose
very different degress of cognitive threat; those degrees are
represented, in increasing order, by the sequence (2) through (5).

The Sources of Conceptual Problems

In discussing external conceptual problems, I was deliberately
vague about what sorts of theories or beliefs can generate
conceptual problems for a scientific theory. I] have avoided this
issue thus far because | wanted to focus first on the kinds of
connections between theories which could generate conceptual
problems. The time has come, however, to spell out the other
side of the issue by asking what sorts of theories can qualify to
CONCEPTUAL PROBLEMS 55

be paired with a scientific theory in order to generate a concep-

tual problem; for unless we can answer that question coher-
ently, one could trivially and mechanically generate conceptual
problems for any theory simply by conjoining it arbitrarily with
any ‘wild’ belief we liked. For instance, we could create a
problem for modern quantum theory by pointing out its lack of
relevancy for Zen Buddhism! So far as I can tell, there are at
least three distinct classes of difficulties which can generate
external conceptual problems: (1) cases where two scientific
theories from different domains are in tension; (2) cases where
a scientific theory is in conflict with the methodological theories
of the relevant scientific community; and (3) cases where a
scientific theory is in conflict with any component of the
prevalent world view. Each merits serious discussion.

Intra-scientific difficulties. It is very often the case that a new

theory in some scientific domain will make assumptions about
the world which are incompatible with the assumptions of
another scientific theory, a theory which we have good inde-
pendent grounds for accepting. Thus, the astronomical system
of Copernicus—while not a theory of physics in itself—nonethe-
less made a number of assumptions about the motion of bodies
which were inconsistent with the then accepted Aristotelian
mechanics. One of the strongest sixteenth-century arguments
against the Copernican system consisted in pointing out that the
theory of Copernicus, although perhaps adequate so far as the
astronomical evidence went, was unacceptable because it ran
counter to the tenets of the best established physical theory.
Even worse, Copernicus really had no well-articulated alterna-
tive system of mechanics with which to rationalize the assump-
tions he was making about the motion of the earth. It was
Galileo’s signal contribution to deal with this conceptual
problem, by recognizing the incompatibility between Aristotelian
physics and Copernican astronomy and by remedying the
situation by designing a new physics which was independently
plausible and compatible with Copernican astronomy.
The recognition and resolution of such conceptual problems
has been one of the morefertile processes in the history of the
56 CONCEPTUAL PROBLEMS

natural and the social sciences.’ If two scientific theories are

inconsistent or mutually implausible, there is a strong presump-
tion that at least one of them should be abandoned. That much
is straightforward. What is more interesting is the fact that one
generally cannot simply jettison one or the other of an
inconsistent pair without wreaking havoc with the rest of
scientific knowledge. Because theories in certain domains(say,
astronomy) seem to require for their comprehension and
empirical assessment the existence of theories in other do-
mains (say, mechanics or optics),'' the decision to abandon
one of a pair of inconsistent theories and to retain the other
member of the pair usually involves a commitment to develop
an adequatealternative to the rejected theory.
As a result, such conceptual problems are generally much
easier to recognize than to resolve. Rarely, if ever, can we
resolve such problems by the simple device of rejecting one of
the offending pair. Moreover, as we have already seen, there is
nothing built into the process of scientific evaluation which can
inform us in advance which member of an inconsistent pair
ought to be rejected. That is a question which can be resolved
only after the fact, i.e., once we have tried giving up one, then
the other, and have observed with what success we can
construct an adequate pair-memberfor the retained theory.
Two final points about intra-scientific conceptual problems
should be madein passing. It should be stressed, first, that the
fact that a particular theory is incompatible with another
accepted theory creates a conceptual problem for both theories.
The inconsistency relation is symmetrical, and we must not lose
sight of the fact that intra-scientific conceptual problems
inevitably raise presumptive doubts about both membersof the
incompatible pair. Second, we should observe that the noting of
a logical inconsistency or a relation of non-reinforcement
between two theories need not force scientists to abandon one,
or the other, or both. Just as it can sometimes be rational to
retain a theory in the face of anomalous evidence, so, too, can it
be sometimes rational to retain a theory in the face of an
inconsistency between it and some other accepted theory. What
we must recognize is that the occurrence of such an inconsis-
tency indicates a weakness, a reason for considering the
abandonment of one or the other theory (or perhaps both).
CONCEPTUAL PROBLEMS 57

Among the most vivid examples of intra-scientific difficulties

were the controversies between biologists, geologists, and phy-
sicists in the late nineteenth century over the chronology of
the earth. On the geological and biological side was an
enormous amount of evidence to support the view that the earth
was very old indeed, -that it was partially fluid under the
surface, and that physical conditions on its surface had
remained largely unchanged for hundreds of millions of years.
Both uniformitarian geology and evolutionary biology rested
upon such assumptions. The physicist Lord Kelvin, however,
found himself unable to reconcile these core postulates with
thermodynamics. Specifically, he showed that the second law of
thermodynamics (entailing an increase in entropy) was incom-
patible with an evolutionary account of species and that both
the first and second laws were incompatible with the geologist’s
hypothesis that the energy reserves in the earth had remained
constant through much of the geological past. General perplex-
ity abounded. Thermodynamics had much going for it in
physics, but the dominant geological and biological theories also
could point to a huge reserve of solved problems. The dilemma
wasacute: ought one abandon thermodynamics, reject uniform-
itarian geology, or repudiate evolutionary theory? Or was there
some other option? As it turned out, though no one could have
foreseen this in advance, all three could be retained, since the
discovery of radioactivity made it possible to circumvent the
problems about energy conservation. What matters here, for
our purposes, is that the emergence of this incompatibility
created acute conceptual problems for a// the sciences con-
cerned. If the route to a resolution of the problems was murky,
it was generally perceived that these conceptual problems, until
resolved, raised strong doubts about the problem-solving effic-
acy of a wide rangeof scientific theories.

Normative difficulties. Science, as it is often said, is an

activity, an activity conducted by seemingly rational agents. As
such, it has certain aims and goals. The rational assessment of
science must therefore be, in large measure, a matter of
determining whether the theories of science achieve the cogni-
tive goals of scientific activity. What are these goals and how do
we achieve them? It is one of the central functions of any
 58 CONCEPTUAL PROBLEMS
philosophy or methodology of science to specify those
goals and
to indicate the most effective means for achieving them.
The
whole point of a methodological rule (such as Newton's
classic
dictum, “hypotheses non fingo’”) is to offer a norm for
scientific behavior; to tell us what we should,
or should not, do
in order to achieve the cognitive, epistemic, and practic
al goals
of the scientific enterprise.
Since antiquity, philosophers and philosopher-scientist
s have
sought to define sets of norms, or methodological
rules, which
are expected to govern the behavior of the
scientist. From
Aristotle to Ernst Mach, from Hippocrates to Claude
Bernard,
thinkers concerned about science have attempted
to legislate
concerning the acceptable modes of scientific inferen
ce. In the
early seventeenth century, the dominant image
was mathe-
matical and demonstrative, an image that becam
e canonical in
Descartes’ famous Discourse on Method. In the eighte
enth and
early nineteenth century, by contrast, most natura
l philosophers
were convinced that the methods of science should
be inductive
and experimental. Not surprisingly, every histori
cal epoch
exhibits one or more dominant, normative images
of science. It
would be a serious mistake to imagine, as manyhi
storians do,
that these norms are just the concern of the
professional
philosopher or logician. Every practicing scienti
st, past and
present, adheres to certain views about how
science should be
performed, about what counts as an adequate
explanation,
about the use of experimental controls, and
the like. These
norms, which a scientist brings to bear in his
assessment of
theories, have been perhaps the single major source
for most of
the controversies in the history of science, and
for the genera-
tion of many of the most acute conceptual proble
ms with which
scientists have had to cope.
It is still widely maintained that the methodology
to which a
scientist subscribes is really little more than perfun
ctory window
dressing, which is honored more in the
breach than in the
observance. Prominentscientists and historical
scholars of our
own era (most notably Einstein and Koyré’’) have
scoffed at the
idea that a scientist’s explicit views about method
ology can exert
much impact on hisscientific beliefs and activiti
es. Moreover,
there are significant cases (e.g., Newton and Galileo
) in which a
CONCEPTUAL PROBLEMS 59

scientist’s actual research violates almost every methodological

rule to which he pays lip service. How, under those circum-
stances, can I argue here that methodology is a potent source
for the evaluation of scientific theories and for the generation of
conceptual problems?
Fortunately, the work of several historians in the last twenty
years has provided overwhelming evidence that the methodolog-
ical beliefs of scientists often do profoundly effect their research
and their appraisals of the merits of scientific theories.'?> What
all these investigations make clear (contra-Einstein and Koyré)
is that the fate of most of the important scientific theories in the
past have been closely bound up with the methodological
appraisals of these theories; methodological well-foundedness
has been constitutive of, rather than tangential to, the most
important appraisals of theories.
It is for precisely that reason that perceived methodological
weaknesses have constituted serious, and often acute, concep-
tual problems for any theory exhibiting them. It is for the same
reason that the elimination of incompatibilities between a theory
and the relevant methodology constitutes one of the most
impressive ways in which a theory can improve its cognitive
standing.
The resolution of a ‘‘tension” between a methodology and a
scientific theory is often achieved by modifying the scientific
theory so as to reconcile it to the methodological norms. But
such problemsare not always resolved in this fashion. In many
cases, it is the methodology itself which is altered. Consider, as
but one example, the development of Newtonian theory in the
eighteenth century. By the 1720s, the dominant methodology
accepted alike by scientists and philosophers was an inductivist
one. Following the claims of Bacon, Locke, and Newton
himself, researchers were convinced that the only legitimate
theories were those which could be inductively inferred by
simple generalization from observable data. Unfortunately,
however, the direction of physical theory by the 1740s and
1750s scarcely seemed to square with this explicit inductivist
methodology. Within electricity, heat theory, pneumatics,
chemistry and physiology, Newtonian theories were emerging
which postulated the existence of imperceptible particles and
 60 CONCEPTUAL PROBLEMS

fluids—entities which could not conceivably be “inductively

inferred” from observed data. The incompatibility of these new
theories with the explicit methodology of the Newtonian re-
search tradition produced acute conceptual problems. Some
Newtonians(especially those in the so-called ‘‘Scottish School’)
soughtto resolve the conceptual problems by simply repudiating
those physical theories which violated the accepted methodolog-
ical norms.’* Other Newtonians (e.g., LeSage, Hartley, and
Lambert) insisted the norms themselves should be changed so
as to bring them into line with the best available physical
theories.'* This latter group took it on themselves to hammer
out a new methodology for science which would provide a
license for theorizing about unseen entities. (In its essentials,
the methodology they produced was the hypothetico-deductive
methodology, which even now remains the dominant one.) This
new methodology, by providing a rationale for ‘“‘micro-theoriz-
ing,’’ eliminated what had been a major conceptual stumbling
block to the acceptance of a wide range of Newtonian theories
in the mid and late eighteenth century. (Here, as above,
historians with purely empiricist models of science have com-
pletely missed the occurrence, let alone the significance, of
these developments in the evolution of the Newtonian research
tradition.)
Other cases of methodologically induced conceptual problems
abound. Much of the debate about uniformitarian geology,
much of the controversy about atomism, the bulk of the
opposition to psychoanalysis and behaviorism, and many of the
quarrels in quantum mechanics focus upon the methodological
strengths and weaknesses of the scientific theories in question.
Cases of this kind make it clear that the recognition of
normative conceptual problems is a much more potent force in
the historical evolution of science than some historians of
science have recognized.
But if historians have sometimes underestimated the impor-
tance of such conceptual problems, their culpability is insignifi-
cant when comparedto the utter failure of philosophers to find
any role for this sort of problem in their accounts of scientific
change. Even those philosophers who have been liberal enough
to find a role for metaphysics in scientific development have
CONCEPTUAL PROBLEMS ol

completely ignored the fact that the methodology to which a

scientist subscribes does, and should have, a major role to play
in determining that scientist’s assessment of the rational merits
of competing scientific theories. If a scientist has good grounds
for accepting some methodology and if some scientific theory
violates that methodology, then it is entirely rational for him
to have grave reservations about the theory. (It is one of the
crueler ironies of recent epistemology that epistemologists
themselves have never cometo terms with, nor founda rationale
for, the decisive role which epistemology and methodology have
enjoyed in the rational developmentof the sciences.)

Worldview difficulties. The third type of external conceptual

problem arises when a particular scientific theory is seen to be
incompatible with, or not mutually reinforcing for, some other
body of accepted, but prima facie nonscientific, beliefs. Within
any culture, there are widely accepted beliefs which go beyond
the scientific domain. Although the exact proportion of scien-
tific and nonscientific propositions within the total population
of reasonable beliefs changes with time, there has never been a
period in the history of thought when the theories of science
exhausted the domain of rational belief. What I am calling
worldview difficulties are like intra-scientific difficulties, except
that here the inconsistency, or lack of mutual reinforcement, is
not within the framework of science itself, but rather between
science and our “‘extra-scientific beliefs.’’ Such beliefs fall in
areas as diverse as metaphysics, logic, ethics and theology.
For example, one of the central conceptual problems con-
fronting the Newtonians in the eighteenth century concerned the
ontology of forces. How, critics such as Leibniz and Huygens
had asked, can bodies exert force at points far removed from
the bodies themselves? What substance carries the attractive
force of the sun through 90 million miles of empty space so that
the earth is pulled towards it? How, at the more prosaiclevel,
can a magnet draw towardsitself a piece of iron several inches
away? Such phenomena seemed to defy the very logic of
speaking about substances and properties since properties (e.g.,
the power of attraction) seemed to be capable of disentangling
themselves from the material bodies of which they were the
 62 CONCEPTUAL PROBLEMS

properties. As Buchdahl,'® Heimann and McGuire'’ have

convincingly argued, sorting out this issue became one of the
central philosophical and scientific problems of the Enlighten-
ment. Not satisfied with the Cotesian denial that this was an
acute conceptual problem (Cotes was prepared to say that
nature was generally unintelligible and that the unintelligibility
of distance forces was no particular source of cognitive con-
cern'*), philosophers and scientists all over Europe began to
re-evaluate such traditional issues as the nature of substance,
the relations of properties to substances, and, particularly, the
1 etnnn ediene

nature of our knowledge of substance. What resulted from this

reappraisal at the hands of Kant, Priestley, Hutton, and
others
was a new ontology which argued for the priority of force over
matter and which made the powers of activity (rather than
passive powers like mass and inertia) into the basic building
blocks of the physical world. The emergence of this new
ontology did several things at once: it eliminated the most acute
conceptual problem for Newtonian science by exhibiting
the
“intelligibility” of action-at-a-distance; it brought the ontology
of philosophy and the ontology of physics back into harmony;
and it made possible the subsequent emergence of theories of
the physical field.*?
Those “‘positivist’’ philosophers and historians of science who
see the progress of science entirely in empirical terms have
completely missed the huge significance of these developments
for science as well as for philosophy. Convinced that meta-
physics is foreign, even alien, to the development of scientific
ideas, they have written about the history of Newtonianism
without even perceiving thevital bearing of these metaphysical
controversies on the historical career of Newtonian doctrines.
Traditionally, worldview difficulties have tended to emerge
most often as a result of tensions between science, on the one
hand, and either theology, philosophy or social theory, on
the
other hand.”° It is well known, for instance, that one
of the
major difficulties for the mechanistic scientific program of the
seventeenth and eighteenth centuries was the perceived discrep-
ancy between a theory which reduced the cosmos to a
self-
operating machine and certain ‘‘activist”’ theologies which
CONCEPTUAL PROBLEMS 63

sought to preserve an important role to God in the day-to-day

maintenance of the universe. The famous Leibniz-Clarke Cor-
respondence, one of the major documentsof the early Enlight-
enment, is replete with controversies that illustrate what I call
world-view difficulties. Similarly, one major stumbling block to
the emergence of evolutionary theory was the conviction, based
on the best available philosophical insights, that species must be
separate and distinct.*' More recently, one of the most
persistent set of conceptual problems in twentieth century
physics has been the dissonance between quantum mechanics
and our ‘‘philosophical”’ beliefs about causality, change, sub-
stance and “reality.”
It is not only incompatibilities between science and philos-
ophy or between science and theology which can lead to world-
view difficulties. Conflicts with a social or moral ideology can
produce similar tensions. In our own time, for instance, there
are several instances where seemingly serious arguments have
been lodged against a scientific theory because of moral or
ethical worldview difficulties. In the Soviet Union, the Lysenko
affair is a case in point. Because evolutionary biology, with its
denial of the transmission of acquired characteristics, ran
counter to the Marxist view that man’s very nature could be
changed by his environment, there were strong reservations
voiced against Darwinism and Mendelism and much support
was given to a scientific research effort like Lysenko’s which
sought to find scientific evidence for the Marxist philosophy of
man. In the West, similar constraints have recently confronted
researchers and theorists examining the possibility of racial
differences. It has been suggested that any scientific theory
which would argue for differences of ability or intelligence
between the various races must necessarily be unsound, because
such a doctrine runs counter to our egalitarian social and
political framework.
There is a prominent group of thinkers in contemporary
science and philosophy who have argued that world view
difficulties are only pseudo-problems.’? They claim that scien-
tific theories can stand alone and that any element of our
worldview which does not square with science should simply be
64 CONCEPTUAL PROBLEMS

abandoned. I shall take issue in the next chapter with this

positivistic doctrine, but for now, I should make a few
disclaimers, lest I be taken for asserting more than I am:
1. It is not my claim that a scientific theory should necessar-
ily be abandoned when it encounters worldview problems; in
asserting the existence of conceptual problems of this type, I am
only asserting the fact that a tension often exists between our
‘scientific’ beliefs and our “‘nonscientific’’ ones, and that such
a tension does pose a problem for both sets of beliefs. How that
tension is to be resolved depends on the particularities of the
case.
2. It is not my claim that every worldview problem con-
stitutes a serious ground for reservations about a scientific
theory. How serious the problem is for the theory depends upon
how well entrenched the nonscientific belief is and upon what
problem-solving capabilities we would lose by abandoningit.

The Relative Weighting of Conceptual Problems

Having examined in a little more detail how conceptual
problems are generated, we can now think about how to assess
their relative importance. It is vital to stress, at the outset, that
a conceptual problem will, in general, be a more serious one
than an empirical anomaly. No one, for instance, proposed
abandoning Newtonian mechanics when it could not accurately
predict the motion of the moon. But many thinkers (such as
Leibniz, Huygens, and Wolff) were seriously prepared to
dismiss Newtonian physics because its ontology was incompat-
ible with the accepted metaphysics of the day. This difference
in weighting arises, not because science is more rationalistic
than empirical; but rather becauseit is usually easier to explain
away an anomalous experimental result than to dismiss out of
hand a conceptual problem.”? (Let me add that I am not
suggesting that all conceptual problems are more important
than all empirical problems. I am rather making the more
modest claim that most conceptual problems are of greater
moment than most empirical anomalies.)
Within the domain of conceptual problems, there are certain
circumstances which tend to promote or demote the initial
CONCEPTUAL PROBLEMS 65

importance of such problems. There are at least four situations

which should be distinguished here:
1. As we have already seen, the nature of the logical relation
between two theories exhibiting a conceptual problem can vary
enormously from inconsistency (in its most acute form) to
mutual support. Other things being equal, the greater the
tension between two theories, the weightier the problem will be.
2. When a conceptual problem arises as a result of a conflict
between two theories, T, and T,, the seriousness of that
problem for 7, depends on how confident we are about the
acceptability of T,. If T, has proven to be extremely effective at
solving empirical problems and if its abandonment would leave
us with many anomalies, then matters are very difficult for the
proponents of 7,. If, on the other hand, 7,’s record as a
problem solver is very modest, then T,'s incompatibility with
T, will probably not count as a major conceptual problem
for T,.
3. Another case in which it becomes meaningful to speak of
the grading of conceptual problems on a scale of importance
occurs when—within a particular scientific domain—we have
two competing (as opposed to complementary) theories, T, and
T,. If both T, and T, exhibit the same conceptual problem(s),
then those problems count no more against one than against the
other and become relatively insignificant in the context of
comparative theory appraisal. However, if T, generates certain
conceptual problems which T, does not, then those problems
become highly significant in the appraisal of the relative merits
of T, and T,.
4. A final determinant of the importance of a conceptual
problem (as with anomalies) has to do with the ‘‘age’’ of that
problem. If it has only recently been discovered that a theory
poses a certain conceptual problem (for instance, an internal
inconsistency), there is usually some groundsfor hope that, with
very minor modifications in the theory, we can bringit into line
and thus eliminate the problem. The threat which the problem
poses to the theory is generally offset by an optimism that it can
be readily dealt with—an optimism that is often justified. If, on
the other hand, a theory has been known to have a particular
conceptual problem for some length of time, if partisans of that
 66 CONCEPTUAL PROBLEMS

theory have tried, repeatedly and unsuccessfully, to make the

theory consistent, or to reconcile it with our norms and our
other accepted beliefs, then that problem assumes an ever
greater importance with time, and assumes an ever greater
significance in debates about the acceptability of the theory (or
theories) which generate(s) it.

Summary and Overview

Quite simply, the claim of this chapter is that no major
contemporary philosophyof science allows scope for the weighty
tole which conceptual problems have played in the history of
science. Even those philosophers who claim to take the actual
evolution of scienceseriously (e.g., Lakatos, Kuhn, Feyerabend,
and Hanson) have made no serious concessions to the non-
empirical dimensions ofscientific debate. We now know enough
about the importance of these nonempirical factors within the
evolution of science to say with some confidence that any theory
about the nature of science which finds no role for conceptual
problems forfeits any claim to being a theory about how science
has actually evolved.
Although the analytic machinery thus far developed is still
insufficent for constructing a general model of scientific
progress and growth, we now possess enough pieces of the
puzzle that we can begin to talk in an approximative way
about what a problem-solving model of progress might look
like. The core assumptions of such a model are simple ones: (1)
the solved problem—empirical or conceptual—is the basic unit
of scientific progress; and (2) the aim of science is to maximize
the scope of solved empirical problems, while minimizing the
scope of anomalous and conceptual problems.
The more numerous and weightier the problems are which a
theory can adequately solve, the better it is. If one theory can
solve more significant problems than a competitor, then it is
preferable to it. At one level, this is a noncontroversial claim. If
we interpret problems exclusively in the sense of what we have
called “solved empirical problems,’’ many philosophers of
science would accept that progress does amount to the solution
of such problems. But, as we have seen, there are problems in
CONCEPTUAL PROBLEMS 67

science other than solved empirical ones, specifically anomalous

and conceptual problems. My definition of progress chiefly be-
comes controversial (and potentially interesting) when we inter-
pret it as applying to the latter as well as to the former. My
reasons for wishing to broaden the basein this way should now be
clear. If it counts in favor of a theory when it can accumulate
solved empirical problems (as the standard view allows), then it
should also count against a theory if it generates anomalous and
conceptual problems. Indeed, the problem-solving effectiveness
of a theory depends on the balance it strikes between its solved
problems and its unresolved problems. How exactly does this
work?
Let us begin with a very crude model of scientific evolution.
Imagine some domain in which we notice a certain puzzling
phenomenon, p. The phenomenon p constitutes an unsolved
problem for the scientist who wishes to develop a theory, T7,,
specifically with a view toward resolving p. Once T, is
announced, several things are likely to happen simultaneously.
Some fellow scientist may observe that 7, predicts other
phenomena in the domain besides p. These predictions will be
tested, and, very often, some of them will not be borne out in
our observation. Thus, the observation of these discrepant
results will constitute one or more anomalies for 7,. At the same
time, it may be pointed out that 7, makes certain assumptions
about natural processes which run counter to some of our most
widely accepted theories, or that it is incompatible with our
methodological norms. This will constitute one or more concep-
tual problems for theory T,.
Thus far in this imaginary chronology, we are not clear
whether any progress has been made. It is true that 7, has
solved its initial empirical problem, p, and to that extent, we
can say that ‘‘progress” has been made. Unfortunately, how-
ever, the very theory, Z7,, which cleared up that problem, has
generated several others; in this case, anomalies and conceptual
problems. It is entirely possible that more serious problems
have been generated than resolved by the invention of 7,. But
let us carry the example through in time for a while. Suppose
that a second theorist comes along who is convinced that he can
improve 7,. What does improving T, mean? Very roughly, such
68 CONCEPTUAL PROBLEMS

improvement would be exhibited by showing that a new theory,

T,, could explain the initial empirical problem of T, without
generating the same, or as many, anomalies and conceptual
problems as T, produced. If 7, managed to do as much workat
the empirical problem level as T, did, without all of 7T,’s
attendant empirical and conceptual difficulties, we could all
agree that it would be more reasonable to accept 7, than to
accept 7,; that, indeed, the acceptance of T, was progressive
and that the continued espousal of 7, was unprogressive or
regressive.
Generalizing from this simple example, we could define an
appraisal measure for a theory in the following way: the overall
problem-solving effectiveness of a theory is determined by
assessing the number and importance of the empirical problems
which the theory solves and deducting therefrom the number
and importance of the anomalies and conceptual problems
which the theory generates.
The step from here to a rudimentary notion of scientific
progress is straightforward. Given that the aim of science is
problem solving (or, more precisely, the mini-max strategy
sketched above), progress can occur if and only if the succession
of scientific theories in any domain shows an increasing degree
of problem solving effectiveness. Localizing the notion of
progress to specific situations rather than to large stretches of
time, we can say that any time we modify a theory or replace it
by another theory, that change is progressive if and only if the
later version is a more effective problem solver (in the sense Just
defined) than its predecessor.
There are many ways in which such progress can occur. It
may come about simply by an expansion of the domain of solved
empirical problems with all the other appraisal vectors remain-
ing fixed. In such a case, the replacement of 7, by 7, (which
solves more empirical problems) is clearly progressive. Progress
can also result from a modification of the theory which
eliminates some troublesome anomalies or which resolves some
conceptual problems. Most often, of course, progress occurs as
a result of all the relevant variables shifting subtly.
Given the exclusive emphasis by most philosophers on
empirical problems, and their solution, it is important to stress
 69
CONCEPTUAL PROBLEMS

that, on the model outlined here, (1) progress can occur without
an expansion of the domain of solved empirical problems, and
is even conceivable when the domain of such problems con-
tracts; and (2) a theory change may conceivably be non-progres-
sive or regressive, even when the index of solved empirical
problems increases, specifically, if the change leads to more
acute anomalies or conceptual problems confronting the new
theory than those exhibited by the predecessor theory.
Although an outline of a theory of cognitive progress is now
emerging, there is still one crucial dimension missing. In ali the
talk about problem solving, there has been some confusion
about what kinds of things solve problems. I have been using
the term “theory” to designate those complexes whose problem-
solving capacities must be appraised; in order to clarify the
types of problems in science, I have had to postpone a discus-
sion about what kinds of things can solve problems. We must
examine that side of the problem-solving equation before the
rough-hewn model of progress outlined here can be refined into
a valuable tool of analysis.
 Chapter Three
From Theories to
Research Traditions

The intellectualfunction of
an established conceptual
Scheme is to determine the
patterns of theory, the mean
-
ingful questions, the legitima
te
interpretations... 5. TOULMI
N (1970), p. 40

Theories are inevitably involved

in the solution of problems; the
very aim of theorizing is to
provide coherent and adequa
solutions to the empirical pro te
blems which stimulate inquiry.
Theories, moreov er, are designed to avoid (or to
various conceptual and anomal resolve) the
ous problems which their pred
cessors generate. If one looks e-
at inquiry in this way, if one
theories from views
this perspective, it become
s clear that the central
cognitive test of any theory invo
lves assessing its adequacy as
solution of certain emptrical a
and conceptual problems. Hav
developed in earlier chapters ing
a taxonomy for describing
kinds of problems which the
confront theories, we must now lay
down adequacy conditions
for determining when a theo
Provides an acceptable solu ry
tion to the problems which
front it. con-
But before we embark
on that task, we must
theories are and how they clarify what
function, for a failure to
make some
70
FROM THEORIES TO RESEARCH TRADITIONS 71

rudimentary distinctions here has brought grief to more than

one major philosophy of science. Entire books have been
devoted to the structure of scientific theory; I am attempting
nothing that ambitious. Rather, I shall want to insist on only
two major points with respect to an analysis of theories.
In the first place, to make explicit what has bten implicit all
along, the evaluation of theories is a comparative matter. What
is crucial in any cognitive assessment of a theory is how it fares
with respect to its competitors. Absolute measures of the
empirical or conceptual credentials of a theory are of no
significance; decisive is the judgment as to how a theory stacks
up against its known contenders. Much of the literature in the
philosophy of sctence has been based upon the assumption that
theoretical evaluation occurs in a competitive vacuum. By
contrast, I shall be assuming that assessments of theories
always involve comparative modalities. We ask: is this theory
better than that one? Is this doctrine the best among the
available options?
The second major claim of this chapter is that it is necessary
to distinguish, within the class of what are usually called
“scientific theories,’’ between two different sorts of proposi-
tional networks.
In the standardliterature on scientific inference, as well as in
common scientific practice, the term ‘‘theory”’ refers to (at least)
two very types of things. We often use the term ‘‘theory’’ to
denote a very specific set of related doctrines (commonly called
“hypotheses” or ‘‘axioms’’ or “principles’’) which can be
utilized for making specific experimental predictions and for
giving detailed explanations of natural phenomena. Examples
of this type of theory would include Maxwell's theory of
electromagnetism, the Bohr-Kramers-Slater theory of atomic
structure, Einstein’s theory of the photoelectric effect, Marx's
labor theory of value, Wegener’s theory of continental drift, and
the Freudian theory of the Oedipal complex.
By contrast, the term “theory” is also used to refer to much
more general, much less easily testable, sets of doctrines or
assumptions. For instance, one speaks about ‘“‘the atomic
theory,” or “‘the theory of evolution,” or ‘‘the kinetic theory of
gases."’ In each of these cases, we are referring not to a single
theory, but to a whole spectrum of individual theories. The term

De
 72 FROM THEORIES TO RESEARCH TRADIT
IONS
“evolutionary theory”’ for instance, does not refer
to any single
theory but to an entire family of doctrines, histori
cally and
conceptually related, all of which work
from the assumption
that organic species have common lines of descent.
Similarly,
the term “atomic theory” generally refers
to a large set of
doctrines, all of which are predicated
on the assumption that
matter is discontinuous. A particularly vivid
instance of one
theory which includes a wide variety of specifi
c instantiations is
offered by recent ‘quantum theory.” Since
1930, that term has
included (among other things) quantum
field theories, group
theories, so-called S-matrix theories,
and renormalized field
theories—between any two of which there are
huge conceptual
divergences.
The differences between the two types of
theories outlined
above are vast: not only are there contrasts
of generality and
specificity between them, but the modes
of appraisal and
evaluation appropriate to each are radically
different. It will be
the central claim of this chapter that until
we become mindful
of the cognitive and evaluational differences
between these two
types of theories, it will be impossible to
have a theory of
scientific progress which is historically sound
or philosophically
adequate.
But it is not only fidelity to scientific practi
ce and usage
which requiresus to take these larger theore
tical units seriously.
Muchof the research done by historians and
philosophers of
science in the last decade suggests that these
more general units
of analysis exhibit many of the epistemic
features which,
although most characteristic of science,
elude the analyst who
limits his range to theories in the narrower
sense. Specifically,
it has been suggested by Kuhn and Lakat
os that the more
general theories, rather than the more specifi
c ones, are the
Primary tool for understanding and
appraising scientific
progress.
I share this conviction in principle, but
find that the accounts
hitherto given of what these larger theori
es are, and how they
evolve, are not fully satisfactory. Becau
se the bulk of this
chapter will be devoted to outlining a new
account of the more
global theories (which I shall be calling resear
ch traditions), it is
appropriate that I should indicate what
I find chiefly wanting in
FROM THEORIES TO RESEARCH TRADITIONS 73

the best known efforts to grapple with this problem. Of the

many theories of scientific evolution that have been developed,
two specifically address themselves to the question of the nature
of these more general theories.

Kuhn’s Theory of Scientific ‘‘Paradigms’’

In his influential Structure of Scientific Revolutions, Thomas
Kuhn offers a model of scientific progress whose primary
element is the ‘‘paradigm.” Although Kuhn’s notion of para-
digms has been shown to be systematically ambiguous’ (and
thus difficult to characterize accurately), they do have certain
identifiable characteristics. They are, to begin with, ‘‘ways of
looking at the world’; broad quasi-metaphysical insights or
hunches about how the phenomena in some domain should be
explained. Included under the umbrella of any well-developed
paradigm will be a number of specific theories, each of which
presupposes one or more elements of the paradigm. Once a
paradigm is accepted by scientists (and one of Kuhn’s more
extreme claimsis that in any ‘‘mature’’ science,’ every scientist
will accept the same paradigm most of the time), they can
proceed with the process of ‘“‘paradigm articulation,” also
known as “normal science.’’ In periods of normal science, the
dominant paradigm will itself be regarded as unalterable and
immune from criticism. Individual, specific theories (which
represent efforts ‘‘to articulate the paradigm,” i.e., to apply it
to an ever wider range of cases) may well be criticized,
falsified and abandoned; but the paradigm itself is unchal-
lenged. It remains so until enough ‘anomalies’? accumulate
(Kuhn never indicates how this point is determined) that
scientists begin to ask whether the dominant paradigm is
really appropriate. Kuhn calls this time a period of ‘‘crisis.”’
During a crisis, scientists begin for the first time to consider
seriously alternative paradigms. If one of those alternatives
proves to be more empirically successful than the former
paradigm, a scientific revolution occurs, a new paradigm is
enthroned, and anotherperiod of normal science ensues.
There is much that is valuable in Kuhn’s approach. He
recognizes clearly that maxi-theories have different cognitive
 74 FROM THEORIES TO RESEARCH TRADITIONS

and heuristic functions than mini-theories.

He has probably
been the first thinker to stress the tenac
ity and persevering
qualities of global theories—even when confronted
with serious
anomalies.‘ He has correctly rejected the
(widely assumed)
cumulative character of science.> But for
all its many Strengths,
Kuhn's model of scientific progress suffer
s from some acute
conceptual and empirical difficulties. For
instance, Kuhn's
account of paradigms and their careers has
been extensively
criticized by Shapere, who has highlighte
d the obscure and
opaque character of the paradigm itself
by pointing out many
inconsistencies in Kuhn's use of the notio
n.® Feyerabend’ and
others have stressed the historical incor
rectness of Kuhn's
Stipulation that “normal science” is
in any way typical or
normal. Virtually every major period
in the history of science
is characterized both by the co-existen
ce of numerous com-
peting paradigms, with none exerting
hegemony over the field,
and by the persistent and continuous
manner in which the
foundational assumptions of every paradigma
re debated within
the scientific community. Numerous critic
s have noted the
arbitrariness of Kuhn's theory ofcrisis: if
(as Kuhn Says) a few
anomalies do not produce crisis, but
“many” do, how does
the scientist determine the “crisis point?
’’ There are other
serious flaws as well. In my view, the most
significant of these
are:
1. Kuhn's failure to see the role of conce
ptual problems in
Scientific debate and in paradigm evaluation
. Insofar as Kuhn
grants that there are anyrational criteria
for paradigm choice,
or for assessing the “progressiveness”
of a paradigm, those
criteria are the traditional positivist ones
such as: Does the
theory explain more facts than its prede
cessor? Can it solve
some empirical anomalies exhibited by
its predecessor? The
whole notion of conceptual problems and
their connection with
progress finds no serious exemplification
in Kuhn’s analysis.
2. Kuhn never really resolves the crucia
l question of the
relationship between a Paradigm and its
constituent theories.
Does the paradigm entail or merely
inspire its constituent
theories? Do these theories, once developed,
justify the para-
digm, or does the paradigm justify them?
It is not even clear,
+"

FROM THEORIES TO RESEARCH TRADITIONS 75

in Kuhn’s case, whether a paradigm precedes its theories or

arises nolens volens after their formulation. Although this issue
is extremely complex, any adequate theory of science is going to
have to come to grips with it more directly than Kuhn has.
3. Kuhn’s paradigms have a rigidity of structure which
precludes them from evolving through the course of time in
response to the weaknesses and anomalies which they generate.
Moreover, because he makes the core assumptions of the
paradigm immune from criticism, there can be no corrective
relationship between the paradigm and the data. Accordingly,
it is very difficult to square the inflexibility of Kuhnian
paradigmswith the historical fact that many maxi-theories have
evolved through time.
4. Kuhn’s paradigms, or “‘disciplinary matrices,”’ are always
implicit, never fully articulated.* As a result, it is difficult to
understand how he can account for the many theoretical
controversies which have occurred in the development of
science, since scientists can presumably only debate about
assumptions which have been made reasonably explicit. When,
for instance, a Kuhnian maintains that the ontological and
methodological frameworks for Cartesian or Newtonian physics,
for Darwinian biology, or for behavioristic psychology were
only implicit and never received overt formulation, he is
running squarely in the face of the historical fact that the core
assumptions of all these paradigmswere explicit even from their
inception.
5. Because paradigms are so implicit and can only be
identified by pointing to their ‘‘exemplars” (basically an
archetypal application of a mathematical formulation to an
experimental problem), it follows that whenever two scientists
utilize the same exemplars, they are, for Kuhn, ipso facto
committed to the same paradigm. Such an approach ignores
the persistent fact that different scientists often utilize the same
laws or exemplars, yet subscribe to radically divergent views
about the most basic questions of scientific ontology and
methodology. (For instance, both mechanists and energeticists
accepted identical conservation laws.) To this extent, analysing
science in terms of paradigms is unlikely to reveal that ‘strong
 76 FROM THEORIES TO RESEARCH TRADITIONS
network of commitments—conceptual,
theoretical, instrumen-
tal, and metaphysical’? which Kuhn hoped
to localize with his
theory of paradigms.

Lakatos’ Theory of ‘Research Programmes’”’

Largely in response to Kuhn’s assault
on some of the
cherished assumptions oftraditional philos
ophyof science, Imre
Lakatos has developed an alternative theor
y about the role of
these “super-theories” in the evolution of
science. Calling such
general theories “research programmes,”
Lakatos argues that
research programmes have three elements:
(1) a “hard-core”
(or “negative heuristic’) of fundamenta
l assumptions which
cannot be abandoned or modified witho
ut repudiation of the
research programme;'® (2) the “positive heuristic,”” which
contains “a partially articulated set of
suggestions or hints on
how to change, .. . modify, sophisticate
[sic] our specific
theories whenever we wish to improve
them, and (3) ‘‘a series
of theories, T,, T,, T,, . . .”’ where each
subsequent theory
“results from adding auxiliary clauses
to. . . the previous
theory.’’'? Such theories are the Specific
instantiations of the
general research programme. Research
programmes can be
progressive or regressive in a variety of ways:
but progress, for
Lakatos even more than for Kuhn, is
a function exclusively of
the empirical growth of a tradition. It
is the possession of
greater “empirical content,” or of a highe
r ‘‘degree of empirical
corroboration’’ which makes one theory
superior to, and more
progressive than, another.
Lakatos’ model is, in many respects, a decid
ed improvement
on Kuhn’s. Unlike Kuhn, Lakatos
allows for, and stresses, the
historical importance of the co-existence
of several alternative
research programmes at the same
time, within the same
domain. Unlike Kuhn, whooften takes the
view that paradigms
are incommensurable™ and thus not open
to rational compari-
son, Lakatosinsists that we can objectively
comparetherelative
progress of competing research traditions.
More than Kuhn,
Lakatostries to grapple with the thorny
question of the relation
of the super-theory to its constituent mini-
theories.
FROM THEORIES TO RESEARCH TRADITIONS 77

But against that, Lakatos’ model of research programmes

shares many of the flaws of Kuhn’s paradigms, and introduces
somenew onesas well:
1. As with Kuhn, Lakatos’ conception of progress is exclus-
ively empirical; the only progressive modifications in a theory
are those which increase the scope of its empirical claims.
2. The sorts of changes which Lakatos allows within the
mini-theories which constitute his research programme are
extremely restricted. In essence, Lakatos only permits, as the
relation between any theory and its successor within a research
programme, the addition of a new assumption or a semantic
re-interpretation of terms in the predecessor theory. On this
remarkable view of things, two theories can only be in the same
research programme if one of the two entails the other. As we
shall see shortly, in the vast majority of cases, the succession
of specific theories within a maxi-theory involves the elimination
as well as the addition of assumptions, and there are rarely
successor theories which entail their predecessors.
3. A fatal flaw in the Lakatosian notion of research pro-
grammesis its dependence upon the Tarski-Popper notions of
“empirical and logical content.’’ A// Lakatos’ measures of
progress require a comparison of the empirical content of every
member of the series of theories which constitutes any research
programme.'* As Griinbaum and others have shown convinc-
ingly, the attempt to specify content measures for scientific
theories is extremely problematic if not literally impossible.'®
Because comparisons of content are generally impossible,
neither Lakatos nor his followers have been able to identify any
historical case to which the Lakatosian definition of progress
can be shownstrictly to apply.’®
4. Because of Lakatos’ idiosyncratic view that the acceptance
of theories can scarcely if ever be rational, he cannot translate
his assessments of progress (assuming he could make them!)
into recommendations about cognitive action.'’ Although one
research programme may be more progressive than another,
we can, on Lakatos’ account, deduce nothing from that about
which research programme should be preferred or accepted.
As a result, there can never be a connection between a theory
 78 FROM THEORIES TO RESEARCH TRADITIONS

of progress and a theory of rational acceptability (or, to use

Lakatos’ language, between methodological “‘appraisal’’ and
“tadvice’’).
5. Lakatos’ claim that the accumulation of anomalies has no
bearing on the appraisal of a research programme is massively
refuted by the history of science.
6. Lakatos’ research programmes, like Kuhn’s paradigms,
are rigid in their hard-core structure and admit of no funda-
mental changes.'*
What should beclear, even from this very brief survey of two
of the major theories of scientific change, is that there
are a
number of analytical and historical difficulties confronting
existing attempts to understand the nature and role of
maxt-
theories. With some of those difficulties in mind, we can
turn
now to explore an alternative model ofscientific progress,
built
upon elements outlined in the previous chapters. One crucial
test of that model will be whether it can avoid some of
the
problems which handicap its predecessors. Although there
are
numerous common elements between my model and those
of
Kuhn and Lakatos (and I readily concede a great
debt to
their pioneering work), there are a sufficiently large number
of
differences that I shall try to develop the notion of a research
tradition more or less from scratch.

The Nature of Research Traditions

We havealready referred to a few classic research traditio
ns:
Darwinism, quantum theory, the electromagnetic theory
of
light. Every intellectual discipline, scientific as well as nonscie
n-
tific, has a history replete with research traditions: empiricism
and nominalism in philosophy, voluntarism and necessitarian-
ism in theology, behaviorism and Freudianism in psychology,
utilitarianism and intuitionism in ethics, Marxism and capital-
ism in economics, mechanism and vitalism in physiology,
to
name only a few. Such research traditions have a number
of
common traits:
1. Every research tradition has a number of specific theorie
s
which exemplify and partially constitute it; some of
these
theories will be contemporaneous, others will be tempor
al
successors of earlier ones;
FROM THEORIES TO RESEARCH TRADITIONS 79

2. Every research tradition exhibits certain metaphysical and

methodological commitments which, as an ensemble, individu-
ate the research tradition and distinguish it from others;
3. Each research tradition (unlike a specific theory) goes
through a number of different, detailed (and often mutually
contradictory) formulations and generally has a long history
extending through a significant period of time. (By contrast,
theories are frequently short-lived.)
These are by no means the only important characteristics of
research traditions, but they should serve, for the time being, to
identify the kinds of objects whose properties I would like to
explore.
In brief, a research tradition provides a set of guidelines for
the development of specific theories. Part of those guidelines
constitute an ontology which specifies, in a general way, the
types of fundamental entities which exist in the domain or
domains within which the research tradition is embedded. The
function of specific theories within the research tradition is to
explain all the empirical problems in the domain by “‘reducing”’
them to the ontology of the research tradition. If the research
tradition is behaviorism, for instance, it tells us that the only
legitimate entities which behavioristic theories can postulate are
directly and publicly observable physical and physiological
signs. If the research tradition is that of Cartesian physics, it
specifies that only matter and minds exist, and that theories
whichtalk of other types of substances(or of ‘‘mixed”’ mind and
matter) are unacceptable. Moreover, the research tradition
outlines the different modes by which these entities can interact.
Thus, Cartesian particles can only interact by contact, not by
action-at-a-distance. Entities, within a Marxist research tradi-
tion, can only interact by virtue of the economic forces
influencing them.
Very often, the research tradition will also specify certain
modes of procedure which constitute the legitimate methods of
inquiry open to a researcher within that tradition. These
methodological principles will be wide-ranging in scope, addres-
sing themselves to experimental techniques, modes of theoret-
ical testing and evaluation, and the like. For instance, the
methodological posture of the scientist in a strict Newtonian
80 FROM THEORIES TO RESEARCH TRADITIONS

research tradition is inevitably inductivist, allowing for the

espousal of only those theories which have been ‘‘inductively
inferred’ from the data. The methods of procedure outlined
for a behavioristic psychologist are what is usually called
“operationalist."” Put simplistically, a research tradition is thus
a set of ontological and methodological “do's” and ‘‘don'ts.”’
To attempt what is forbidden by the metaphysics and metho-
dology of a research tradition is to put oneself outside that
tradition and to repudiate it. If, for instance, a Cartesian
physicist starts talking about forces acting-at-a-distance, if a
behaviorist starts talking about subconscious drives, if a
Marxist begins speculating about ideas which do not arise in
response to the economic substructure; in each of these cases,
the activity indicated puts the scientist in question beyond the
pale. By breaking with the ontology or the methodology of the
research tradition within which he has worked, he has violated
the strictures of that research tradition and divorced himself
from it. Needless to say, that is not necessarily a bad thing.
Some of the most important revolutions in scientific thought
have come from thinkers who had the ingenuity to break with
the research traditions of their day and to inaugurate new ones.
But what we must preserve, if we are to understand either the
logic or the history of the natural sciences, is the notion of the
integrity of a reseach tradition, for it is precisely that integrity
which stimulates, defines and delimits what can count as a
solution to many of the most important scientific problems. *°
Althoughit is vital to distinguish between the ontological and
the methodological components of a research tradition, the two
are often intimately related, and for a very natural reason:
namely, that one’s views about the appropriate methods of
inquiry are generally compatible with one’s views about the
objects of inquiry. When, for instance, Charles Lyell defined the
“uniformitarian’’ research tradition in geology, his ontology was
restricted to presently acting causes and his methodology
insisted that we should “explain past effects in terms of
presently acting causes.”’ Without a “‘presentist’”’ ontology, his
uniformitarian methodology would have been inappropriate;
and without the latter, the presentist ontology would not have
FROM THEORIES TO RESEARCH TRADITIONS 81

allowed Lyell to explain the geological past. Similarly, the

mathematical ontology of the Cartesian research tradition (an
ontology which argued that ail physical changes were entirely
changes of quantity) was very closely connected with the
(mathematically inspired) deductivist and axiomatic metho-
dology of Cartesianism. As we shall see later, it does not always
happen that the ontology and methodology of a research
tradition are so closely intertwined (for instance, the inductivist
methodology of the Newtonian research tradition had only the
weakest of connections with that tradition’s ontology), but such
cases are the exception rather than the rule.
So a preliminary, working definition of a research tradition
could be put as follows: a research tradition is a set of general
assumptions about the entities and processes in a domain of
study, and about the appropriate methods to be used for
investigating the problems and constructing the theories in that
domain.

Theories and Research Traditions

Every research tradition will be associated with a series of
specific theories, each of which is designed to particularize the
ontology of the research tradition and to illustrate, or satisfy, its
methodology. The mechanistic research tradition in seventeenth
century optics, for example, includes several of Descartes’
theories as well as the optical theories of Hooke, Rohault,
Hobbes, Régis, and Huygens.*° The phlogistic tradition in
eighteenth century chemistry received more than a dozen
specific theoretical formulations.*! Many of the theories within
any evolving research tradition will be mutually inconsistent
rivals, precisely because sometheories represent attempts, within
the framework of the tradition, to improve and correct their
predecessors.
The individual theories constituting the tradition will gener-
ally be empirically testable for they will entail (in conjunction
with other specific theories) some precise predictions about how
objects in the domain will behave. By contrast, research
traditions are neither explanatory, nor predictive, nor directly
 82 FROM THEORIES TO RESEARCH TRADITIONS

testable. Their very generality, as well as their normative

elements, precludes them from leading to detailed accounts of
specific natural processes.
Except at the abstract level of specifying what the world is
made of, and how it should be studied, research traditions do
not provide detailed answers to specific questions. A research
tradition will not tell us what happens to light when it is
refracted at an interface between water and air: it will not tell
us what happens if we put an eight-month-old female mouse
into a maze; it will not tell us why lead melts at a lower
temperature than copper. But it would be a mistake to conclude
from the fact that research traditions do not offer solutions to
specific problems that they are outside of the problem-solving
process. To the contrary, the whole function of a research
tradition is to provide us with the crucial tools we need for
solving problems, both empirical and conceptual. (As we shall
see later, the research tradition even goes so far as to define
partially what the problems are, and what importance should be
attached to them.) It is for just this reason that the objective
evaluation of any research tradition is crucially linked with the
problem-solving process. The very idea that an entity like a
research tradition—which makes no predictions, which solves
no specific problems, which is fundamentally normative and
metaphysical—could be objectively evaluated may seem para-
doxical. But nothing could be further from the case, for we can
say quite simply that a successful research tradition is one
which leads, via its component theories, to the adequate
solution of an increasing range of empirical and conceptual
problems. Determining whether a tradition is successful in this
sense does not mean, of course, that the tradition has been
“confirmed”’ or “‘refuted."’ Nor can such an appraisal tell us
anything about the truth or falsity of the tradition.?? A research
tradition may be enormously successful at generating fruitful
theories and yet flawed in its ontology or methodology. Equally,
one can conceive that a research tradition might be true, and
yet (perhaps because of the unimaginativeness of its propon-
ents) unsuccessful at generating theories which were effective
problem solvers. Hence to abandonor reject a reseach tradition
FROM THEORIES TO RESEARCH TRADITIONS 83

is not (or ought not be) to judge that tradition false. Nor, in
rejecting a research tradition as momentarily unsuccessful, are
we necessarily relegating it to permanent oblivion; to the
contrary, we can explicitly stipulate conditions which, if
satisfied, would revive and recussitate it. Thus, when wereject
a research tradition, we are merely making a tentative decision
not to utilize it for the moment becausethere is an alternative to
it which has proven to be a more successful problem solver.
Just as the fortunes of a research tradition are linked closely
to the problem-solving effectiveness of its constituent theories,
so too is the determination of the adequacy of a specific
theory inextricably bound up with an assessment of the
problem-solving effectiveness of the entire set of theories
spawned by the research tradition of which that theory is a
part.?> If a theory is closely linked to an unsuccessful research
tradition, then—whatever the problem-solving merits of that
particular theory—it is likely to be regarded as highly suspect.
For instance, Count Rumford’s theories of heat conduction and
convection were far superior to any alternative theories of
thermalflow in fluids available in the period from 1800 to 1815.
Nonetheless, few scientists took Rumford’s theories seriously
because (as they saw it) the research tradition in which
Rumford worked (deriving from Boerhaave) had been discred-
ited by the emergence of rival research traditions in chemistry
(especially Joseph Black’s), which suggested that heat was a
substance rather than, as Rumford imagined, the random
motion of particles. Rumford’s specific theories only became
fashionable in the 1840s and 1850s because by that time the
balance between various research traditions had shifted suffic-
iently that many scientists were more prepared to consider
seriously specific theories (like Rumford’s) which grew out of a
kinetic research tradition.
Contrariwise, a theory, even an inadequate one, will have
some strong argumentsin its favorif it is linked with a research
tradition that is otherwise highly successful. Thus, theories of
mechanistic physiology in the late seventeenth century (such as
those of Borelli and Pitcairn) were highly regarded in many
circles where the mechanistic research tradition was flourishing
84 FROM THEORIES TO RESEARCH TRADITIONS

even though, judged entirely on their own merits, they were

significantly inferior to certain theories in other, less successful
research traditions. ”*
Up to this point, I have been deliberately vague about
describing the kind of relation which exists between a theory
and its “‘parent™’ research tradition. 1 have spoken of research
traditions “‘inspiring’’ or ‘‘containing”’ or “‘generating”’ theories,
and about theories “‘presupposing” or “‘constituting’’ or even
“defining” research traditions. This is an extremely complex
matter; the ambiguity of the metaphors I have invoked to
characterize the theory/research tradition connection is a
symptom of the difficulty of tackling this problem head-on.
But that task cannot be further postponed. I shall begin by
saying what the relation between theories and research tradi-
tions is not. It is not, for instance, one of entailment. Research
traditions do not entail their component theories; nor do those
theories, taken either singly or jointly, entail their parent
research traditions. One might wish it were otherwise, for then
it would be a simple matter to determine mechanically which
theories belonged to any given research tradition, or the
research tradition(s) lurking behind any theory. But to see the
theory/research tradition connection in such formal terms is
completely to misunderstand the differences in kind between
the two. A research tradition, at best, specifies a general
ontology for nature, and a general method for solving natural
problems within a given natural domain. A theory, on the other
hand, articulates a very specific ontology and a number of
specific and testable laws about nature. To be told, as the
Newtonian research tradition in mechanics tells us, that we
should treat all nonrectilinear motions as cases of centrally
directed forces, does not entail any specific theory about how to
explain, say, the motion of a compass needle in the vicinity of a
current-carrying wire. To develop a “‘Newtonian”’ theory for that
particular phenomenon, we must (as Ampére did) go far beyond
the deductive consequences of the Newtonian research tradition.
To be told, as the nineteenth-century ‘“‘mechanical” research
tradition tells us, that heat is simply a form of motion, does not
deductively lead us to Boltzmann’s version of the kinetic
theory of gases orto statistical thermodynamics.
FROM THEORIES TO RESEARCH TRADITIONS 85

Similar considerations apply to the reverse relation between

theories and research traditions. For example, given the theory
of impact as developed by Huygens, we cannot deduce the basic
assumptions of the research tradition within which Huygens
worked. (We may, of course, be able to deduce that Huygens
was working in a research tradition in which collision phenom-
ena constituted an important unsolved problem, for if not, why
should Huygens have bothered working out a theory of colli-
sion?). But it is never possible to deduce the whole of a research
tradition from one, or even all, of the theoriesallied toit.
The reason why entailment will not help here is very simple:
there are a number of mutually inconsistent theories which can
claim allegiance to the same research tradition, and there are a
numberof different research traditions which can, in principle,
provide the presuppositional base for any given theory.
Examples of both phenomena abound: manyscientists in the
Cartesian optical tradition argued that light travelled faster in
optically denser media; other theorists, within the same tradi-
tion, asserted the converse. Staying within the history of optics,
there are numerous examples of competing research traditions
claiming to justify the same theory. For instance, Newton's
theory that light has certain periodic properties was accepted
alike by scientists in the wave and corpuscular traditions. If
entailment were the relation between research traditions and
theories, then it would be impossibie for such situations to arise.
Since the relation we are trying to explore is evidently not one of
entailment, what can wesay positively about it?
There are at least two specific modes by which theories and
research traditions are related: one is historical, the other is
conceptual. It is a matter of historical fact that mostif not all of
the major theories of science have emerged when the scientist
who invented them was working within one or another specific
research tradition. Boyle’s theory of gases developed within the
framework of the mechanical philosophy. Buffon’s embryo-
logical theories were developed as efforts to apply the Newtonian
research tradition to biological phenomena. Hartley’s theories of
sensation were developed within the research tradition of
associationist psychology. Hertz’s electrical theories were linked
in important ways with the Maxwellian research tradition.
86 FROM THEORIES TO RESEARCH TRADITIONS

A specific theory, abstracted from its historical context, may

not give unambiguous clues as to the research tradition (or
traditions) with which it is associated. It is just this fact which
has lead many scientists and philosophers to imagine that
theories are usually appraised and assessed independently of the
research traditions of which they are a part. But we should not
be misled by the fact that a theory, taken abstractly, does not
have its “‘parent’’ research tradition stamped all over it.
Historical research can always (at least in principle) identify the
research tradition(s) with which a particular theory has been
associated. In this sense, the connection between a theory and a
research tradition is as real as any fact of the past, and it is as
important as the most important facts of the past. In order to
see how important these connections are, we need to look at the
ways in which theories and research traditions can interact.
The most important modes of interaction are generally
influences of the research tradition upon its constituent theories.
These influences take a variety of forms:

The problem determining role of research traditions. Even

before specific theories are formulated within a tradition, and
continuously thereafter, a research tradition will often strongly
influence (though it does not fully determine) the range and the
weighting of the empirical problems with which its component
theories must grapple. Equally, research traditions have a
decisive influence on what can count as the range of possible
conceptual problems which the theories in that tradition can
generate. These two processes are important ones and should be
discussed in some detail.
1. Among the other roles of a research tradition, it is
designed to delimit, at least partially and in outline, the domain
of application of its constituent theories. It does this by
indicating that it is appropriate to discuss certain classes of
empirical problems in the given domain, whilst others belong to
foreign domains, or are “‘pseudo-problems” which can be
legitimately ignored. Either the ontology or the methodology of
the research tradition can influence what are to count as
legitimate problemsfor its constituent theories. If, for instance,
the methodology of a research tradition specifies—as it usually
FROM THEORIES TO RESEARCH TRADITIONS 87

will—certain experimental techniques which alone are the

legitimate investigational modes for determining what are the
data to be explained, then it is clear that only “phenomena”
which can be explored by those means can, in principle, count
as legitimate empirical problems for theories within that
tradition. A classic example of this process is offered by nine-
teenth-century phenomenological chemistry. Scientists in this
tradition argued that the only legitimate problems to be solved
by the chemist were those which concerned the observable
reactions of chemcial reagents. Thus, to ask how this acid and
this base react to form this salt is to pose an authentic problem.
But to ask how atoms combine to form diatomic molecules
cannot conceivably count as an empirical problem because the
methodology of the research tradition denies the possibility of
empirical know!edge of entities the size of atoms and molecules.
For other research traditions in nineteenth-century chemistry,
questions about the combining properties of certain entities not
directly observable constituted authentic problems for empirical
research.’* (Contemporary behavioristic psychology and quan-
tum mechanics likewise have methodologies which strongly
preclude from consideration as problems certain ‘“‘phenomena”’
which other research traditions countenance.)
Similarly, the ontology of a research tradition may exclude
certain situations from, or include them within, the appropriate
domain. Thus, the rise of the Cartesian mechanistic research
tradition in the seventeenth century radically transformed the
accepted problem domain for optical theories. It did so by
arguing, or rather by simply postulating, that problems of
perception and vision—problems which had classically been
regarded as legitimate empirical problems for any optical
theory—should be relegated to psychology and to physiology,
fields outside the domain of optics, so that such empirical
problems could be safely ignored by the mechanistic optical
theorist.
A different kind of example is provided by late nineteenth-
century physics, where the subtle fluid tradition (of Faraday,
Maxwell, Hertz, and others) countenanced as legitimate empir-
ical problems queries about the properties of the electromag-
netic aether. Indeed, the classic Michelson-Morley experiments
88 FROM THEORIES TO RESEARCH TRADITIONS

were originally conducted in order to determine the drag

coefficient of bodies moving through such an aether. With the
emergence of special relativity theory, however, a new research
tradition and its related ontology cut out from the domain of
the empirical problem of physics all questions about the
elasticity, density, and velocity of the aether—questions which
had been central empirical problems between 1850 and 1900.*°
These few examples should make it clear that research
traditions can play a decisive role in specifying the sorts of
things that are to count as potentially solvable empirical
problemsfor their constituent theories.
2. Equally important is the way in which a research tradition
can generate conceptual problems for its constituent theories.
Indeed, the bulk of the conceptual problems which any theory
may face will arise because of tensions between that theory and
the research tradition of which it is a part. It often happens
that the detailed articulation of a theory will lead to the
adoption of assumptions which run counter to those allowed by
the research tradtion of that theory. In such a situation, it is
commonplacefor critics of the theory to point to such a tension
as a major conceptual problem for it. When, for instance,
Huygens came to develop a general theory of motion, he found
that the only empirically satisfactory theories were those which
assumed vacua in nature. Unfortunately, Huygens was working
squarely within the Cartesian research tradition, a tradition
which identified space and matter and thus forbade empty
spaces. As Leibniz and others pointed out to Huygens, his
theories were running counter to the research tradition which
they claimed to instantiate. This was an acute conceptual
problem of the first magnitude, as Huygens himself sometimes
acknowledged. Similarly, when Thomas Young—working within
the Newtonian optical research tradition—found himself offer-
ing explanations for optical interference which presupposed a
wave-theoretic interpretation of light, he was chastised for not
recognizing the extent to which his wave theory violated certain
canons of the research tradition to which he seemingly paid
allegiance.?”. Here again, we can see how the dissonance
between a research tradition and its component theories can
generate acute conceptual problems.
FROM THEORIES TO RESEARCH TRADITIONS 89

The constraining role of research traditions. As we have

already said, it is the primary function of a research tradition
to establish a general ontology and methodologyfor tackling all
the problems of a given domain, or set of domains. As such, it
acts negatively as a constraint on the types of theories which
can be developed within the domain. If the ontology of the
research tradition denies the existence of forces acting-at-a-
distance, then it clearly rules out as unacceptable any specific
theory which relies on noncontact action. It was precisely for
this reason that “‘Cartesians’’ such as Huygens and Leibniz
(committed to an ontology of pushes and pulls) found Newton’s
theory of celestial mechanics so otiose. Einstein’s theory of the
equivalence of matter and energy excludes from consideration
any specific theory which postulates the absolute conservation of
mass. The mechanistic tradition in heat theory (with its
corollary that heat can be turned into work) precludes the
development of theories which assume the materiality of heat,
or heat conservation.
There are also many occasions where the methodology of a
research tradition rules out certain sorts of theories. For
instance, any research tradition which has a strongly inductivist
or observationalist methodology will regard as inadmissible
“specific’’ theories postulating entities which cannot be ob-
served. Much of the opposition to subtle fluid theories in the
eighteenth century and to atomic theories in the nineteenth
century was due to the fact that the dominant methodology of
the period denied the epistemic and scientific well-foundedness
of theories which dealt with ‘‘unobservable entities.”’?*
In all these cases, the research tradition within which a
scientist works precludes him from adopting specific theories
which are incompatible with the metaphysics or methodology
of the tradition.
Thus far, we have focussed attention primarily on the
negative manner in which research traditions exclude certain
problems and theories. They also have, however, two very
positive functions.

The heuristic role of research traditions. Precisely because

they postulate certain types of entities and certain methods for
90 FROM THEORIES TO RESEARCH TRADITIONS

investigating the properties of those entities, research traditions

can play a vital heuristic role in the construction of specific
scientific theories. Not, of course, because theories can in any
sense be deduced from research traditions; but rather because
research traditions can provide vital clues for theory construc-
tion. Consider the case of Benjamin Franklin and his efforts to
articulate a theory of static electricity. Franklin was famifiar
with certain phenomena(particularly, electrification by friction,
electroscopes, and the Leyden jar). Working within a research
tradition which postulated the existence of electrical matter,
Franklin needed a theory which could explain how friction
electrifies bodies, how electrical bodies could attract and repel,
howelectricity could be stored in a condensor, and why certain
bodies were conductors and others were insulators. In the
early stages of the development of his theory, Franklin came to
the view that positive electrification consisted in the accumula-
tion within bodies of an excess amount of this electrical fluid,
while negative electrification was caused by a deficiency of this
fluid. If these specific theoretical assumptions are linked
together with the ontology of his research tradition, an ontology
which postulated that electricity was a form of matter and
therefore conserved in the same way that ordinary matter was,
it became natural to assume that electrical charge must be
conserved. This important theoretical insight, subsequently
confirmed in Franklin’s experiments, emerged as an almost
inevitable result of Franklin’s thinking about the relations
between his emerging theory and its parent research tradition.
It did not follow logically from either the early theory itself, nor
from the research tradition. It was the juxtaposition of the two
that madepossible this vital theoretical extension.
A different sort of heuristic role is illustrated by the early
history of thermodynamics. When Sadi Carnot set out to
develop a theory of steam engines, he sought to do so within the
research tradition of the caloric doctrine of heat. Within this
tradition, heat was conceived as a material, conserved substance
capable of moving between the constituent parts of macroscopic
bodies. Carnot, familiar with the work that could be performed
by such simple mechanical systems as a water wheel, tried to
conceive of heat flow on analogy with the fall of water, with the
5a

FROM THEORIES TO RESEARCH TRADITIONS 91

temperature gradient between input and output corresponding

to the top and bottom heights of the waterfall. It is in terms of
this analogy that Carnot develops the “proof” of his theory. It is
clear that, if Carnot had not conceived of heat as a conserved
substance capable of flowing from one point to another without
loss of its quantity, he almost certainly would not have
enunciated his theory. But that way of conceiving heat was a
natural result of the research tradition within which Carnot
worked.
One final example may make the point still clearer. When
Descartes attempted to develop a theory of light and colors, he
had already defined his general research tradition. In brief, it
amounted to the assertion that the only properties which bodies
can have are those of size, shape, position, and motion. The
research tradition did not, indeed could not, specify precisely
what sizes, shapes, positions, and motions particular bodies
could exhibit. But it did make it clear that any specific physical
theory, in optics or elsewhere, would have to deal exclusively
with these four parameters. As a result, Descartes knew—when
he set out to explain optical refraction, the colors of the
rainbow, and the path of light through lens and prisms—that
his optical theories would have to be constructed along such
lines. So, he sought to explain colors in terms of the shape and
rotational velocity of certain particles; he explained refraction in
terms of differential velocities of such particles in different
media. Moreover, since his general research tradition madeit
clear that particles of light are exactly like other material
bodies, he recognized that he could apply general mechanical
theorems (such as the laws of impact and the principle of
conservation of motion) to a theoretical analysis of light. Again,
none of his theories followed logically from his research
tradition; but, in the ways indicated, that research tradition
directed the construction of Cartesian theories in a number of
subtle and important ways.
In all the cases mentioned thus far, the research tradition
functions heuristically to suggest an initial theory for some
domain. A second important heuristic role for the research
tradition, as Lakatos has pointed out, arises when one of its
constituent theories requires modification (because of its lack of
92 FROM THEORIES TO RESEARCH TRADITIONS

problem-solving ability). Any sound research tradition will

contain significant guidelines about how its theories can be
modified and transformed, so as to improve its problem-solving
capacity.
For instance, when early versions of the kinetic theory of
gases were confronted by someserious predictive failures, there
was enormous “‘flexibility’’ within the research tradition which
pointed the way towards natural modifications that might be
made. If more degrees of freedom were needed to accommodate
seeming energy losses, kineticists could introduct molecular
spin or alter their assumptions about molecularelasticities. If
gases did not condense in accordance with theoretical predic-
tions, the addition of weak intermolecular attractions could do
the job. All these, and manysimilar “‘gambits’’ emerge quite
plausibly from regarding matter as possessing a molecular and
mechanical composition, ?°

The justificatory role of research traditions. It is one of the

important functions of research traditions to rationalize or to
Justify theories. Specific theories make many assumptions about
nature, assumptions which are generally not justified either
within the theoryitself or by the data which confirm the theory.
These are usually assumptions about basic causal processes and
entities, whose existence and operation the specific theories
take ‘‘as given.’’ When, for instance, Sadi Carnot developed his
theory of the steam engine, his working out of that theory
presupposed that no heat was lost in performing the work of
driving a piston. (That assumption later turned out to be
unacceptable, of course; but it is an assumption which is
absolutely crucial to Carnot’s “proof” of his theory.) Carnot
offered no rationale for that assumption, and, quite rightly, felt
no need to; the caloricist research tradition, within which he was
working, laid it down as a primary postulate that heat was
always conserved. Carnot was thus able to presuppose, in
developing his theory, certain things about nature which his
theory could notitself establish, not even in principle.
A century earlier, when Stephen Hales had developed his
theory about the nature of ‘air’ (i.e., gases), he was able to
take it almost for granted that gases were composed of mutually
repellent particles, and was able to use repulsion to explain such
FROM THEORIES TO RESEARCH TRADITIONS 93

phenomenaas elasticity and gaseous mixing. Had Hales been

working in research traditions other than the Newtonian one,
such an assumption would have been unthinkable, or at least in
need of elaborate justification. (At a minimum, his theory
would have had to addressitself to justifying that assumption.)
But, as a Newtonian, Hales could assume, alfnost without
argument, that it was appropriate and legitimate to conceive of
gases as swarms of mutually repellent particles. By sanctioning
certain assumptions in advance, the research tradition thus
frees the scientist working within it from having to justify all of
his assumptions, and gives him the time to pursue specific
problems of interest. Although critics outside the research
tradition may fault a scientist for constructing theories based on
such assumptions, the scientist knows that his primary audience
—fellow researchers within the same tradition—will not find
his working assumptions problematic.
Research traditions thus identify for the scientist working
within them three classes of assumptions: those which are
unproblematic, because justified by the research tradition;
those which are forbidden by the research tradition; and, of
course, those which, while not forbidden by the research
tradition, definitely require a rationale within the theory (for the
research tradition itself provides no rationale for them). Among
scientists working within any one research tradition, there will
be a broad consensus about where any given statement falls as
between those three pigeonholes.
Summing up the discussion thus far, we have seen that such
research traditions can justify many of the assertions which their
theories make; they can serve to stamp certain theories as
inadmissible because they are incompatible with the research
tradition; they can influence the recognition and weighting of
empirical and conceptual problems for their component
theories, and they can provide heuristic guidelines for the
generation or modification of specific theories.

The Separability of Theories from Research Traditions

Up to now, I have stressed that virtually all theoretical
activity takes place within the context of a research tradition,
that such traditions constrain, inspire, and serve to justify the
 94 FROM THEORIES TO RESEARCH TRADITIONS

theories which are subsumed under them. Without wishing to

negate any of that, it is equally important to recognize that
there are circumstances in which theories can break away from
the research traditions which initially inspired or justified them.
Galileo’s theory of fall, for instance, has (since the 1650s) been
treated separately from the Galilean research tradition; similar
things could be said about Pasteur’s theory of disease,
Maxwell’s theory of electromagnetism, Lavoisier’s theory of
oxidation, and Planck’s theory of black-body radiation, to name
only a few cases. Indeed, it is just the eventual possibility of
separating a theory from a given research tradition which gives
the misleading impression that theories exist independently of,
and owe nothing to, research traditions.
This process of theory separation is a fascinating one and
deserves to be studied in some detail. I shall limit my remarks
here to pointing out that the separation of a theory from its
parent research tradition generally takes place only when that
theory can be taken over, either intact or by small-scale modi-
fications in it, by an alternative research tradition. Theories
rarely can exist on their own, and even whenthey doitis only
for short periods of time. The reasons for this are clear.
Theories are never self-authenticating; they invariably make
assumptions about the world for which they provide no
rationale. Since it is one of the functions of a research tradition
to provide just such a rationale for a theory, it is normally the
case that a theory is separated from one research tradition only if
it can be absorbed (i.e., justified) within another and more
successful research tradition.
The doctrines of early thermodynamics, to which we have
referred earlier, are a case in point. Originally developed within
a caloricist research tradition (based on substantial, nonkinetic
theories of heat) by Carnot and Clapeyron, the theory of
thermodynamics proved an embarrassment during the late
1840s and 1850s, by which time the research tradition that
inspired it had been largely discredited. There was wide
agreement that the theory of thermodynamics was worth
preserving, but not (manyfelt) at the price of subscribing to the
research tradition which had generated it. At the same time, the
kinetic, anti-caloric research tradition was making great strides
FROM THEORIES TO RESEARCH TRADITIONS 95

forward in other domains, but was thought to be weak insofar

as it had been unable to equal, within the area of thermo-
dynamics, the successes which its competitor, the caloric
tradition, had achieved. It was Rudolf Clausius, writing in the
1850s, who was able to show that the theory of thermodynamics
could be developed and rationalized within the kinetic tradition,
independently of the caloricist assumption of the conservation
of heat. Clausius thereby showed that the theory of thermo-
dynamics was not inexorably wedded to the caloricist research
tradition and could be absorbed by the kineticist tradition. In
one fell swoop, Clausius thus managed to strengthen the case
for both thermodynamicsand for kineticism, by removing what
had been a serious conceptual problem for them both. In like
manner, Newton (as a vehement opponent of the Cartesian
research tradition) was able to show that his own research
tradition could absorb the Huygensian theory of impact—a
theory which had originally been developed squarely within the
Cartesian tradition.
The multitude of cases one could cite of this process ought
not lead us to underestimate its difficulty. Precisely because a
research tradition plays an important justificatory role for its
constituent theories, any alternative research tradition which is
to play the same role mustbe sufficiently rich conceptually, and
its partisans sufficiently imaginative, to allow it to justify and
rationalize theories which prima facie are more naturally related
to very different metaphysical and methodological traditions. (1
shall have more to say later about this process of ‘‘theory
appropriation,” for it is one of the most important ways in
which mew research traditions establish their scientific cre-
dentials.)

The Evolution of Research Traditions

Research traditions, as we have seen, are historical creatures.

They are created and articulated within a particular intellectual
milieu, they aid in the generation of specific theories—andlike
all other historical institutions—they wax and wane. Just as
surely as research traditions are born and thrive, so they die,
and cease to be seriously regarded as instruments for furthering
 96 FROM THEORIES TO RESEARCH TRADITIONS

the progress of science. I shall consider below how research

traditions are displaced by other ones, for the aetiology of
research tradition “decay” and ‘‘putrefaction”’ is crucial to the
processes which must be understood. For now, however, I want
to talk about the ways in which important and substantive
changes can occur within an ongoing research tradition. These
changestake two distinct forms.
The most obvious way in which a research tradition changes
is by a modification of some of its subordinate, specific theories.
Research traditions are continuously undergoing changesof this
type. Researchers in the tradition often discover that thereis,
within the framework of the tradition, a more effective theory
for dealing with some of the phenomena in the domain than
they had realized previously. Slight alterations in previous
theories, modifications of boundary conditions, revisions of
constants of proportionality, minor refinements of terminology,
expansion ofthe classificatory network of a theory to encompass
newly discovered processes or entities; these are but a few of the
many ways in which the scientist may seek to improve on the
problem-solving success of any of the theories within the
research tradition. Whenever he discovers a theory which is a
significant improvement on its predecessor he drops the latter
immediately. Precisely because a scientist’s cognitive loyalties
are based primarily in the research tradition rather than in any
of its specific theories, he generally has no rational vested
interest in hanging onto those individual theories. (It is for just
this reason that most individual theories have very short
life-spans—in many cases amounting to no more than a few
months or even weeks.) Because theories change so rapidly, the
history of any flourishing research tradition will exhibit a long
succession of specific theories.
But there is another important way in which research
traditions evolve; this second class of changes involves, not the
specific theories within the research tradition, but a change of
some of its most basic core elements. \ must discuss this type of
transformation in some detail, since there are many philos-
ophers who have denied that research traditions are capable of
any significant internal modification. Both Kuhn and Lakatos,
for instance, usually suggest that entities such as research
FROM THEORIES TO RESEARCH TRADITIONS 97

traditions have a rigid and unchanging set of doctrines which

identify and define them. Any change in those doctrines, it is
suggested, produces a different research tradition. Since,
Lakatos argues, we define a research tradition or research
programmein terms of its central doctrines (doctrines which
Lakatos argues we make true by fiat or by convention), any
change in those central tenets is de facto the abandonment of
the research tradition which was defined as the set of those
tenets.°° As tempting as this approach is (for, if true, it would
make the process of identifying research traditions relatively
straightforward), I shail be arguing that we mustrejectit, forit
can only obfuscate our effort to get some understanding of the
historical processes of science.
If one looks at the great research traditions in the history of
scientific thought—Aristotelianism, Cartesianism, Darwinism,
Newtonianism, Stahlian chemistry, mechanistic biology, or
Freudian psychology, to name only a few—one can see
immediately that there is scarcely any interesting set of
doctrines which characterizes any one of these research tradi-
tions throughout the whole of its history. Certain Aristotelians,
at times, abandoned the Aristotelian doctrine that motion in a
void is impossible. Certain Cartesians, at times, repudiated the
Cartesian identification of matter and extension. Certain New-
tonians, at times, abandoned the Newtonian demand that all
matter has inertial mass. But need it follow that these seeming
“renegades”? were no longer working within the research tradi-
tion to which they earnestly claimed to subscribe? Does Thomas
Aquinas cease to be an Aristotelian because he rejects portions
of Aristotle’s analysis of motion? Does Huygens become a
non-Cartesian because he admits the possibility of void spaces?
Certain advantages will accrue if we can plausibly answer these
questions negatively. To show how that is possible is the task
before us.
A research tradition, we have said, is a set of assumptions:
assumptions about the basic kinds of entities in the world,
assumptions about how those entities interact, assumptions
about the proper methods to use for constructing and testing
theories about those entities. In the course of their develop-
ment, research traditions and the theories they sponsor
98 FROM THEORIES TO RESEARCH TRADITIONS

encounter a number of problems; anomalies are discovered;

basic conceptual problemsarise. In some cases, proponents of a
research tradition will find themselves unable, by modifying
specific theories within the tradition, to eliminate these anom-
alous and conceptual problems. In such circumstances, it is
common for partisans of a research tradition to explore what
sorts of (minimal) changes can be made in the deep-level
methodology or ontology of that research tradition to eliminate
the anomalies and conceptual problems confronting its constit-
uent theories. Sometimes, scientists will find that there is no
amount of tinkering with one or another assumption of the
research tradition which will eliminate its anomalies and
conceptual problems. This becomes strong grounds for aban-
doning the research tradition (provided there is some alternative
in sight). But, perhaps more often, scientists find that by
introducing one or two modifications in the core assumptions of
the research tradition, they can both solve the outstanding
anomalies and conceptual problems and preserve the bulk of
the assumptions of the research tradition in tact.
In the latter case, it is positively misleading to speak of the
creation of a ‘“‘new’’ research tradition, for such language
conceals from us the crucial conceptual ancestry and similarity
which such cases exhibit. We should speak, rather, of a natural
evolution in the research tradition; an evolution which repre-
sents a change, to be sure, but a change that is far from
repudiation of a former research tradition and the creation of a
new one.*?
There is much continuity in an evolving research tradition.
From one stage to the next, there is a preservation of most of
the crucial assumptions of the research traditions. Most of the
problem-solving techniques and archetypes will be preserved
through the evolution. The relative importance of the empirical
problems which the research tradition addresses will remain
approximately the same. But the emphasis here must be on
relative continuity between successive stages in the evolutionary
process. If a research tradition has undergone numerous
evolutions in the course of time, there will probably be many
discrepancies between the methodology and ontology of its
FROM THEORIES TO RESEARCH TRADITIONS 99

earliest and its latest formulation. Thus, the Cartesianism of a

Bernoulli, writing a century after Descartes’ death, is very
different from the Cartesianism of the master. The Newtonian
research tradition in Michael Faraday’s handsis a far cry from
that of Newton’s first followers. But a finer-grained analysis of
the historical evolution of these research traditions will show
that there was a continuous intellectual descent from Descartes
to Bernoulli, and from Newton to Faraday, and that the
Cartesian and Newtonian research traditions, as different as
their end-points may look from their beginnings, exhibited
enormous continuity in the character of their transformations.’
But such an approach leaves itself open to the obvious
challenge: if a research tradition can undergo certain deep-level
transformations and still remain in some sense the ‘‘same”
tradition, how do we distinguish change within a research
tradition from the replacement of one research tradition by
another?
A partial answer to the question comes from recognizing that
at any given time certain elements of a research tradition are
more central to, more entrenched within, the research tradition
than other elements. It is these more central elements which are
taken, at that time, to be most characteristic of the research
tradition. To abandon them is indeed to move outside the
research tradition, whereas the less central tenets can be
modified without repudiation of the research tradition. Like
Lakatos, then, I want to suggest that certain elements of a
research tradition are sacrosanct, and thus cannot berejected
without repudiation of the tradition itself. But unlike Lakatos,
I wantto insist that the ser of elements falling in this (unreject-
able) class changes through time. What was taken to charac-
terize the unrejectable core of the Newtonian tradition in
eighteenth-century mechanics (e.g., absolute space and time)
was no longer regarded as such by mid-nineteenth-century
Newtonians. What constituted the essence of the Marxist
research tradition in the late nineteenth century is substantially
different from the “‘essence’’ of Marxism a half century later.
Lakatos and Kuhn were right in thinking that a research
programme or paradigm always has certain nonrejectable
 100 FROM THEORIES TO RESEARCH TRADITIONS
elements associated with it; but they were mistaken in failing
to
see that the elements constituting this class can shift through
time. By relativizing the “essence” of a research tradition with
respect to time, we can, I believe, come much closer to
capturing the way in which scientists and historians of science
actually utilize the concept of a tradition.
Of course, this still leaves unanswered howit is that scientists
decide at any given time which elements of a maxi-theory or
research tradition are to be treated as “unrejectable” (a
problem likewise unanswered by Kuhn and Lakatos). I cannot
give a fully satisfactory answer to the question, but some
hunches are probably worth exploring. Both Kuhn and Lakatos
seem to believe that the decision about which elements of
a
maxi-theory fall into this privileged class is arbitrary and not
governed by rational considerations: on their account, it simply
“happens.’’*? I am unable to give a full Specification of all
the
factors which influence the selection of the core of a research
tradition, but there are clearly dimensions of the choice which
are rational. For instance, one of the major factors influencing
the entrenchmentof any element of a research tradition
is its
conceptual well-foundedness. The core assumptions of any
given research tradition are continuously undergoing conceptu
al
scrutiny. Some of those assumptionswill, at any given time, be
found to be strong, and unproblematic. Others will be regarded
as less clear, less well-founded. As new arguments emerge
which
buttress, or cast doubt on, different elements of the
research
tradition, the relative degree of entrenchment of the different
components will shift. During the evolution of any active
research tradition, scientists learn more about the
conceptual
dependence and autonomy ofits various elements; when it can
be shown that certain elements, previously regarded
as essential
to the whole enterprise, can be jettisoned without compromising
the problem-solving success of the tradition itself, these
ele-
ments cease to be a part of the “‘unrejectable core” of
the
research tradition. (For instance, after Mach and Frege
argued
that none of the other elements of the Newtonian tradition
required the absoluteness of space and time, these
notions
moved perceptibly towards the periphery of the Newtoni
an
research tradition.)
FROM THEORIES TO RESEARCH TRADITIONS 101

Research Traditions and Changes in Worldviews

Wehave stressed, both here and in the previous chapter, how

research traditions and theories can encounter serious cognitive
difficulties if they are incompatible with certain broader systems
of belief within a given culture. Such incompatibilities consti-
tute conceptual problems which may seriously challenge the
acceptability of the theory. But it may equally well happen that
a highly successful research tradition will lead to the abandon-
ment of that worldview which is incompatible with it, and to
the elaboration of a new worldview compatible with the re-
search tradition. Indeed, it is in precisely this manner that
many radically new scientific systems eventually come to be
“canonized” as part of our collective “common sense.” In the
seventeenth and eighteenth centuries, for instance, the new
research traditions of Descartes and Newton went violently
counter to many of the most cherished beliefs of the age on such
questions as “‘man’s place in Nature,”’ the history and extent of
the cosmos, and, more generally, the nature of physical
processes. Everyone at the time acknowledged the existence of
these conceptual problems. They were eventually resolved, not
by modifying the offending research traditions to bring them in
line with more traditional world views, but rather by forging
a new world view which could be reconciled with the scientific
research traditions. A similar process of re-adjustment occurred
in response to the Darwinian and Marxist research traditions in
the late nineteenth century; in each case, the core, ‘‘nonscien-
tific’ beliefs of reflective people were eventually modified to
bring them in line with highly successful scientific systems.
But it would be a mistake to assume that worldviews always
crumble in the face of new scientific research traditions which
challenge them. To the contrary, they often exhibit a remark-
able resilience which belies the (positivistic) tendency to dismiss
them as mere fluff. The history of science, both recent and
distant, is replete with cases where worldviews have not evap-
orated in the face of scientific theories which challenged
them. In our own time, for instance, neither quantum mechan-
ics nor behavioristic psychology have shifted most people’s
beliefs about the world or themselves. Contrary to quantum
102 FROM THEORIES TO RESEARCH TRADITIONS

mechanics, most people still conceive of the world as being

populated by substantial objects, with fixed and precise prop-
erties; contrary to behaviorism, most people still find it helpful
to talk about the inner, mental states of themselves and others.
Confronted with such examples, one might claim that these
research traditions are still new and that older world views
predominate only because the newerinsights have not yet pene-
trated the general consciousness. Such a claim may prove to be
correct, but before we accept it uncritically, there are certain
more striking historical cases that need to be aired. Ever since
the seventeenth century, the dominant research traditions
within the physical sciences have presupposed that all physical
changes are subject to invariable natural laws (either statistical
or nonstatistical). Given certain initial conditions, certain
consequences would inevitably ensue. Strictly speaking, this
claim should be as true of man and other animals as it is of
stars, planets, and molecules. Yet in our own time, as much as
in the seventeenth century, very few people are prepared to
abandon the conviction that human beings (and some of the
higher animals) have a degree of undeterminiation in their
actions and their thoughts. Virtually all of our social institu-
tions, most of our social and political theory, and the bulk of
our moral philosophy is still based on a worldview seemingly
incompatible with a law-governed universe. Despite repeated
efforts in the last three centuries to explain away this conceptual
problem, it is fair to say that this is one important area where
the traditional worldview has made very few concessions to the
“broader implications” of some highly successful scientific
traditions.**
It has jong been fashionable to imagine that the worldview or
“Zeitgeist’”’ of any epoch always plays a purely conservative role,
suppressing intellectual innovation and encouraging the reten-
tion of the scientific status quo. Exponents of scientific progress
have frequently bemoaned the use of ‘‘worldview”’ considera-
tions which invariably stifle the emergence of new scientific
ideas. E. G. Boring spoke for many scientists and philosophers
when he insisted that: “Inevitably by definition the Zeitgeist
favors conventionality . . . [and] works against originality.’’*S
FROM THEORIES TO RESEARCH TRADITIONS 103

This position is bad philosophy and false history. It is philo-

sophically weak insofar as it ignores the fact that there is no
reason, in principle, why an entrenched worldview could not
provide a more convincing rationale for an innovative theoret-
ical development than for a traditional theory. Boring’s claim
that a Zeitgeist automatically favors traditional theories is thus
without cognitive foundation. The view is equally misleading
historically. It is well known, for instance, that the Zeitgeist of
late seventeenth-century England did much to hasten the
replacement of the older mechanical philosophy by the newer
science of Newton, precisely because Newton’s research tradi-
tion could be morereadily justified within that framework than
the mechanistic science of Descartes could. More recently, the
emergence of ‘“‘new’’ quantum mechanics in the late 1920s
found a quick and ready reception among the many intellec-
tuals who were already convinced that the rigid causal cate-
gories of classical science were unreliable.

The Integration of Research Traditions

Up to now I have spoken as if research traditions were
invariably in competition with one another, suggesting moreover
that the resolution of such a conflict comes when one alone
among the competing traditions dominates and when its
competitors are effectively vanquished. This is often the case.
But it would be a serious error to assumethat a scientist cannot
consistently work in more than one research tradition. If these
research traditions are inconsistent in their fundamentals, then
the scientist who accepts them both raises serious doubts about
his capacity for clear thinking. But there are times when two or
more research traditions, far from mutually undermining one
another, can be amalgamated, producing a synthesis which is
progressive with respect to both the formerresearch traditions.
It is the dynamics of such situations which I want to discuss
briefly here.
There are basically two ways in which different research
traditions can be integrated. In some cases, one research
tradition can be grafted onto another, without any serious
104 FROM THEORIES TO RESEARCH TRADITIONS

modification in the presuppositions of either. Thus, in eigh-

teenth-century natural philosophy, many scientists were simul-
taneously Newtonians and subtle fluid theorists. Their adher-
ence to the research tradition of subtie fluids (which was as
much Cartesian as it was Newtonian) led them to postulate
imperceptible aetherial fluids in order to explain the phenom-~
ena of electricity, magnetism, heat, perception, and a range of
other empirical problems. Their Newtonianism, on the other
hand, led them to assumethat the constituent particles of such
fluids interacted (not by contact, as the Cartesians tried to
suggest) but rather by means of strong forces of attraction and
repulsion, acting-at-a-distance across empty space. The fusion
of these two research traditions was to constitute itself a major
research tradition, one which Schofield has labelled ‘‘material-
ism.’°¢ While undermining the presuppositions of neither of
its predecessors, the amalgamation suggested important new
lines of research, and put scientists in a position to deal with
empirical and conceptual problems which neither of the
ancestor traditions alone could resolve satisfactorily.
In other cases, however, the amalgamation of two or more
research traditions requires the repudiation of some of the
fundamental elements of each of the traditions being combined.
In these cases, the new researchtradition, if successful, requires
the abandonmentof its predecessors. (It is, incidentally, in just
this way that most so-called scientific revolutions take place; not
by the articulation of a research tradition whose ingredients are
revolutionary and new, but rather by the development of a
research tradition whose novelty consists in the way in which
old ingredients are combined.) There are many examples of this
process in the history of any discipline, scientific or otherwise.
To consider some scientific cases first, eighteenth- and nine-
teenth-century natural philosophy is replete with such integra-
tions. Roger Boscovich, for instance, set out deliberately to
develop a new “‘system of nature,” by picking and choosing
from among the assumptions of two incompatible research
traditions, Newtonianism and Leibnizianism, Maupertuis at-
tempted something similar. The work of their contemporary,
Daniel Bernoulli, illustrates an analogous attempt to forge a
FROM THEORIES TO RESEARCH TRADITIONS 105

compromise between the research traditions of Cartesian and

Newtonian physics. In the eighteenth and nineteenth centuries,
geological followers of Hutton were hammering out a new
tradition which drew on elements of caloricist heat theories and
Vulcanist geology. These research traditions could not be
preserved intact and, as a result, the Huttonians had to forge
what was regarded as a ‘revolutionary’ research tradition
which incorporated elements of traditions which had been
previously incompatible. Within economics, Kari Marx drew on
elements from the idealism of Hegel, the materialism of
Feuerbach, and the “capitalism” of Adam Smith and his
English followers.

‘‘Nonstandard”’ Research Traditions

It would be dishonest to leave the topic of research traditions
without adding a caveat, although how important it is remains
to be seen. We have thus far characterized research traditions
as rather ambitious and grandiose entities, replete with ontol-
ogies and methodologies. There is no doubt in my mind that
many of the best known research traditions in science possess
both these characteristics. But there also seem to be traditions
and schools in science which, although lacking one or the
other (or in some cases both), have nonetheless had a genuine
intellectual coherence about them. For instance, the tradition of
psychometrics in the early twentieth century seems to have been
held together by little more than the conviction that mental
phenomena could be mathematically represented. Equally, the
tradition of rational mechanics in the eighteenth century seems
to have cut horizontally across almost every conceivable meta-
physical and methodological tradition and to have drawn
together a group of thinkers committed simply to the mathe-
matical analysis of motion and rest. The important tradition of
“analytic physics” in early nineteenth century France (including
Biot, Fourier, Ampére, and Poisson) seems to have had no
common ontology, although its partisans doubtlessly shared a
common methodology. In our own time, cybernetics and
information theory seem to be “‘schools’’ without well-defined
106 FROM THEORIES TO RESEARCH TRADITIONS

ontologies. Whether, on further investigation, it will turn out

that these ‘‘nonstandard”’ research traditions do have ontolog-
ical and methodological elements or whether, failing that, they
will behave differently from “‘richer’’ research traditions are
still unanswered questions. Much research is still needed on
these units that are too narrow to be full-blown research
traditions but too global to be mere theories.

The Evaluation of Research Traditions

Our focus thus far has been on the temporal dynamics of

research traditions. We have learned something about how such
traditions evolve, how they interact with their constituent
theories and with wider elements of the worldview and the
problem situation.
However, I have said nothing yet about how, if at all, it is
possible for scientists to make sensible choices between alterna-
tive research traditions, nor about how single tradition can be
appraised relative to its acceptability. This is a crucial issue, for
until and unless we can articulate workable criteria for choice
between the larger units I am calling research traditions, then
we have neither a theory of scientific rationality, nor a theory of
progressive, cognitive growth.
In the next few pages, I shall be defining somecriteria for the
evaluation of research traditions, and discussing some of the
different contexts in which cognitive evaluations can be made.

Adequacy and Progress

Even though research traditions in themselves entail no
observable consequences, there are several different ways in
which they can berationally evaluated and thus compared. Two
chief modes of appraisal, however, are the most common and
the most decisive. One of these modesis synchronic, the other is
diachronic and developmental.
Wemay, to begin with, ask about the (momentary) adequacy
of a research tradition. We are essentially asking here how
effective the /atest theories within the research tradition are at
solving problems. This, in turn, requires us to determine the
problem-solving effectiveness of those theories which presently
FROM THEORIES TO RESEARCH TRADITIONS 107
constitute the research tradition (ignoring their predecessors).
Since we already discussed how to evaluate the problem-solving
effectiveness of individual theories,?? we need only combine
those appraisals to find the adequacy of the broader research
tradition.
Alternatively, we may ask about the progressiveness of a
research tradition. Here our chief concern is to determine
whether the research tradition has, in the course of time,
increased or decreased the problem-solving effectiveness of its
components, and thus its own (momentary) adequacy. This
matter is, of course, unavoidably temporal; without a knowl-
edge of the history of the research tradition, we can say nothing
whatever about its progressiveness. Under this general rubric,
there are two subordinate measures which are particularly
important:
1. the general progress of a research tradition—this is
determined by comparing the adequacy of the sets of theories
which constitute the oldest and those which constitute the most
recent versions of the researchtradition;
2. the rate of progress of a research tradition—here, the
changes in the momentary adequacy of the research tradition
during any specified time span are identified.
It is important to note that the general progress and the rate
of progress of a research tradition may be widely at odds. For
instance, a research tradition may show a high degree of
general progress, and yet show low rate of progress, especially
in its recent past. Alternatively, a research tradition may have
a high rate of progress during its recent past while exhibiting
limited general progress.
Likewise, and even more importantly, the appraisals of a
research tradition based uponits progressiveness (either general
or time-dependent) may be very different from those based on
its momentary adequacy. One can conceive of cases, for
example, where the adequacyof a researchtradition is relatively
high and yet it shows no general progress or even a negative
rate of progress. (In fact, many actual research traditions have
this character.) Alternatively, there are cases(e. g., behavioristic
psychology and early quantum theory) where the general
progress and the rate of progress of a research tradition are

re
108 FROM THEORIES TO RESEARCH TRADITIONS

high, but where the momentary adequacyofthe tradition isstill

quite low.
Needless to say, the appraisals will not always point in
contrary directions, but the very fact that they can (and
sometimes have) emphasizes the need to attend very carefully to
the various contexts in which cognitive appraisals of research
traditions are made. It is that issue which must occupyus next.

The Modalities of Appraisal: Acceptance and Pursuit

Almost all the standard writings on scientific appraisal,
whether we look to philosophical or historical discussions of
science, have two commonfeatures: they assume that there is
only one cognitively legitimate context in which theories can be
appraised; and they assume that this context has to do with
determinations of the empirical well-foundedness of scientific
theories. Both these assumptions probably need to be aban-
doned: the first because it is false, the second because it is
too limited.
1 shall be arguing that a careful examination of scientific
practice reveals that there are generally two quite different
contexts within which theories and research traditions are
evaluated.’ I shall suggest that, within each of these contexts of
inquiry, very different sorts of questions are raised about the
cognitive credentials of a theory, and that much scientific
activity which appears irrational—if we insist on a uni-con-
textual analysis—can be perceived as highly rational if we allow
for the divergent goals of the following two contexts:

The context of acceptance. Beginning with the more famil-

iar of the two, it is clear that scientists often choose to accept
one among a group of competing theories and research tra-
ditions, i.e., to treat it as if it were true. Particularly in cases
where certain experiments or practical actions are contem-
plated, this is the operative modality. When, for instance, a
research immunologist must prescribe medication for a volun-
teer in an experiment, when a physicist decides what measuring
instrument to use for studying a problem, when a chemist is
seeking to synthesize a compound with certain properties; in all
al

FROM THEORIES TO RESEARCH TRADITIONS 109

these cases, the scientist must commit himself, however tenta-

tively, to the acceptance of one group of theories and research
traditions and to the rejection of others.
How can he make a coherent decision? There are a wide
range of possible answers here: inductivists will say ‘““choose the
theory with the highest degree of confirmation”; or “choose the
theory with the highest utility’; falsificationists—if they give any
advice at ail—will say ‘‘choose the theory with the greatest
degree of falsifiability.”’ Still others, such as Kuhn, would insist
that no rational choice can be made.*° I have already indicated
why none of these answersare satisfactory. My own reply to the
question, of course, would be, “choose the theory (or research
tradition) with the highest problem-solving adequacy."
On this view, the rationale for accepting or rejecting any
theory is thus fundamentally based on the idea of problem-
solving progress. If one research tradition has solved more
important problems than its rivals, then accepting that tradition
is rational precisely to the degree that we are aiming to
“‘progress,”’ i.e., to maximize the scope of solved problems. In
other words, the choice of one tradition over its rivals is a
progressive (and thus a rational) choice precisely to the extent
that the chosen tradition is a better problem solver than its
rivals.
This way of appraising research traditions has three distinct
advantages over previous modeis of evaluation: (1) it is workable:
unlike both inductivist and falsificationist models, the basic
evaluation measures seem (at least in principle) to pose fewer
difficulties; (2) it simultaneously offers an account of rational
acceptance and of scientific progress which shows the two to be
linked together in ways not explained by previous models; and
(3) it comes closer to being widely applicable to the actual
history of science than alternative models have been.

The context of pursuit. Even if we had an adequate account

of theory choice within the context of acceptance, however, we
would still be very far from possessing a full account of rational
appraisal. The reason for this is that there are many important
situations where scientists evaluate competing theories by
110 FROM THEORIES TO RESEARCH TRADITIONS

criteria which have nothing directly to do with the acceptability

or ‘“‘warranted assertibility” of the theories in question.
The actual occurrence of such situations has often been
observed. Paul Feyerabend, in particular, has identified many
historical cases where scientists have investigated and pursued
theories or research traditions which were patently less accept-
able, fess worthy of belief, than their rivals. Indeed, the
emergence of virtually every new research tradition occurs under
Just such circumstances. Whether we look to Copernicanism,
the early stages of the mechanical philosophy, the atomic
theory in the first half of the nineteenth century, early psycho-
analytic theory, the preliminary efforts at the quantum mechan-
ical approach to molecular structure, we see the same pattern:
scientists often begin to pursue and to explore a new research
tradition long before its problem-solving success (orits inductive
support, or its degree of falsifiability, or its novel predictions)
qualifies it to be accepted over its older, more successful rivals.
Another side to the same coin is the historical fact that a
Scientist can often be working alternately in two different, and
even mutually inconsistent, research traditions. Particularly
during periods of “scientific revolution,” it is commonly the
case that a scientist will spend part of his time working on the
dominant research tradition and a part of his time working on
one or more of its less successful, less fully developed rivals. If
we take the view that it is rational to work with and explore
only the theories one accepts (and its corollary that one ought
not accept or believe mutually inconsistent theories) then there
can be no way of making sense of this common phenomenon.
Hence neither the use of mutually inconsistent theories nor
the investigation of less successful theories—both well-attested
historical phenomena—can be explained if we insist that the
context of acceptance exhaustsscientific rationality. Confronted
by such cases, we would have to conclude, with Feyerabend
and Kuhn,“ that the history of science is largely irrational. But
if, on the other hand, we realize that scientists can have good
reasons for working on theories that they would not accept, then
this frequent phenomenon may be more comprehensible.
To see what could count as “good reasons’ here, we must
return to some earlier discussions. It has often been suggested
FROM THEORIES TO RESEARCH TRADITIONS 111

in this essay that the solution of a maximum number of

empirical problems, and the generation of a minimum number
of conceptual problems and anomalies is the central aim of
science. We have seen that such a view entails that we should
accept at any time those theories or research traditions which
have shown themselves to be the most successful problem
solvers. But need the acceptance of a given research tradition
preclude us from exploring and investigating alternatives which
are inconsistent with it? Under certain circumstances, the
answer to this question is decidedly negative. To see why, we
need only consider the following general kind of case: suppose
we have two competing research traditions, RT and RT’;
suppose further that the momentary adequacy of RT is much
higher than that of RT’, but that the rate of progress of RT’ is
greater than the related value for R7. So far as acceptance is
concerned, RT is clearly the only acceptable one of the pair. We
may nonetheless decide to work on, further articulate, and
explore the problem-solving merits of RT’, precisely on the
grounds that it has recently shown itself to be capable of
generating new solutions to problems at an impressive rate. This
is particularly appropriate if RT' is a relatively new research
tradition. It is common knowledge that most new research
traditions bring new analytic and conceptual techniques to bear
on the solution of problems. These new techniques constitute
(in the cliché) ‘fresh approaches” which, particularly over the
short run, are likely to pay problem-solving dividends. To
accept a budding research tradition merely because it has had a
high rate of progress would, of course, be a mistake; but it
would be equally mistaken to refuse to pursue it if it has
exhibited a capacity to solve some problems (empirical or
conceptual) which its older, and generally more acceptable,
rivals have failed to solve.
Putting the point generally, we can say that it is always
rational to pursue any research tradition which has a higher rate
of progress than its rivals (even if the former has a lower
problem-solving effectiveness). Our specific motives for pur-
suing such a research tradition could be one of many: we
might have a hunchthat, with further development, RT’ could
become more successful than RT; we might have grave doubts
112 FROM THEORIES TO RESEARCH TRADITIONS

about RT’ ever becoming generally successful, but feel that

some of its more progressive elements could eventually be
incorporated within RT. Whatever the vagaries of the individual
case, if our general aim is increasing the number of problems
we can solve, we cannot be accused of inconsistency or
irrationality if we pursue (without accepting) some highly
progressive research tradition, regardless of its momentary
inadequacy (in the sense defined above).
In arguing that the rationality of pursuit is based on relative
progress rather than overall success, I am making explicit what
has been implicitly described in scientific usage as “‘promise”’ or
“fecundity.”” There are numerous cases in the history of science
which illustrate the role which an appraisal of promise or
progressiveness can have in earning respectability for a research
tradition.
The Galilean research tradition, for instance, could not in its
early years begin to stack up against its primary competitor,
Aristotelianism. Aristotle’s research tradition could solve a great
many more important empirical problems than Galileo’s.
Equally, for all the conceptual difficulties of Aristotelianism, it
really posed fewer crucial conceptual problems than Galileo's
early brand of physical Copernicanism—a fact that tends to be
lost sight of in the general euphoria about the scientific
revolution. But what Galilean astronomy and physics did have
going for it was its impressive ability to explain successfully
some well-known phenomena which constituted empirical
anomalies for the cosmological tradition of Aristotle and
Ptolemy. Galileo could explain, for example, why heavier bodies
fell no faster than lighter ones. He could explain the irregular-
ities on the surface of the moon, the moons of Jupiter, the
phases of Venus, and the spots on the sun. Although Aristote-
lian scientists ultimately were able to find solutions for these
phenomena (after Galileo drew their attention to them), the
explanations proferred by them smacked of the artificial and
the contrived. Galileo was taken so seriously by later scientists
of the seventeenth century, not because his system as a whole
could explain more than its medieval and renaissance predeces-
sors (for it palpably could not), but rather because it showed
promise by being able, in a short span of time, to offer solutions
FROM THEORIES TO RESEARCH TRADITIONS 113

to problems which constituted anomalies for the other research

traditions in thefield.
Similarly, Daltonian atomism generated so muchinterest in
the early years of the nineteenth century largely becauseof its
scientific promise, rather than its concrete achievements. At
Dalton’s time, the dominant chemical research tradition was
concerned with elective affinities. Eschewing any attempt to
theorize about the microconstituents of matter, elective affinity
chemists sought to explain chemical change in terms of the
differential tendencies of certain chemical elements to unite
with others. That chemical tradition had been enormously
successful in correlating and predicting howdifferent chemical
substances combine. Dalton’s early atomic doctrine could
claim nothing like the overall problem-solving success of elective
affinity chemistry (this is hardly surprising, for the affinity
tradition was a century old by the time of Dalton’s New
System of Chemical Philosophy), still worse, Dalton’s system was
confronted by numerous serious anomalies.‘’ What Dalton was
able to do, however, was to predict—as no other chemical
system had done before—that chemical substances would
combine in certain definite ratios and multiples thereof, no
matter how much of the various reagents was present. This
phenomenon, summarized by what we now call the laws of
definite and multiple proportions, created an immediate stir
throughout European science in the decade after Dalton’s
atomic program was promulgated. Although most scientists
refused to accept the Daltonian approach, many nonetheless
were prepared to take it seriously, claiming that the serendipity
of the Daltonian system made it at least sufficiently promising
to be worthy of further development and refinement.
Whether the approach taken here to the problem of ‘‘rational
pursuit” will eventually prevail is doubtful, for we have only
begun to explore some of the complex problems in this area;
what I would claim is that the linkage between progress and
pursuit outlined above offers us a healthy middle ground
between (on the one side) the insistence of Kuhn and the
inductivists that the pursuit of alternatives to the dominant
paradigm is never rational (except in times of crisis) and the
anarchistic claim of Feyerabend and Lakatos that the pursuit
 114 FROM THEORIES TO RESEARCH TRADITIONS

of any research tradition—no matter how regressive it is—can

always be rational.

Adhocness and the Evolution of Research Traditions

No discussion of the various appraisal vectors utilized in
science would be complete without including the notion of
adhocness (an issue often discussed under the rubric ‘‘inde-
pendent testability’). At least since the seventeenth century,
but particularly in our own era, ad hoc stratagems and
hypotheses have received much attention from scientists and
philosophers alike.‘? The determination that a theory or
theoretical modification is ad hoc gives us grounds, on the usual
account, for dismissing it as illegitimate and unscientific. If we
are to accept the claims sometimes made by such philosophers
as Popper, Griinbaum, and Lakatos,** it is irrational or
unscientific ever to accept a theory which is ad hoc. What does
such adhocness amount to, and why, if at all, is it such a
liability for theories which exhibit it?
The issue of adhocness arises most often in connection with
the evolution of theories and the manner in which they handle
anomalies. We are usually asked to imagine a situation in which
some theory, 7,, encounters a refuting instance, A. In response
to A, some modification is introduced into T,, producing
T,. The conventional wisdom insists that the later theory 7, is
ad hoc if: T, can solve A, and the other known problems T,
could solve, but 7, has no non-trivial, testable implications
other than those of 7, and A. Putting it in the languageof this
monograph, a theory 7, ts ad hoc if it can solve only those
empirical problems solved by its predecessor‘7,, and those
which constitute refuting instances for 7,, but no further
problems.
There are several difficulties with this approach to adhocness.
In the first place, we generally have no way of knowing at any
given time whether a new theory 7, will at some later point be
able to solve new problems. To make such a judgment sensibly
would require a super-human clairvoyance about what empir-
ical problems and what auxiliary theories (which, when con-
joined with the theory, might lead to the solution of new

1
T°"

FROM THEORIES TO RESEARCH TRADITIONS 115

problems) are going to emerge in the future. However, taking a

cue from Adolf Griinbaum, we can relativize the above
definition to situations of belief and say that a theory T, is ad
hocif it is believed to solve only those empirical problems which
were solved by, or refuting instances for, 7,.**
But serious difficulties still remain. As Duhem taught us,
individual theories in isolation generally solve no problems. It
is, rather, complexes of theories which are involved in problem
solution.‘ Hence, we must modify the traditional characteriza-
tion once again, yielding a definition such as the following: a
theory is ad hoc if it is believed to figure essentially in the
solution of all and only those empirical problems which were
solved by, or refuting instances for, an earlier theory.
Clumsyasit is, this characterization of adhocness seems to do
justice to some of the most sophisticated accounts of adhocness
developed in the last decade. Assuming that adhocness is
understood in this way, we are entitled to ask: what is
objectionable about it? If some theory T, has solved more
empirical problems than its predecessor—even just one more—
then JT, is clearly preferable to 7,, and, ceteris paribus,
represents cognitive progress with respect to 7,. However, we
can go further than this to claim that the resort to ad hoc
stratagems, as defined immediately above, is perfectly consis-
tent with the general aim of increasing our problem-solving
capacities. Ad hoc modifications, by their very definition, are
empirically progressive.
This result should not be surprising. Indeed, much of what
we mean by such clichés as “learning from experience’ and
“the self-correction of science” is represented by situations in
which, when a theory encounters an anomaly, we alter the
theory so as to transform the anomaly into a solved problem.
While it would be a nice bonus if every theory modification
could immediately solve some new problemsas well as someold,
unsolved ones, to insist on that requirement (as, for instance,
Popper, Lakatos, and Zahar have) is to repudiate the doctrine
that theories which solve more problems about the world are
preferable to those which solve fewer.
In urging that adhocness (so defined) is a cognitive virtue
rather than a vice, I am clearly not implying that ad hoc
116 FROM THEORIES TO RESEARCH TRADITIONS

theories are invariably better than non-ad hoc ones. My claim,

rather, is that an ad hoctheory is preferable to its non-ad hoc
predecessor (which was confronted with known anomalies). To
believe otherwise is to deny a vital aspect of the problem-
solving characterof scientific inquiry. *°
But it might be argued that I have missed the point of the
critics of adhocness. They might say, ““Yes, of course, T, is
better than its refuted predecessor 7,; but the relevant com-
parison is between the ad hoc T, and someother theory 7, which
is not ad hoc butstill solves as many problemsas T,.’’ Einstein’s
special theory of relativity might exemplify 7, while the
Lorentz-modified aether theory was 7,.‘’ The obvious reply
to such criticism is to ask why the admittedly ad hoc character
of the Lorentz contraction constitutes a decisive handicap
against it in comparingit with special relativity. If the empirical
problem-solving capacities of the two theories are, so far as we
can tell, equivalent, then they are (empirically) on a par;
defenders of the view that the adhocness of T, makes it
distinctly inferior to 7,, must spell out why, in such cases, the
comparable problem-solving abilities and equivalent degrees of
empirical support can be thrown to the winds simply by stipu-
lating that ad hoctheories are intrinsically otiose.
What seemsto lie behind many discussions of adhocness is a
conviction—often present but rarely defended—that there is
something suspicious about any change in a theory which is
motivated by the desire to remove an anomaly. The presump-
tion is that we cannot really trust such cosmetic surgery
because, once we know what the anomaly is, it is little more
than child’s play to produce some face-saving change in the
theory which turns the anomaly into a positive instance. I doubt
that where “‘real”’ science is concerned, this task is such an easy
one. We must remember that, as adhocness has been defined,
any ad hoc change must increase rather than decrease the
problem-solving capacity of the theory in question. Most of the
obvious and trivial ways of eliminating anomalies—e.g., arbi-
trarily restricting the boundary conditions, eliminating those
postulates of the theory which entailed the anomaly (assuming
they could be localized!), redefining terms or correspondence
rules—would generally result in decreasing the problem-solving
effectiveness of a theory. Hence, such maneuvres—which we
tae

FROM THEORIES TO RESEARCH TRADITIONS 117

might well wish to criticize**—-do not qualify as ad hoc. The

detractors of adhocness have yet to show that the emendation
of a theory to preserve its problem-solving capacity and to save
it from an anomaly requires any less theoretical imagination or
serendipity than the construction of a new theory from scratch.
To the extent that these same detractors set an epistemic
premium on theories which work the first time around, without
any juggling or ad hoc adjustments, we are entitled to ask for
the rationale for such a preference.
To this philosophical worry, we should briefly add a historical
one. Most of the major theories in science—including Newton-
ian mechanics, Darwinian evolution, Maxwellian electromag-
netic theory and Daltonian atomism—were all ad hoc in the
sense defined above. Those modern philosophers andscientists
who wish to make adhocness a debilitating handicap for any
theory which exhibits it must explain why the most ‘‘successful’’
theories of the past were also highly ad hoc.
There is a grain of truth, however, in the worries of many
scientists and philosophers about adhocness. To locate it,
we must direct our attention away from the empirical level
and towards the conceptual one. In many of the classic epi-
sodes where charges of adhocness have been made (e.g., Ptol-
emic astronomy, Cartesian physics, phrenology, and the Lorentz-
FitzGerald contraction), the cognitive features of the situa-
tion can be characterized as follows: a theory, 7,, has en-
countered an anomaly, A. 7, has been replaced by T,,
which solves A, and T,’s previous solved problems, but is not
known to be able to solve any other empirical problems. At the
same time, J, has generated more acute conceptual problems
than T, exhibited (perhaps by making assumptions contrary to
the ontology of 7,’s research tradition, or by running counter
to other acceptable theories). In such cases, the empirical gains
made by T, may be more than offset by its conceptual losses,
resulting in a diminished overall problem-solving effectiveness.
Here we would be warranted in refusing to accept JT, in
preference to T,. Viewed in this light, the only legitimately
pejorative sense of ‘‘adhocness”’ reduces to a situation in which
a theory's overall problem-solving effectiveness decreases, by
virtue of its increasing conceptual difficulties. This sort of
adhocness is common in science, and is a frequently cited
118 FROM THEORIES TO RESEARCH TRADITIONS

ground for rejecting theories. But it is important to stress that

the concept of adhocnessitself, thus understood, adds nothing
whatever to our analytic machinery for appraising theories,
since it is itself just a special case of conceptual problem
generation.
I am, by no means, the first to suggest a conceptual
interpretation of adhocness; Lakatos, Zahar, and Schaffner
have developed similar interpretations recently.*? In all their
discussions, however, conceptual adhocness remains but one of
many species of adhocness, rather than the only legitimate
sense. Still worse, none of these writers has indicated how
conceptual adhocness is to be assessed, nor even what it
amounts to. Equally, all these writers leave us in the dark about
howseriously, if at all, it should count against a theory if it is ad
hoc. The seeming virtue of the approach taken here is that it
separates spurious senses of the ad hoc from legitimate ones,
and it gives us machinery for assessing the degrees of cognitive
threat posed by adhocness to the theories which exhibit it.

Anomalies Revisited
Chapter one contained the paradoxical claim that the refut-
ing instances of a theory are not necessarily anomalous
problems, along with a promissory note to clarify that claim
once the machinery was available to do so. The evaluational
methods outlined here allow us to return to this issue. I said
before that a problem was only anomalous (i.e., cognitively
threatening) for some theory, 7, if that problem was unsolved
by T but solved by one of its competitors. Clearly, some re-
futing instances will satisfy this definition, but many will not.
It is often the case that some prediction of a theory fails to
square with the data, but no other available theory can solve
the data either. In the latter situation, why should the data not
count as a threatening anomaly for T?
In brief, the answer is this: Whenever a theory encounters a
refuting instance, it is possible to modify the interpretative rules
associated with the theory so as to disarm the ‘‘refuting” data.
If, for instance, we have a theory, 7, that ‘all planets move in
ellipses’ and then discover a satellite of the sun, S, which
FROM THEORIES TO RESEARCH TRADITIONS 119

movesin a circle, we can always modify the interpretative rules

governing the term “‘planet’’ so as to exclude S, thus preserving
our theory intact and eliminating any appearance of refutation.
If there is no other theory extant which can explain the motion
of S, the exclusion of § from 7's domain is perfectly reasonable
and progressive—for we lose none of our previously won
problem-solving successes by legislating S out of the relevant
domain. By contrast, if some alternative to T can solve S, then
T's legislation of S outside the domain is a regressive step, open
to rational criticism precisely because T's abandonmentof S as
a legitimate problem entails that we sacrifice some of our
demonstrated problem-solving capacity.
What this amounts to is that the modification of a theory
arbitrarily in order to eliminate a refuting instance is open to
criticism only if such a move would lead to a diminished
problem-solving efficiency. That can generally be shown to
happen only if the refuting instance is solved by some theory in
the domain. Hence, a refuting instance only counts as a serious
anomaly when it has been solved by some theory or other.

Summary: A General Characterization

of Scientific Change
Drawing together the various strands of argument developed
in this chapter, we can conclude that:
1. The adequacy or effectiveness of individual theories is a
function of how manysignificant empirical problems they solve,
and how many important anomalies and conceptual problems
they generate. The acceptability of such theories is related both
to their effectiveness and to the acceptability of their related
research tradition.
2. The acceptability of a research tradition is determined by
the problem-solving effectiveness of its latest theories.
3. The promise, or rational pursuitability, of a research
tradition is determined by the progress (or rate of progress) it
has exhibited.
4. Acceptance, rejection, pursuit, and non-pursuit constitute
the major cognitive stances which scientists can legitimately
take towards research traditions (and their constituent theories).
aa cw oa wlll

120 FROM THEORIES TO RESEARCH TRADITIONS

Determinations of truth and falsity are irrelevant to the

acceptability or the pursuitability of theories and research
traditions.
5. All evaluations of research traditions and theories must be
made within a comparative context. What matters is not, in
some absolute sense, how effective or progressive a tradition or
theory is, but, rather, how its effectiveness or progressiveness
compares with its competitors.
We can now move on to discuss the implications of this
model of scientific progress for an understanding of some of the
central historical and philosophical questions about the cogni-
tive growth of science,
 Chapter Four
Progress and Revolution
The revolutionary can only regard
his revolution as a progress in so
far as he is also an historian. COLLINGWOOD (1956), p. 326

The analytic machinery developed in the preceding chapters

raises numerous significant questions about the historical
evolution and cognitive status of the sciences. The function of
this chapter is to examine the ways in which a problem-solving
approachtoscientific inquiry can throw new light on a number
of central historical and philosophical problems about science,
and to show howscientific progress, scientific rationality, and
the nature of scientific revolutions can all be profitably
discussed in terms of the problem-oriented model outlined
above.

Progress and Scientific Rationality

One of the thorniest questions of twentieth-century philos-
ophy concerns the nature of rationality. Some philosophers
suggest that rationality consists in acting to maximize one’s
personal utilities; others suggest that rationality consists in
believing in, and acting on, only those propositions which we
have good grounds for believing to be true (or at least to be

121

EEE
 122 PROGRESS AND REVOLUTION

more likely than not); others hint that rationality is a function

of cost-benefit analysis; still others claim that rationality
amounts to no more than putting forward statements which can
be refuted. A great deal has been written on these, as well as
other, notions of rational belief and rational action. But,
~ ignoring the fact that none of these explications of rationality
has been shown to be free of logical and philosophical
difficulties, it has never been shown that any of them are rich
enough to fit our intuitions about the rationality inherent in
much of the history of scientific thought. To the contrary, it is
relatively easy to show that there are numerous cases in the
history of science—cases in which almost everyone would agree
intuitively that rational analysis was occurring—which run
counter to each of the models of rationality mentioned above.
The theory of research traditions and progress which is
outlined in the preceding chapters constitutes a significant
improvement on the theories of rationality now in common
parlance among philosophers (if by improvement we mean
providing a more accurate explication of the cognitive factors
present in actual casesof scientific decision making).
As the previous discussion has shown, there are important
historical cases where: (1) scientists have invoked what I have
called ‘‘non-refuting’’ anomalous problems as major objections
to theories; (2) scientists have devoted themselves to the
clarification of concepts and to the reduction of other sorts of
conceptual problems; (3) scientists have pursued and investi-
gated promising (i.e., highly progressive) theories, even when
those theories were less adequate than rivals; (4) scientists have
utilized metaphysical and methodological arguments against
and in favor of scientific theories and research traditions; (5)
scientists have accepted theories confronted by numerous
anomalies; (6) the importance of a problem, and even its
status as a problem, has exhibited wild fluctuations; (7)
scientists have accepted theories which did rot solve all the
empirical problems of their predecessors.
Although cases exhibiting (1) through (7) have not always
been rational and cognitively well-founded, the model I have
developed allows us to specify circumstances under which any
one of these ploys would be rationally justified. The same claim

ol
PROGRESS AND REVOLUTION 123

cannot be made, I believe, for any other extant model of

scientific growth and progress.
But it might well be argued against this model, that it is
purely descriptive, with no rational or normative force; that it
offers, at best, a taxonomy for identifying certain variables in
scientific controversies, but that it does not show why any of
those variables should play a role in the appraisal of scientific
theories. It could be pointed out that nowhere do I show how
the capacity of a theory to solve problems bears on the truth or
the probability of the theory in question. It could be shown that
nowhere do I establish that problem-solving ability provides
grounds for rational belief. ©
Some of these criticisms are entirely correct; I do not even
believe, let alone seek to prove, that problem-solving ability has
any direct connection with truth or probabilities. But I deny
that the circumvention of such epistemic questions deprives the
model of normative and explanatory force; equally, I deny that
a model of rational theory appraisal must issue in judgments of
truth, falsity, probability, confirmation, or corroboration.
To make these denials plausible, | must tackle directly, if
briefly, the question (so far skirted in this essay) of the
connections between rationality and truth.
At its core, rationality—whether we are speaking about
rational action or rational belief—consists in doing (or believ-
ing) things because we have good reasons for doing so. That
does not solve the problem, of course, but only restates it. The
restatement, however, is a useful one, for it makes clear that if
we are going to determine whether a given action or belief is (or
was) rational, we must ask whether there are (or were) sound
reasons for it. It is vital to be clear at the outset that many
things that would count as good reasons outside science cannot
constitute good reasons within science. To take a trivial
example, I might have a good reason for saying that “2 + 2 =
5,” if | know that someone will punish me severely if I refuse to
say it. Similarly, I might have a good personal reason for trying
to resurrect the Ptolemaic theory (if, for instance, I am poor
and a research institute of the Vatican begins awarding grants
for such projects). But what can count as a good personal
reason for doing something does not necessarily count as a good
 124 PROGRESS AND REVOLUTION

scientific reason for doing it. So what does count as a good

reason in science? To answer that question, we must consider
the aims of science. For if we can show that doing one
particular action, rather than another, would be conducive to
achieving the aims of the scientific enterprise, then we would
have shown the rationality of doing the one and the irrationality
of doing the other within the framework ofscience.
I have tried to argue that the single most general cognitive
aim of science is problem solving. I have claimed that the
maximization of the empirical problems we can explain and the
minimization of the anomalous and conceptual problems we
generate in the process are the raison d‘étre of science as a
cognitive activity. I have claimed that any research tradition
which can exemplify this process through time is a progressive
one. It follows from this that the chief way of being scientifically
reasonable or rational is to do whatever we can to maximize the
progress of scientific research traditions. Equally, this line of
attack suggests that rationality consists in accepting the best
available research traditions. There are, however, other com-
ponents of rationality which follow from this way of looking at
the matter. The model I have outlined suggests, for instance,
that scientific debate is rational so long as it involves a discus-
sion of the empirical and conceptual problems which theories
and research traditions generate. Thus, contrary to common
belief, it can be rational to raise philosophical and religious
objections against a particular theory or research tradition, if
the latter runs counter to a well-established part of our general
Weltbild—even if that Weltbild is not ‘“‘scientific’’ (in the usual
sense of the word). The model suggests that the rational
appraisal of a theory or research tradition necessarily involves
an analysis of the empirical problems which it solves, and the
conceptual and anomalous problems which it generates. The
model, finally, insists that any appraisal of the rationality
of accepting a particular theory or research tradition is trebly
relative: it is relative to its contemporaneous competitors,
it is relative to prevailing doctrines of theory assessment, and
it is relative to the previous theories within the research
tradition.

'
PROGRESS AND REVOLUTION 125

In arguing for this approach to science, I am deliberately

driving a wedge between several issues that have hitherto been
closely intertwined. Specifically, it has normally been held that
any assessment of either rationality or scientific progress is
inevitably bound up with the question of the truth of scientific
theories. Rationality, it is usually argued, amounts to accepting
those statements about the world which we have good reason for
believing to be true. Progress, in its turn, is usually seen as a
successive attainmentof the truth by a process of approximation
andself-correction. [ want to turn the usual view on its head by
making rationality parasitic upon progressiveness. To make
rational choices is, on this view, to make choices which are
progressive (i.e., which increase the problem-solving effective-
ness of the theories we accept). By thus linking rationality to
progressiveness, I am suggesting that we can have a theory of
rationality without presupposing anything about the veracity or
verisimilitude of the theories we judge to be rational or
irrational.
If this effort to talk about the cognitive status of scientific
knowledge without relating it to the truth claims of such
knowledge seems bizarre, one need only consider the circum-
stances which have motivated this way of tackling the problem.
Philosophers and scientists since the time of Parmenides and
Plato have been seeking to justify science as a truth-seeking
enterprise. Without exception, these efforts have foundered
because no one has been able to demonstrate that a system like
science, with the methods it has at its disposal, can be
guaranteed to reach the ‘‘Truth,” either in the short or in the
long run. If rationality consists in believing only what we can
reasonably presume to be true, and if we define “‘truth” in its
classical, non-pragmatic sense, then science is (and will forever
remain) irrational. Realizing this dilemma, some philosophers
(notably Peirce, Popper, and Reichenbach) have sought to link
scientific rationality and truth in a different way, by suggesting
that although our present theories are neither true nor prob-
able, they are closer approximations to the truth than their
predecessors. Such an approach offers few consolations, how-
ever, since no one has been able even to say what it would mean
er “4

126 PROGRESS AND REVOLUTION

to be “closer to the truth,” let alone to offer criteria for

determining how we could assess such proximity.' Hence, if
scientific progress consists in a series of theories which represent
an ever closer approximation to the truth, then science cannot
be shownto be progressive. If, on the other hand, we accept the
proposal developed in this essay and take the view that science
is an inquiry system for the solution of problems, if we take the
view that scientific progress consists in the solution of an
increasing number of important problems, if we accept the
proposal! that rationality consists in making choices which will
maximize the progress of science, then we may be able to show
whether, and if so to what extent, science in general, and the
specific sciences in particular, constitute a rational and progres-
Sive system.
The price we have to pay for this approach may be regarded
by some people as too high, for it entails that we may find
ourselves endorsing theories as progressive and rational which
turn out, ultimately, to be false (assuming, of course, that we
could ever definitely establish that any theory was false). But
there is no reason for dismay at this conclusion. Most of the
past theories of science are already suspected of being false;
there is presumably every reason to anticipate that current
theories of science will suffer a similar fate. But the presumptive
falsity of scientific theories and research traditions does not
render science either irrational or non-progressive.
The model under discussion here offers a means of showing
how, even granting the fact that every theory of science may
well be false, science may nonetheless turn out to be a worthy
and intellectually significant enterprise. There will be those who
will charge that such an approach its patently instrumentalist
and that it entails that science is a hollow set of symbols and
sounds, with no bearing on ‘‘the real world” or on the ‘‘truth.”
Such an interpretation is very wide of the mark. There is
nothing in this model which rules out the possibility that, for all
we know, scientific theories are true; equally, it does not
preclude the possibility that scientific knowledge through time
has moved closer and closer to the truth. Indeed, there is
nothing 1 have said which would rule out a full-bodied,
“realistic” interpretation of the scientific enterprise. But what I

a
 127
PROGRESS AND REVOLUTION
am suggesting is that we apparently do not have any way of
is
knowing for sure (or even with some confidence) that science
truth. Such
true, or probable, or that it is getting closer to the
know
aims ate utopian, in the literal sense that we can never
being achieved . To set them up as goals for
whether they are
inquiry may be noble and edifying to those who
scientific
can
delight in the frustration of aspiring to that which they
very helpful
never (know themselves to) attain; but they are not
if our object is to explain how scientific theories are (or should
be) evaluated.’
The workability of the problem-solving model is its greatest
virtue. In principle, we can determine whether a given theory
can
does or does not solve a particular problem. In principle, we
determine whether our theories now solve more importa nt
problems than they did a generati on or a century ago. If we
in
have had to weaken our notions of rationality and progress
weare at least now in a position to be
order to achieve this end,
sive—a
able to decide whether science is rational and progres
denied to us if we retain the classical
crucial necessity
connections between progress, rationality, and truth.
How precisely do we go about making this decision? Inevi-
the
tably, it involves the assessment of specific cases drawn from
rational and
history of science; whether science as a whole is
set of
progressive depends, of course, upon whether the
traditio ns: has
individual choices of theories and research
exhibited progress and rationality. Thus, we may ask whether
on
the reaction of the scientific community to Einstein’s paper
a progress ive modifica tion in the
the photoelectric effect led to
the
theories of physics. At another level, we may ask whether
Newtoni an research traditio n over the
overall triumph of the
ian research traditio ns in the eightee nth
Cartesian and Leibniz
ive. In answeri ng such question s, we must
century was progress
scten-
attend very carefully to the parameters of contemporary
and controve rsy, for it is precisel y therein that the
tific debate
l and
historian can find out what the acknowledged empirica
it is there that he can get a
conceptual problems were;
of those
reasonably clear sense of the weight or importance
case (and not
problems. By a subtle analysis of the actual
the histori an—or
4 so-called rational reconstruction of it),
 128 PROGRESS AND REVOLUTION
the contemporary scientist—can usually determine the degree
to which competing research traditions, or competing theories
within the same research tradition, were progressive
in their
modifications.
Whatis crucial here is that we must cast our nets of apprais
al
sufficiently widely that we include al/ the cognitively
relevant
factors which were actually present in the historical
situation.
We must not assume a priori, as some historians
of science
have, that the only important parameters were experim
ental or
other obviously “‘scientific’’ ones. Because theories
and research
traditions have to be accommodated within a broader
network
of beliefs and preconceptions, any accurate apprais
al of an
episode must attend carefully to the philosophical,
theological
and other intellectual currents which were brought
to bear on
the case at hand. The fact that a twentieth-century
scientist
might not recognize the cogency of an objection to
a theory on
philosophical or religious grounds manifestly does
not mean
that an understanding of the rationality of earlier science
can be
acquired by ignoring such factors. If a culture at a
particular
time has a strongly entrenched set of religious or
philosophical
doctrines which thinkers in that culture believe to
be crucial to
an understanding of nature, then it is pertectly rational
to
appraise new scientific theories or research traditions
in light of
their ability to be accommodated within that prior
system of
beliefs and presuppositions.
There are doubtless those who would argue that such
an
approach sorelativizes our standards of rationality that
it will
justify any set of beliefs. If such a criticism were true,
then there
would be grave problems with the notion of rational
ity being
defended here. But that is far from the case. To
suggest that
“anything goes,” that any combination of beliefs would
emerge
as rational and progressive on this model, is profoun
dly to
misunderstand the high standards of rational behavio
r which it
requires. Nor does the model involve the complete
surrender of
our standards of rationality to the exigencies of
earlier times
and places.
This point is worth discussing at some length, for
it bears
crucially on many of the central dilemmas in the histori
ography
and sociology of science. Many philosophers have
sought to set
 PROGRESS AND REVOLUTION 129

up standards of rationality or progress which are valid for all

times and places. They see the task of the historian-philosopher
of science as that of evaluating historical episodes entirely with
respect to modern theories of rational acceptance and appraisal.
In some cases, proponents of such an approach have gone so
far as to claim that all the actual standards of rational appraisal
have remained constant through time. Israel Scheffler, for
instance, summarizesthis view as follows:
Underlying historical changes of theory, . . . [is] a constancy of logic and
method, which unifies each scientific age with that which preceded it... .
Such constancy comprises not merely the canons of forma! deduction, but also
those criteria by which hypotheses are confronted with the test of experience
and subjected to comparative evaluation.’
Weneed waste little time on this approach. Virtually all the
scholarly literature on the history of methodology shows unam-
biguously that such componentsof rational appraisal as criteria
of explanation, views about scientific testing, beliefs about the
methods of inductive inference and the like have undergone
enormous transformations.
A second group, represented by Popper and Lakatos, ac-
knowledges that scientific standards of rationality have evolved,
but insists that we should evaluate historical episodes using our
standards and simply ignore the appraisals made by the relevant
scientists about the rationality of what they were doing. On this
approach, we pay no heed to whether an experiment was viewed
as reliable, whether a theory was regarded as intelligible, or
whether an argument was perceived as cogent.* What matters,
rather, is whether, by our lights, a particular theory was
well-founded.
Understandably, historians have been dismayed by both these
approaches. What is the point, they ask, of analyzing the
rationality of past science unless we take into account the
views of historical agents about the rationality of what they were
doing? Unencumbered by modern notions of rationality, scien-
tists of the past had to make decisions about the acceptability of
contemporary theories by their criteria rather than by ours. We
may have the hubris to imagine that our theories of rationality
are better than theirs (and they may well be), but how does it
help historical understanding to evaluate the cogency of past

|
 130 PROGRESS AND REVOLUTION
theories utilizing evaluative measures which we know were not
operative (not even in an approximative form) in the case
at hand?
But the historian is confronted by the other horn of the
dilemma. If he simply takes at face value every actual appraisal
by past scientists of the rationality of a belief, he will never be in
any position to judge whether such appraisals were, even by the
appropriate standardsof the time, well-founded. Obviously, the
fact that some historical agent says, “theory A is better than
theory B,’’ does not necessarily makeit so. If the historian is to
i explain why certain theories triumphed and others perished,
then he must (unless he takes the view that theory choice is
always irrational) be able to show that some theories—by the
best available rational standards of the time—were superior to
others.
Hence, the central problem seems to be this: how can we,
with the philosophers, continue to talk normatively about the
rationality (and irrationality) of theory choices in the past, while
at the same time avoid the grafting of anachronistic criteria of
rationality onto those episodes?
The model I have outlined resolves part of that difficulty by
exploiting the insights of our own time about the general nature
of rationality, while making allowances for the fact that many of
the specific parameters which constitute rationality are time-
and culture-dependent. It transcends the particularities of the
past by insisting that for all times and for all cultures, provided
those cultures have a tradition of critical discussion (without
which no culture can lay claim to rationality), rationality consists
in accepting those research traditions which are most effective
problem solvers. It insists that for scientists in any culture to
espouse a research tradition or a theory which is less adequate
than other ones available within that culture is to behave
irrationally. In these important respects, the model argues that
there are certain very general characteristics of a theory of
rationality which are trans-temporal and trans-cultural, which
are as applicable to pre-Socratic thought, or the developmentof
ideas in the Middle Ages, as they are to the more recent
history of science. On the other hand, the model also insists that
what is specifically rational in the past is partly a function of

_
 {31
ON
PROGRESS AND REVOLUTI
of things which count as
time and place and context. The kinds
the sorts of objections that are
empirical problems,
of intelligibility,
recognized as conceptual problems, the criteria
the importance or
the standards for experimental control,
of the methodo-
weight assigned to problems, are all a function
unity of thinkers.
logical-normative beliefs of a particular comm
advantage of
The model under discussion here possesses the
ical norms of a
allowing us to integrate the specific histor
more gener al, time- independent fea-
previous epoch and the
tures of rational decision making.*
nal choice is to
Toignore the time-specific parameters of ratio
geous position of
put the historian or philosopher in the outra
achievements in the
indicting as irrational some of the major
irrational when he
history of ideas. Aristotle was not being
science of physics
claimed, in the fourth century B.c., that the
by, metaphysics—
should be subordinate to, and legitimated
and places, might well
even if that same doctrine, at other times
nas or Robert
be characterized as irrational. Thomas Aqui
d or preju diced when they
Grosseteste were not merely stupi
must be compa tible with
espoused the belief that science
religious beliefs.
disagree with
We in the twentieth century may vehemently
them obscu ranti st and harmf ul to the
such views, thinking
ve we are
development of science. And in so disagreeing, I belie
that theories and
right. One of the things that time has shown is
s) flourish best
research traditions sometimes (though not alway
ogical and meta-
when they are not subordinated to the theol
ific community.
physical doctrines dominant outside the scient
we have come to
But it is with the advantage of hindsight that
ience of the last
that conclusion. In the absence of the exper
d to assu me that it
three centuries, it would be palpably absur
theol ogy and metap hysics
wasirrational to imagine that science,
view that scien ce is quast-
could be mutually supportive. The
is itself a resea rch tradit ion, one
independent of such disciplines
is a kind of resea rch tradi tion that
of relatively recent origin. It
e of progress.
has, in its way, generated a considerable degre
ratio nal in the twent ieth century to
That is why it may be
that a belief is ratio nal in the prese nt age,
accept it. But the fact
matte r, does not neces saril y entail that it
or in any age for that

ee
132 PROGRESS AND REVOLUTION

was rational at other times and places. Quite the reverse is more
often the case.
It should be clear by now that, in arguing that the cultural
exigencies and pressures exerted on science must be taken into
account, I am neither abandoning the possibility of rational
appraisal nor am | insisting that nonscientific factors are
present in every case of scientific choice. I am simply suggesting
that we need a broadened notion of rationality which will show
how the “‘intrusion’’ of seemingly ‘‘nonscientific’’ factors into
scientific decision making is, or can be, an entirely rational
process. Far from viewing the introduction of philosophical,
religious and moral issues into science as the triumph of
prejudice, superstition and irrationality, this model claims that
the presence of such elements may be entirely rational; further,
that the suppression of such elements mayitself be irrational
and prejudicial.
Of course, whether it is rational to use theological, moral, or
philosophical arguments for (or against) a new scientific theory
or research tradition is a contingent matter which depends on
how rational and progressive are the research traditions which
provide such arguments. To argue against modern theories of
chemical combustion on the grounds that such theories are
incompatible with the myth of Vulcan is patently absurd, for
the Greek myths have scarcely established themselves as a body
of rational and progressive dogma. To argue against Marxist
economics on the grounds that it is contrary to Christian
morality is, again, to use a singularly non-progressive tradition
as a tool for criticizing a relatively progressive ‘‘scientific’’
tradition. The rationality or irrationality of any episode where
“nonscientific,’’ but intellectual, factors play a role must be
assessed on a case-by-case basis. But the guiding principles here
should be these: (1) in the case of competing scientific research
traditions, if one of those traditions is compatible with the most
progressive ‘‘worldview” available, and the other is not, then
there are strong grounds for preferring the former; (2) if both
traditions can be legitimated with reference to the same world-
view, then the rational decision between them may be made on
entirely ‘“‘scientific’” grounds; (3) if neither tradition is com-
patible with a progressive worldview, their proponents should

Se
PROGRESS AND REVOLUTION 133

either articulate a new, progressive worldview which does justify

them, or develop a new research tradition which can be made
compatible with the most progressive extant worldview. ,

Scientific Revolutions
For well over a century, it has been commonplace to focus on
“scientific revolutions” as one of the core concepts for historical
narration and exegesis. Within the last two decades, the idea of
a revolution has become canonized in Thomas Kuhn's classic
The Structure of Scientific Revolutions. Although far from his
intent (since Kuhn was primarily concerned to draw attention to
non-revolutionary, “normal science’’), his book has led many
scientists, philosophers and historians alike to compartment-
alize the evolution of science into widely separated periods of
revolutionary activity, and to imagine that the scientific revolu-
tion (with its attendant “change of paradigm’’) is the basic
category for discussing the evolution of science.
Although scientific revolutions are undoubtedly important
historical phenomena, they possess neither the importance nor
the cognitive character often associated with them. They have
assumed this privileged position largely because their structure
has been mis-described in ways that make them seem radically
unlike science in its usual state; the exaggeration of the
difference between ‘‘normal”’ and “revolutionary science” in its
turn has led some writers to lay heavier stress on ‘‘periods of
revolutionary activity’ than they probably deserve.
Consider, for instance, Kuhn’s account of scientific revolu-
tions, For him, a revolution is marked by the emergence of a
new theoretical ‘‘paradigm’’ which, in a short span of time,
discredits the older paradigm and draws the virtually unani-
mous adherence of every member of the relevant scientific
community. Revolutions, on his view, are preceded by short
periods of frenetic theoretical activity during which many
alternative viewpoints vie for the allegiance of the scientific
community. Elements of the previous paradigm which were
previously sacrosanct suddenly become objects of lively debate
and heated controversy. A wide range of alternative viewpoints
is explored until eventually (usually in less than a generation)

eh
134 PROGRESS AND REVOLUTION

one of these new views vanquishesall the others and becomes

established as the new paradigm, demanding unquestioning
adherence from scientists in the field. Indeed, Kuhn even goes
so far as to say that a discipline is not scientific if the discussion
of critical, foundational problems continues unabated.® If
revolutions really had such a character, if they really differed so
much from “normal science,”’ they would, of course, be
singularly interesting historical phenomena (both from a con-
ceptual and from a sociological point of view).
There is much evidence to suggest, however, that scientific
revolutions are not so revolutionary and normal science not so
normal as Kuhn's analysis would suggest. As we have already
observed, debate about the conceptual foundations of any
paradigm or research tradition is a historically continuous
process. The posing and resolving of conceptual problems—a
phenomenon Kuhn relegates chiefly to short-lived periods of
crisis—continues unabated throughout the life of any active
research tradition. As several critics have noted, Kuhn and his
followers have been unable to point to any lengthy period in the
history of any major paradigm when its partisans closed their
eyes to the conceptual problems which the paradigm generated.
One important reason why these basic framework questions
rarely go away springs from another feature of science which
Kuhn has ignored; namely, the rarity with which any one
paradigm achieves that hegemony in the field which Kuhn
requires for “normal science.”” Whether we look at nineteenth-
century chemistry, eighteenth-century mechanics, twentieth-
century quantum theory; whether we examine evolutionary
theory in biology, mineralogy in geology, resonance theory
in chemistry or proof theory in mathematics, we see a far
more diversified situation than Kuhn’s account allows. Two (or
more) research traditions in each of these areas have been the
rule rather than the exception. Indeed it is difficult to find any
extended period of time (even on the order of a decade) when
only one research tradition or paradigm stood alone in any
branchofscience.
It may be helpful to select some of Kuhn's own examples to
see how badly his analysis founders:
PROGRESS AND REVOLUTION 135

The Newtonian revolution in mechanics. Like several other

scholars, Kuhn’s archetypal example of a scientific revolution is
the development of Newtonian mechanics from 1700 to the
middle of the nineteenth century; this is scarcely surprising
since there can have been few more successful paradigms or
research traditions than this one. But eighteenth-century me-
chanics offers few consolations for a Kuhnian theory of
revolutions. From its first reception at the hands of Huygens
and Leibniz, its core assumptions were under continuous critical
scrutiny, even from many physicists who readily conceded its
mathematical virtuosity and its empirical triumphs.’ George
Berkeley, several of the early Bernoullis, Maupertuis, the
Hutchinsonians, Boscovich, the young Kant, and even Euler
raised a number of fundamental problems about the ontological
foundations of Newtonian mechanics. At the same time, many
other scientists (e.g., Hartley, LeSage, Lambert) were taking
issue with the methodological assumptions of the Newtonian
tradition.* Although there can be no doubt that the Newtonian
tradition had a tremendous impact on eighteenth-century
rational mechanics, that tradition exhibited neither the unan-
imity of adherence nor the suspension of critical judgment
which, on Kuhn’s view, typify the aftermath of a scientific
revolution.

The Lyellian revolution in geology. On Kuhn’s account, it

was the publication of Charles Lyell’s Principles of Geology
(1830-33) which established the first important scientific tradi-
tion in geology.’ In other words, Lyell’s Principles provided both
a paradigm (“‘uniformitarianism’’) and some working exemplars
for geology, which collectively constituted a scientific revolution.
Even on the most charitable interpretation of the historical
evidence, the Lyellian revolution does not support Kuhn’s
historiography. In the first place, there was nothing global
about the Lyellian revolution. Restricted largely to England and
America, Lyell’s work was hardly ever taken seriously in
Germany and France, and virtually no continental geologist
became a “‘Lyellian.”” Even within the English-speaking world,
Lyell’s ideas—though widely cited—were severely criticized and
 vee A EY

136 PROGRESS AND REVOLUTION

rarely accepted without emendation. Indeed, the most charac-

teristic features of Lyell’s geological system (namely, his ‘‘degree
uniformitarianism,” his theory of climate, and his volcanic
theories) were accepted by very few geologists. Equally, there
Ge a ORENASSEEESCATRNtinne

was none of that cessation of foundational debate associated

with the end of a Kuhnian revolution. In the two generations
after Lyell’s work most geologists, cosmogonists, geographers,
and bio-geologists (most notably Charles Darwin) found it
necessary to abandon many of the most fundamental assump-
tions of the Lyellian paradigm (e.g., Lyell’s conviction that the
entire spectrum of animals and plants is fully represented in
each geological epoch). Even before evolutionary theory discred-
ited Lyellian geology, many critical voices had been raised
against virtually all of its core presuppositions. What is true of
Lyell is equally applicable to the whole of early nineteenth
century geology: there was no geological paradigm which was
either universally or uncritically accepted. A multiplicity of
alternative frameworks was the rule rather than the exception.
It is this perennial co-existence of conflicting traditions of
research which makes the focus on revolutionary epochs so
misleading. These traditions are constantly evolving, their
relative fortunes may shift through time, old traditions may be
largely displaced by new ones, but it is generally unhelpful to
focus attention on certain stages of this process as revolutionary
and on others as evolutionary. The examination of fundamen-
tals, the exploration of alternative frameworks, the replacement
of older perspectives by newer and more progressive ones take
place unceasingly in science—and in every other intellectual
discipline for that matter. This is not to say, of course, that
every scientist is (as Popper would have it) constantly criticizing
the framework or tradition within which he works. Many
scientists at any given time will be taking the tradition as
“given’”’ and will be seeking constructively to apply it to a wider
range of unsolved empirical problems (what Kuhn calls “puzzle
solving’’). But to imagine that all scientists are doing that all of
the time—except in rare periods of crisis—is to take remarkable
liberties with the actual evolution of the sciences.
Clearly, if the notion of a scientific revolution is to be

i
PROGRESS AND REVOLUTION 137

historically fruitful, we must be able to define scientific

revolutions in such a way that their occurrence allows for a
persistent disharmony among scientists concerning the basic
foundationsof their discipline.
One natural approach here would seem to involve a discus-
sion of numbers. One might suggest, for instance, that a
scientific revolution occurs when a sizable numberof influential
scientists in any discipline abandon one research tradition and
espouse another. But what constitutes a ‘‘sizable number’’?
This is not a mere matter of counting heads, or speaking of a
revolution occurring as soon as more than half of the scientific
community adopts one particular research tradition. Revolutions
can be, and often have been, achieved by a relatively small
proportion of scientists in any particular field. Thus, we speak
of the Darwinian revolution in nineteenth-century biology, even
though it is almost certainly the case that only a small fraction
of working biologists in the last half of the nineteenth century
were Datwinians. We speak of a Newtonian revolution in
early eighteenth-century physics, even though most natural
philosophers in the period were not Newtonians. As we have
seen, it is common to speak of Lyell as having wrought a
revolution in geology, even though the bulk of his scientific
contemporaries had grave reservations about the research
tradition which he espoused.
Examples such as these suggest that a scientific revolution
occurs, not necessarily when all, or even a majority, of the
scientific community accepts a new research tradition, but
rather when a new research tradition comes along which
generates enough interest (perhaps through a high initial rate of
progress) that scientists in the relevant field feel, whatever their
own research tradition commitments, that they have to come to
terms with the budding research tradition. Newton created the
stir he did because, once the Principia and the Opticks were
published, almost every working physicist felt that he had to
deal with the Newtonian view of the world. For many, this
meant finding cogent arguments against the Newtonian system.
But what was almost universally agreed was that Newton had
developed a way of approaching natural phenomena which
138 PROGRESS AND REVOLUTION
could not be ignored. Similarly, late nineteenth-century biolo-
gists, whether fervent Darwinians or confirmed anti-evolution-
ists, found themselves having to debate the merits of Darwin-
ism. To put the point in a more general fashion, I am
suggesting that a scientific revolution occurs when a research
tradition, hitherto unknown to, or ignored by, scientists in a
given field, reaches a point of development wherescientists in
the field feel obliged to consider it seriously as a contender for
the allegiance of themselves or their colleagues.
It is worth noticing that I have defined revolutions in such a
way as to presuppose nothing whatever about their inherent
rationality or progressiveness. Scientific revolutions can occur
even when it is entirely irrational or nonrational considera-
tions which bring a new research tradition to everyone’s
attention. A revolution could, in principle, involve the abandon-
ment of more progressive research traditions for less progressive
ones. In short, whether a scientific revolution is rational and
progressive is a contingent matter. In sharp contrast to Kuhn,
who argues that scientific revolutions are ipso facto progres-
sive,'° I want clearly to separate the question of whether a
revolution has occurred from a determination of the progressive-
ness of that revolution. Otherwise, the claim that science is
progressive becomes vacuously true, and therefore cognitively
worthless.
Even so conceived, it must still be stressed that scientific
revolutions are not the core unit for analysis which some
historians and philosophers have imagined. Once we accept that
the emergence of new research traditions, and the criticism and
modification of older ones is the ‘‘normal’’ state of science, then
a preoccupation with revolutions—as historical phenomena
different in kind from ordinary science—must be avoided. But
we can go further than this. If theories and research traditions
are undergoing continuous appraisal and evaluation, then the
natural focus for the historian’s interest should be specific
research traditions and the debates about the relative merits of
the extant traditions in any science. A successful revolution is
nothing more than a consequence of, an obituary for, a
particularly dramatic and decisive encounter between vying
research traditions.
PROGRESS AND REVOLUTION 139

Revolution, Continuity, and Commensurability

Among scholars who have thought about the process of

scientific change, there is a central cleavage between those who
are struck by the successive revolutionary convulsions of
scientific thought, and those who are more vividly impressed by
the remarkable continuities which science exhibits through its
history. The ‘‘revolutionary” school stresses the very different
kinds of metaphysics of nature implicit in successive scientific
periods. Thus, Aristotle believed in a plenum; seventeenth
century atomists, in a void. Eighteenth-century chemists be-
lieved that air was composed of highly reactive chemical
substances and that fire was not an element. Seventeenth-
century and eighteenth-century geologists saw the history of the
earth in terms of processes of change and transformation quite
unlike anything still occurring today on the face of the globe;
some nineteenth-century geologists, on the other hand, were
impressed by the uniformitarian character of the earth’s history.
By contrast, the “‘gradualists” stress the degree to which
science manages to preserve most of what it has discovered.
They point out that, for all the seeming “‘revolutions”’ in optics
since the early seventeenth century, we still espouse substan-
tially the same sine law of refraction that Descartes did. They
point out that, Einstein notwithstanding, contemporary me-
chanics still utilizes almost entirely techniques worked out by
Newtonian scientists, or reasonable approximations thereto.
The gradualists see the process of knowledge acquisition as slow
and cumulative, with new truths, or better approximations
being constantly added to a reservoir of laws about nature which
have accumulated since antiquity. They point out, further, that
many seemingly radical conceptual innovations often amount to
little more than a subtle juxtaposition or realignment of
traditional elements.
Both historiographical schools have focussed on important
traits of the history of science, but neither has managed to
integrate them convincingly. When regarded from the point of
view of a problem-solving approach, it is easier to capture both
sorts of insights. The chief element of continuity, one might say,
is the base of empirical problems. Although there is some
140 PROGRESS AND REVOLUTION

change in the empirical problem domain as a result of both

time and successive research traditions, what continuity thereis
in science tends to be found in the domain of such empirical
problems. Since the 1640s, every optical theory has had to deal
with what happens to light when it is refracted through a
prism. Since antiquity, every astronomical theory has been
obliged to explain solar and lunar eclipses. Since the 1650s,
every theory of matter and of the gaseous state has been forced
to explain the (approximately) inverse relations between pres-
sure and volume of gases. Since about 1800, every theory of
chemistry has had to grapple with the role of air in the process
of combustion. History suggests that problems such as these are
a permanent fixture of the scientific scene and that, however
much the basic ontology of science changes, however many new
research traditions emerge, many of these problems will be
essential explananda for science throughout its evolution.
Where discontinuities occur is not so much at the level of
first order problems as at the level of explanation or problem
solution. There are radical differences between the way in which
a contemporary chemist explains combustion and the manner in
which his eighteenth- or nineteenth-century predecessors did.
There are crucial discontinuities between the quantum physi-
cist’s explanation of black-body radiation and a nineteenth-
century physicist’s account of the same problem. Of course, this
is not to suggest that successive research traditions have nothing
in common except a partial overlap of their empirical problems.
There are often important formal and conceptual relationships
which persist through time, and which are preserved in a
succession of research traditions. But it is basically the shared
empirical problems which establish the important connections
between successive research traditions; it is these, and these
alone, which must be preserved if science is to exhibit that
(partially) cumulative character which is so striking about much
of its history.*’
Many recent writers on the problem of scientific change,
particularly those in the “revolutionary” camp, have been
struck by the radical incommensurability between successive
research traditions. Carrying the “revolutionary” stance to its
extreme, they argue that theories before and after a revolution
+

PROGRESS AND REVOLUTION 141

are so radically different that we cannot even speak meaning-

fully of any similarities between them. Quite rightly pointing out
that a Ptolemaian and a Copernican, or a Lamarckian and a
Darwinian, or a Newtonian and a relativist look at the world in
different ways (perhaps even “look at’ different worlds,
although that strikes me as a bizarre way of stating the issue),
these writers (such as Hanson, Quine, Kuhn, and Feyerabend)
have drawn some very pessimistic conclusions about the possi-
bility of rationality in science. In several cases, they have been
led to conclude that it is, in principle, impossible to establish
that any research tradition ever rationally triumphs over
another. The logic of their arguments (which I shall examine
shortly) leads them to conclude that the history of science is but
a succession of different worldviews, and that rational choice can
never be made between such divergent schemes of the universe.
Because each has its own internal rationale and integrity, no
meaning can be attached to the suggestion that one scheme ts
more(or less) rational than another.
This argument is an important one. If true, it would mean
that science had no particular claim to our cognitive loyalities.
If there are no conceivable grounds for rational choice between
competing research traditions, then science becomes a matter of
whim and caprice, in which that tradition wins which happens
to attract the most influential adherents and the most powerful
propagandists. That may well be how science ts, but before we
accept the rather depressing conclusion that science must
necessarily be that way, it is worth examining with some care
the arguments which its proponents give for such a relativistic
notion of scientific change.
In brief, the central argument runs like this: scientific
theories implicitly define the terms which occur within them.
Hence, if two theories are different then all the terms within
them must have different meanings. (Thus, when an Einstein-
ean physicist refers to the ‘‘mass” of a particle, he means
something different from a Newtonian when the latter refers to
the ‘‘mass” of a particle.) Moreover, the argument continues,
even the so-called observational reports which scientists working
with different theories make are incommensurable, for their
observational terms are theory-laden, i.e., they are given
 142 PROGRESS AND REVOLUTION

meaning by virtue of one or another theory. That meansthat,

although scientists working in different research traditions may
sometimes make the same verbal utterances, we cannot even
then assumethat they are asserting the same thing. To accept a
theory, on this view, is to accept a quasi-private language which
i no one who does not accept the same theory can understand
or comprehend. As a result, scientists working in different
research traditions cannot communicate with, and cannot
i understand the statements of, their fellow scientists in other
i traditions. Given this general incomprehension, science emerg-
ing as a new version of the tower of Babel, theories cannot be
compared and rationally evaluated because such comparison
seemingly requires a common language for speaking about the
world.
I believe this general argument to be faulty in several
respects. It rests on a very idiosyncratic theory about how words
acquire meaning (namely, the theory of implicit definition).'? It
begs a number of questions about synonymy and translation.
Butits central flaw, for our purposes, lies in its presumption
that rational choice can be made between theories only if those
theories can be translated into one another's language or into a
third “theory-neutral"’ language. As Kuhn puts the point,
“‘(the] comparison of two successive theories demands a lan-
guage into which at least the empirical consequences of both
can be translated without loss or change.’’!* I shall maintain,
to the contrary, that even if we accept the view that all
observations are theory-laden to a degree that makes their
contents inseparable from the theory which is used to express
them, it is still possible to outline machinery for objective,
rational comparisons between competing scientific theories and
research traditions. There are two general arguments which lead
me to such a conclusion.

The argument from problem solving. In the heyday of logical

positivism, it was commonly argued that competing theories
could be evaluated by comparing their ‘“‘observational’’ conse-
quences. Given the dominance at that time of the linguistic
metaphor, this was usually conceived as a process of translating

": i
oe]

PROGRESS AND REVOLUTION 143

the predictions of competing theories (via so-called correspon-

dence rules) into some purely observational language. Because
the observational language was held to be free of any specula-
tive, theoretical biases, it was thought to provide objective
groundsfor the empirical appraisal! of vying theories. As doubts
grew about the existence of rules of correspondence and about
the existence of theory-free observation languages, philosophers
such as Kuhn, Hanson, and Feyerabend beganto despair about
the possibility of any objective yardstick for comparing different
theories and suggested that theories were incommensurable and
thus not open to objective comparison.
What this approach ignores is that neither correspondence
rules nor a theory-free observation language are necessary for
comparing the empirical consequences of competing theories.
For even without correspondence rules and without a purely
observational Janguage, we can still talk meaningfully about
different theories being about the same problem, even when the
specific characterization of that problem is crucially dependent
upon many theoretical assumptions.
How, acknowledging the mannerin which theories affect our
characterization of what problemsare, can we nonetheless show
that different theories address the “‘same’’ problem? The
answer is straightforward: the terms in which a problem is
characterized will generally depend upon the acceptance of a
range of theoretical assumptions, T,,T,, ... , T,. These
assumptions may, or may not, constitute the theories which
solve the problem. If a problem can be characterized only
within the language and the framework of a theory which
purports to solve it, then clearly no competing theory could be
said to solve the same problem. However, so long as the
theoretical assumptions necessary to characterize the problem
are different from the theories which attempt to solve it, then it
is possible to show that the competing explanatory theories are
addressing themselves to the same problem. Consider a very
elementary example. Since antiquity, scientists have been
concerned to explain why light is reflected off a mirror or other
polished surface according to a regular pattern. Relating the
incident to the reflected angle, the problem of reflection, thus
 144 PROGRESS AND REVOLUTION

characterized, involves many quasi-theoretical assumptions,

e.g., that light movesin straight lines, that certain obstacles can
changethe direction of a ray of light, that visible light does not
continuously fill every medium, etc. Does the existence of these
theoretical assumptions entail that no two theories can be said
- to solve the problem of reflection? The answer is clearly
seen pe tie eth

negative, provided that the theories which solve the problem are
not inconsistent with those relatively low-level theoretical as-
sumptions required to state the problem.'* Throughout the late
seventeenth century, for example, numerous conflicting theories
of light (including those of Descartes, Hobbes, Hooke, Barrow,
Newton, and Huygens) addressed themselves to the problem of
reflection. The various optical theories were all regarded as
solving the problem of reflection, because that problem could
be characterized in a way which was independent of any of the
theories which sought to solveit.
I do not mean to suggest, of course, that a/l the problems
which a theory or research tradition attempts to solve can be
characterized independently of the theory (or theories) which
solves them. The determination of the “independence”’ of any
specific problem must depend upon the particularities of the
case. It is my impression, however, that there are far more
problems common to competing research traditions than there
are problems unique to a single one. These shared problems
provide a basis for a rational appraisal of the relative problem-
solving effectiveness of competing research traditions.
I must stress again that this argument does not presuppose
that empirical problems can be stated in some purely observa-
tional, nontheoretical language. To speak of (for instance) light
being refracted through a prism makes a numberof theoretical
assumptions (among them, that light moves, that something
happens to light while it is ‘inside’ a prism, etc.). It is not the
atheoretical nature of empirical problems which is being alleged
here. Rather, the weaker claim being madeis this: with respect
to any two research traditions (or theories} in any field of
science, these are some joint problems which can be formulated
so as to presuppose nothing which is syntactically dependent
upon the specific research traditions being compared. Thus,
when eighteenth-century Newtonians and Cartesians spoke
|

PROGRESS AND REVOLUTION 145

about the probiem offree fall, they were identifying the same
problem—for all the profound differences between their re-
spective research traditions. When these same natural philos-
ophers discussed the problem whyall the planets moved in the
same direction about the sun, they were also in complete
agreement about the nature and meaning of the problem
(although they did quarrel about its relative importance as a
problem). When early nineteenth-century geologists debated the
explanation of stratification, they could all—whether uniformi-
| tarian or catastrophist, whether Neptunist or Vulcanist,
' whether Huttonian or Wernerian, whether God-fearing or God-
| denying, whether French, English or German—agree that one
problem for any geological theory was that of explaining how
i such uniform and distinct layers had been formed.
Kuhn has been misled by his discovery that some empirical
problems are not jointly shared between different traditions or
paradigms (which is certainly true) into believing that no
problems are identical. The generalized thesis of problem
incommensurability is as perverse as the limited thesis of partial
non-overlap is profound.

The argument from progress. The argument just given pre-

sumesthat there are ways of identifying and characterizing some
problems which are neutral with respect to the various theories
which attempt to solve those problems. But there are doubtless
those philosophers who will deny that there is any way in which
empirical problems can be characterized so as to allow us to
speak of “two theories (or research traditions) solving (or failing
to solve) the same problem.” I have yet to see any compelling
arguments to that effect, but even if there were—i.e., even if we
grant that it cannot be decided whether theories are dealing
with the same problems—there is still scope for the objective
evaluation and comparison of incommensurable theories and
research traditions. To see why this is so, we need only trace
out certain corollaries of our earlier discussion of scientific
rationality. It was observed there that rationality consisted in
accepting those research traditions which had the highest
problem-solving effectiveness. Now, an approximate determina-
tion of the effectiveness of a research tradition can be made
146 PROGRESS AND REVOLUTION

within the research tradition itself, without reference to any

other research tradition. We simply ask whether a research
tradition has solved the problems which it set for itself; we ask
whether, in the process, it generated any empirical anomalies or
conceptual problems. We ask whether, in the course of time, it
has managed to expand its domain of explained problems and
to minimize the number and importance of its remaining
conceptual problems and anomalies.'* In this way, we can come
up with a characterization of the progressiveness (or regressive-
ness) of the research tradition.
If we did this for all the major research traditions in science,
then we should be able to construct something like a progressive
ranking of all research traditions at a given time. It is thus
possible, at least in principle and perhaps eventually in practice,
to be able to compare the progressiveness of different research
traditions, even if those research traditions are utterly incom-
mensurable in terms of the substantive claims they make about
the world!'®
Hence, even if we could not in principle ever find a way of
translating Newtonian mechanics into relativistic mechanics;
even if we could never find a way of comparing the substantive
claims of twentieth-century particle physics with nineteenth-
century atomism; even if, more generally, we could never say
that two theories dealt with some of the same problems; it
would still be possible in principle for us to make an assess-
ment, on rational grounds, of the relative merits of these (or any
other) research traditions. This point can be readily generalized
by noting that there are many criteria for the comparison of
competing theories which do not require any degree of com-
mensurability at the observational level. We might, for instance,
compare theories with respect to their internal consistency or
their coherence. Equally, we might ask of two (or more)
theories, which is the simpler? or which has been refuted? or
which yields the more precise predictions? Because such
properties (including progressiveness) can be definitely specified,
we can Say that the possible incommensurability of theories and
research traditions (so far as their substantive claims about the
world are concerned) does not preclude the existence of com-
parative appraisals of their acceptability.”

ee
reg

PROGRESS AND REVOLUTION 147

Non-Cumulative Progress
Ever since the appearance of Condorcet’s Sketch Towards a
History of the Progress of the Human Mind, many philosophers
and historians of science have developed, at least in outline
form, theories of cognitive progress. From Whewell, Peirce, and
Duhem through Collingwood, Popper, Reichenbach, Lakatos,
Stegmiiller, and Kuhn, the search for adequate models of
cognitive progress has been, if not commonplace, at least not
rare. For al! their differences, these models of progress—save
Kuhn’s'*—share one common feature: a conviction that it is
only possible to speak of progress if knowledge is acquired
through purely cumulative theories. By ‘‘purely cumulative
theories,’ 1 mean those theories which may add to the store of
solved problems, but which never fail to solve al/ the problems
succesfully solved by their predecessors. Put slightly differently,
these thinkers argue that a necessary condition for one theory,
T,, to represent progress over another, 7,, is that 7, must solve
all the solved problems of 7,. Although this cumulative
conception of progress is usually associated with Popper and
Lakatos, it was probably most succinctly formulated by Colling-
wood when hewrote:

If thought in its first phase, after solving the initial problems of that phase, is
then through solving these, brought up against others which defeat it; and if
the second solves these further problems without losing its hold on the solution
of the first, so that there is gain without any corresponding loss, then there is
progress. And there can be progress on no other terms. [f there is any loss, the
problem of setting loss against gain is insoluble. '°

What kind of insolubility is being asserted here? Collingwood

never tells us, but presumably what lies behind his concern is
the belief that unless the solved problems of one theory form a
proper subset of the solved problemsof a rival, then we have no
way of knowing which theory is the more progressive, for we
cannot then reduce progress to a simple additive relation.
Similar concerns motivate the approach of Popper and
Lakatos to the nature of progress. In his ‘‘requirements for the
growth of knowledge,” for instance, Popperinsists that in order
for us to be able to show that a theory is progressive with respect
to a competitor, we must be able to show that it entails every
148 PROGRESS AND REVOLUTION

fact entailed by its competitor.?° In the absence of such an

entailment, progress (in the Popperian sense) is impossible,
Lakatos, for all his quarrels with Popper, takes the same view
on this issue: a precondition for saying that a series of theories
{i.e., a “research programme’’) is “progressive” is that each
later member in the series must entail all the corroborated
content of its predecessor.”!
Heinz Post has also recently defended the view that new
theories always absorb the problem-solving successes of their
predecessors. Post goes so far as to make the ‘‘claim that, as a
matter of empirical historical fact . . . [past] theories always
explained the whole of [the well-confirmed part of their
predecessors] . . . contrary to Kuhn, there is never any loss of
successful explanatory power.’’??
The appeal of approaches such as those just mentioned lies in
their enormous simplicity. If progress occurred in the manner
they require, then we would not have to worry about the
counting or the weighting of problems. If all the previously
solved problems in any science are always solved by its latest
theories, and if those later theories solve still other problems
(regardless of their number or weight), then it is obviously the
case that the later theories exhibit progress over the earlier ones.
Whatvitiates this approach to the problem of progress is that
the conditions it requires for progress are rarely satisfied in the
history of science. As Kuhn, Feyerabend, and others have
claimed, there are usually problem losses as well as problem
gains associated with the replacement of any older theory by a
newer one.”°
Wecan get a sense of just how substantial these losses can be
by considering a particularly vivid historical example, namely,
the shift in geological problemsin the early nineteenth century.
Prior to Hutton, Cuvier and Lyell, geological theorists had been
concerned with a very wide range of empirical problems, among
them: how deposits get consolidated into rocks; how the earth
originated from celestial matter and slowly acquired its present
form; when and where the various animals and plants origin-
ated; how theearthretains its heat; the subterraneousorigins of
volcanos and hot springs; the origin and constitution of igneous
rocks; how and when various mineral veins were formed.
Solutions, of varying degrees of adequacy, had been offered in

i
PROGRESS AND REVOLUTION 149

the eighteenth century to each of these problems. Yet after

1830, particularly with the emergence of stratigraphy, there
were no serious geological theories which addressed themseives
to many of the problems mentioned above. Does that mean (as
Popper, Lakatos, Collingwood, and others would have it) that
geology was not progressive between 1830 and-about 1900 (when
many of these issues began to re-emerge)? To jump to that
conclusion would be rash, for it would ignore the fact that
geological theories after Cuvier and Lyell successfully addressed
themselves to a very different range of empirical problems,
including those of bio-geography, stratigraphy, climate, erosion,
and land-sea distribution. A full analysis of this shift, for which
there is not space here, would show that the precision and range
of empirical problems that could be solved by mid-nineteenth-
century geology (as well of the acuteness of the conceptual and
anomalous problems generated) compared favorably to the
overall problem-solving success which late cighteenth-century
geological theories could claim for themselves. Although this
example illustrates more than most just how many problems
may fall away from the concernsof a scientific community, the
phenomenon it exemplifies is commonplace.
That phenomenonis illustrated within physics by the failure
of Newton’s optics to solve the problem of refraction in Iceland
spar (which had been explained by Huygens’ optics), and by the
failure of early nineteenth-century caloric theories of heat to
explain phenomena of heat convection and generation, problems
which had been solved by Count Rumford in the 1790s. Within
chemistry, many problems which had been solved by the early
theories of elective affinity were not solved by Dalton’s later
atomistic chemistry.** A still better example is afforded by
Franklinic electrical theory. Prior to Franklin, one of the central
solved problems for electricity was the mutual repulsion of
negatively charged electrical bodies. Various theories, especially
vorticular ones, had solved this problem by the 1740s. Frank-
lin’s own theory, which was widely accepted from the middle to
the end of the eighteenth century, never adequately came to
grips with this problem.’*
As these examples show, empirical problems often are either
abandoned or relegated to insignificance and any adequate
theory of scientific development must presumably allow that
 150 PROGRESS AND REVOLUTION

such constrictions of the problem domain can, under certain

circumstances, be progressive.
I have suggested that such situations can be handled by
TEARCOLIBRAtesoe a

allowing for the relative importance of various empirical

problems. Knowledge of the relative weight or the relative
numberof problems can allow us to specify those circumstances
under which the growth of knowledge can be progressive even
when welose the capacity to solve certain problems. In this way,
we can avoid the alleged Collingwoodian ‘“‘insolubility’’ of how
to make a progressive choice between systems, neither of whose
problem sets falls neatly inside the other’s.”*

In Defense of ‘‘Immature’’ Science

Both Kuhn and Lakatos are committed to the view that there
are two radically different types of science, corresponding
roughly to the ‘‘early’’ and “‘advanced’’ stages of scientific
activity. Although called by different names (for Lakatos,
“immature” and ‘‘mature” science; for Kuhn, ‘‘pre-”” and
“post-paradigm” science),?? and defined differently, both
writers are committed to the view that the various sciences, at
various times, undergo the transition from infancy to majority
and that when they do, the rules of the scientific game change
substantially. For Kuhn, the transition occurs when one
paradigm establishes monopoly over the field and when ‘‘nor-
mal science’ first ensues. For Lakatos, a science reaches
maturity when scientists in that field consistently ignore both
anomalous problems and outside intellectual and social influ-
ences, and focus almost entirely on the mathematical articula-
tion of research programmes. Thus, what chiefly characterizes
a mature science, for both Kuhn and Lakatos, is the emergence
of paradigms(or research programmes) which are autonomous,
and thereby independent from outside criticism. This transition
is more than a nominal one; Kuhn and Lakatosalike insist that
mature science is more progressive, more genuinely scientific
than its immature counterpart.
There are several troubling aspects to the concept of a mature
science (at least as developed by Kuhn and Lakatos). The
suggestion that every (or even that any) science undergoes a

ql
or

PROGRESS AND REVOLUTION 151

permanent transition of the kind which Kuhn and Lakatos

describe does not square with what we know aboutthe evolution
of science. Kuhn can point to no major science in which
paradigm monopoly has been the rule, nor in which founda-
tional debate has been absent. Lakatos, for his part, has
identified no (physical) science in which the disdain for anomaly
and the indifference to extra-programmatic conceptual prob-
lems have been the prevailing features. As a result, it is extremely
unclear whether the notion of a “mature” science finds any
exemplification whatsoever in the history of science.
Even if mature sciences existed, the Kuhn-Lakatosthesis that
they would be intrinsically more progressive and more scientific
than ‘“‘immature’’ ones has not been established. Kuhn has not
shown that more empirical problems necessarily would be solved
if one paradigm tyrannizes a scientific domain. Lakatos has not
made a convincing case for his claim that autonomous, anomaly
ignoring research programmesarelikely to be more progressive
than non-autonomous, anomaly-recognizing ones.”*® In the
absence of cogent arguments for the greater rationality of
mature science, we can only conclude that the preference voiced
by Kuhn and Lakatos for mature sciences is without foundation.
A third difficulty with the doctrine of mature and immature
sciences is the scope it allows the builder of any model of
scientific rationality to dismiss as irrelevant any historical
counter-examples to his model. Since the models are chiefly
designed as replicas of ‘‘mature science,” any actual scientific
examples which fail to fit the models can be explained away as
proto- or pseudo-science rather than being regarded as genuine
exceptions to the models. The mature-immature dichotomy is
thus methodologically suspect because it effectively renders
these models of scientific rationality immune from empirical
criticism.”°
In arguing against the existence and the desirability of
mature science, so construed, I am manifestly not claiming that
the later stages of a science exhibit all the structural and
methodological features of theories in earlier stages. We may yet
find a characterization of mature science which will do justice
both to history and to rationality.°° But the concept of mature
science as conceived by Lakatos and Kuhn unfortunately does
neither,
 Part Two

Applications
 Chapter Five
History and Philosophy of Science
Philosophy of science without history
of science is empty; history of science
without philosophy of science is
blind. 1. LaKaTos (1971), p. 91

Because the stimulus for developing the model in Part One

camechiefly from the writings of historians and philosophers of
science, it is appropriate to begin exploring the ramifications of
that model by examining its consequences for this field (or
fields). To say as muchis already to indicate what must be one
of our core concerns; for the parenthetical expression above
stresses just how unclear scholarsstill are about whether history
of science and philosophy of science are two distinct domains or
whether, as some writers allege, they are so intimately con-
nected as to be onefield, incapable of meaningful separation.
Couched in these terms, the issue might seem to be largely
verbal—one of those boring disputes about where the boun-
daries of one discipline end and another begins. But in this
case, there are some substantive issues that hang on the
separability of history and philosophy of science. Questions
about the aims, the methods of inquiry and the manner of
legitimating both historical and philosophical claims are bound
up in the question: are history of science and philosophy of
science autonomousenterprises?
155
156 HISTORY AND PHILOSOPHY OF SCIENCE

The standard view, of course, regards history of science and

philosophy of science as radically different, if perhaps some-
times complementary, ways of studying science. The historian,
on this view, is dealing with facts and data, seeking to arrange
them into a convincing and coherent tale about how scientific
ideas have evolved. Philosophy of science, by contrast, is
commonly perceived as a normative, evaluational and largely a
priori investigation of how science ought to proceed. On this
view, the gap between history and philosophy of science is as
broad as, indeed is illustrative of, the divide between matters
of fact and matters of value. History is irrelevant to the
philosopher because he is not concerned about what science has
been, but rather how it should be. Philosophy is irrelevant to
the historian because it is not his job to make normative
judgments aboutthe figures he studies.
Research in the last twenty years has done muchto highlight
the weaknesses in the standard account. Agassi,’ Griinbaum?
and others have shown how much writing in the history of
science is laden with implicit philosophical assumptions, as-
sumptions which decisively determine the character of the
history that is produced. (To take an overly simple example, if a
historian is convinced that experiments can be the only decisive
grounds for abandoning a theory, then his history will tend to
focus exclusively on so-called crucial experiments.) The thesis is
not merely that philosophical assumptions have influenced
historical scholarship, but that they must do so, because history
(like science) has no neutral data, and because the treatment of
any particular historical episode is going to be influenced to
some degree by one’s prior philosophical conceptions of whatis
important in science.
The correlative case for the philosophy of science has been
argued with equal fervor by a number of thinkers, including
Whewell, Hanson, Kuhn, Toulmin, Lakatos, McMullin, and
Feyerabend.*> While granting that the aim of philosophical
inqury is the production of a set of norms (e.g., for choosing
between competing theories), these critics of the standard view
point out that any philosophical! theory of science which failed
completely to square with the history of science would be deemed
unacceptable. *Confronted with a philosopher’s account of,
HISTORY AND PHILOSOPHY OF SCIENCE 157

say, rational theory acceptance which entailed that the whole of

the history of science was irrational, we would tend to view that
as a reductio ad absurdum of the theory of rationality rather
than as a demonstration that science itself had been a sequence
of entirely irrational preferences
If these critics are right, there are relations of mutual
dependence between history of science and philosophy of
science which make nonsense of any attempt to allow them an
autonomous development. But there are, prima facie, some
difficulties with an integrated view of history and philosophy of
science, difficulties so acute that most thinkers have remained
unconvinced by the claim of mutual dependence. Foremost
among these difficulties is the vicious circularity which it
seemingly entails. If the writing of history of science presup-
poses a philosophy of science and if a philosophy of science is
then to be authenticated by its capacity to lay bare the ra-
tionality held to be implicit in the history of science, how can
we avoid automatic self-authentication, since the history we
write will presuppose the very philosophy which the written
history will allegedly test? Other difficulties abound. If, as
seems likely, virtually all the available philosophies of science
can do scant justice to the history of science, why should the
historian of science take them seriously as theoretical tools for
organizing his research? Similarly, if most history of science has
been written utilizing discredited philosophical models of
science, why should the philosopher feel constrained to test his
carefully worked out models against historical ‘‘data’’ which
were collected under the aegis of a naive or opposing philosophy
of science? There are some slightly more technical troubles as
well. Even if we grant that in some sense the actual course of
science should have some bearing on the philosophy of science,
how close does the fit between actual history and its normative
reconstruction have to be? Because no one, neither historian
nor philosopher, is committed to the view that the whole of
science is rational,* why should the philosopher be disturbed if,
on his account, many episodes in the history of scientific ideas
turn out to have irrational elements?
These are large, and still unanswered questions. The aim of
this chapter is to provide someof those answers.
158 HISTORY AND PHILOSOPHY OF SCIENCE

The Role of History in the Philosophy of Science

There are, of course, already certain areas of philosophy of

science where a significant empirical input from the sciences is
taken for granted. To take but two examples, the philosophy of
space and time and the philosophy of biology are universally
acknowledged to draw heavily upon the recent state of natural
sciences. But in that portion of philosophy of science concerned
with general methodology (e.g., with norms for theory appraisal
and assessment), there is still a widespread discomfiture with
the suggestion that empirical data about the evolution of science
are relevant or decisive.
Before we attempt to resolve these questions, it will be helpful
to remind ourselves of one elementary but crucial distinction
which is germane to this discussion: specifically, the distinction
between the history of science itself (which, at a first approxi-
mation, can be regarded as the chronologically ordered class of
beliefs of former scientists) and writing about the history of
science {i.e., the descriptive and explanatory statements which
historians make about science). Vital though the distinction is,
it is often forgotten—in part, presumably, because speakers of
the English language use the same name for both. Because
some of the confusion about the relations between history of
science and philosophy of science derives from an equivocation
on these two different senses, I shall use ““HOS,” to refer to the
actual past of science and ‘“‘HOS,”’ to refer to the writings of
historians about that past.
A fresh version of the traditional case for the autonomy of
philosophy of science (in the sense of general methodology)
from HOS, was recently published by Ronald Giere.* His
approach involves the familiar insistence that philosophy of
science is normative, and because one cannot derive norms from
“facts,” he does not see how the history of science could be
relevant to philosophy. He goes on to say that although a
philosopher might come to some new insights by studying
HOS,, such study is no part of the authentication or the
validation of those insights, since (Giere tells us) they could
have been discovered in any case without those historical
 159
SCIENCE
HISTORY AND PHILOSOPHY OF
must not
examples. Finally, Giere insists that the philosopher
roles is to
become a slave to HOS, because one of his primary
criticize the theories of the past. For such criticism to have any
s forit.
bite, we must have independent, nonhistorical ground
Giere’s views (which are, as he says, “fairly representative
plausible at
of the majority of philosophers of science’’®) seem
e under detailed
first glance. But they quickly begin to crumbl
of
examination. As he himself concedes, if any philosophy
s scientific
science were to entail that virtually all our previou
doubts
judgments were irrational, then we would have grave
ophy of science ] to be talking
about the “‘claim [of that philos
ophical
about scientific theories.’ Precisely because “philos
a priori, ’ they must captur e some
theses cannot be completely
hunche s about which theorie s are
of our prephilosophical
not.* If these hunche s do not come from
rational and whichare
obtain them? Giere’s answer gives the game
HOS,, where do we
science that
away: it is, says Giere, to recent and contemporary
tion and
the philosopher of science must look for inspira
“actual
legitimation. Giere fails to see that his use of current
mechanics,
scientific practice’ (his examples are quantum
is itself the
molecular biology, and contemporary psychology’)
claims. The fact
invocation of HOS, to adjudicate philosophical
ly
that a scientific theory is still believed and is current
ahistorical.
undergoing development scarcely makes that theory
philos opher of science will
Every example which a Gierean
from history . Giere’s own
discuss is drawn from the past,
but they are
historical preferences may be for the recent past,
nonetheless historical for all that.
tion
What lies behind Giere’s point, I believe, is a recogni
(note the subscri pt) focusse s on the distant
that much of HOS,
ts of
past, and that thereare asyet all too few historical accoun
science can
recent HOS,. But the fact that philosophy of
the parasitic
dispense with HOS, does not militate against
Whati s clear,
dependence of philosophy of science upon HOS,.
riptive
therefore, is that a resolution of the normative/desc
which are ground ed
paradox is as crucial for those philosophies
phies which
in contemporary science asit is for those philoso
ss to say, a tu quoque
look back beyond our own time. Needle
. ail

160 HISTORY AND PHILOSOPHY OF SCIENCE

argument of this type does not solve that core problem; to the
contrary, it accentuates its importance by revealing its
universality.
I shall propose one possible way out of the paradox. Let us
begin by returning to the distinction between HOS, and HOS,,.
Within HOS,, there is, I shall claim, a subclass of cases of
theory-acceptance and theory-rejection about which most scien-
tifically educated persons have strong (and similar) normative
intuitions. This class would probably include within it many
(perhaps even all) of the following: (1) it was rational to accept
Newtonian mechanics and to reject Aristotelian mechanics by,
say, 1800; (2) it was rational for physicians to reject homeopathy
and to accept the tradition of pharmacological medicine by, say,
1900; (3) it was rational by 1890 to reject the view that heat was
a fluid; (4) it was irrational after 1920 to believe that the
chemical atom had no parts; (5) it was irrational to believe after
1750 that light moved infinitely fast; (6); it was rational to
accept the general theory of relativity after 1925; (7) it was
irrational after 1830 to accept the biblical chronology as literal
account ofearth history.
The precise dates here are not important, nor yet is any single
item on the list. What I shall maintain, however, is that there is
a widely held set of normative judgments similar to the ones
above. This set, constitutes what I shall call our preferred
pre-analytic intuitions about scientific rationality (or ‘“‘PI,”’ for
short). (This set is a very small subset of all our beliefs about
HOS,.) Our convictions about the rationality or irrationality of
such episodes are clearer and more firmly rooted than any of
our overt and explicit theories about rationality in the abstract.
Particularly decisive here are those theories and research
traditions which have been the most global and the most
influential, i.e., those which have for long epochs provided the
motivation and presuppositions for a wide range of detailed
theorizing. Any model of rationality which led to the conclusion
that the acceptance of most of these doctrines were ill-founded
would have few claims on our loyalty."° As a result, our
intuitions about such cases can function as decisive touchstones
for appraising and evaluating different normative models of
rationality, since we maysay thatit is a necessary condition of
HISTORY AND PHILOSOPHY OF SCIENCE 161

any acceptable model of rationality that it square with (at least

some of) our PIs.
How, in practice, can such episodes test a putative model of
rationality? In outline, the procedure is a simple one. Any
philosophical modei will specify certain parameters as being
relevant to the acceptance of a theory (e.g., in the case of the
model in Part One, these would be the solved, anomalous and
conceptual problems exhibited by any theory and its competi-
tors). Historical research into the case in hand would indicate
what their values should be. Once these values are specified, the
model should lead us to a determination of the historical
rationality of accepting the theory in question. If the evaluation
issuing from the model accords with our pre-analytic intuitions,
then the latter provide support for the model, if, on the other
hand, the model’s verdict contradicts our pre-analytic judg-
ments, then the model is in serious jeopardy.
In the extremecase, a proposed model of rationality would be
justifiably dismissed out of hand if, when applied to the cases
involved in PI, it entailed that all our intuitions were incorrect,
for it would have failed to capture the very rationality it was
designed to explicate. We should be very explicit about what we
are committing ourselves to in taking this approach: (1) that at
least certain specified developments in the history of science were
rational; and (2) that the test of any putative model of rational
choice is whether it can explicate the rationality assumed to be
inherent in these developments. Claim (1), modest thoughit is,
remains entirely a matter of faith since there is, in principle, no
way we could prove these cases were rational, for our criterion
of rationality itself will take their rationality for granted.
We have thus far mentioned only the extremal case in which
a methodology is discredited by every element of PI: though an
extreme, it is common enough (indeed many contemporary
philosophies of science are supported by none of the cases
above). Even so, we can move beyondthe extremal case to claim
more generally that the degree of adequacy of any theory of
scientific appraisal is proportional to how many of the PIs it can
do justice to. The more of our deep intuitions a model of
rationality can reconstruct, the more confident will we be that it
is a sound explication of what we mean by “‘rationality.”

DN
 162 HISTORY AND PHILOSOPHY OF SCIENCE

As natural as the proposal to utilize history of science as a

testing ground for philosophical models of rational choice might
seem, there are probably those purists who regard it as
unseemly that philosophy should have to look beyond itself and
its own argumentative strategies for legitimation. But where,
within philosophy, can one find the appropriate decision
criteria? Suppose that we are confronted with two competing
modelsof rationality, MR, and MR,(each of whichis internally
consistent). How, in principle, could we make a rational,
philosophical choice between them? Since both MR, and MR,
purport to define the conditions for rational choice, any choice
between them would presuppose the validity of one or the other
model (or perhaps yet a third model). We clearly have a serious
meta-level problem which can only be solved by testing the
competing models against something besides a theory of
rational choice itself. The proposal in this chapter is that our
criterion for choice between competing theories of rationality
should involve evaluating such models against those archetypal
cases of rationality (PI) which wefind in HOS,.
This proposal for the authentication of philosophical claims
about scientific rationality makes it clear that philosophy of
science depends in two crucial respects upon the history of
science. In the first respect, it aims to explicate the criteria of
rationality implicit in our preferred intuitions about certain
cases within HOS,. In the second respect, the authentication
of
any philosophical model requires careful research in HOS, in
order to assess the applicability of that model to the cases which
constitute PI.
But does this approach then make the philosophy of science
merely descriptive and rob it of any critical force? The general
answer is no. About most episodes in HOS,, we have no strong,
widely shared, pre-analytic convictions. Indeed, the chief point
of constructing a model of rationality is to use it to get
clarification about the ‘“‘fuzzy’”’ cases (which are the overwhelm-
ing majority). With respect to the latter, the philosopher’s
judgment—based on a modelof rationality authenticated by the
set of PIs—must take precedence over whatever weakly held,
pre-analytic hunches we may have about them. Asin ethics, so
in philosophy of science: we invoke an elaborate set of norms,
HISTORY AND PHILOSOPHY OF SCIENCE 163

not to explain the obvious cases of normative evaluation (we do

not need formal ethics to tell us whether the murder of a
healthy child is moral), but rather to aid us in that larger set of
cases where our pre-analytic judgments are unclear.
Hence, the philosophy of science is both descriptive and
normative, both empirical and a priori, but with respect to
different types of historical cases.
There are doubtless other ways in which the HOS, might be
helpful to the philosopher of science, ranging from providing
illustrations of philosophical claims to offering heuristic guid-
ance for the handling of specific issues.'' But the philosopher
does not need HOS, for these ends. The one and only point
where he cannot dispense with HOS, is when it comes to deciding
whether his would-be theories of rationality are, in fact, theories
of rationality.
Imre Lakatos has already made a suggestion akin to mine
aboututilizing HOS, to ‘‘test’” any model of scientific rational-
ity. There are, however, important differences of substance
between our approaches which are worth exploring. Essentially,
Lakatos’ proposal is that the best model of scientific rationality
is the one which, when applied to HOS,, will allow us to
represent the largest portion of scientific history as a rational
enterprise. It is, in short, not a small set of cases about which
we have strong intuitions (as I propose), but the whole of the
history of science (i.e., HOS,) which becomesthe criterion for
choosing between different models of rationality.‘? Lakatos’
approach strikes me as counter-intuitive for a very simple
reason: if we take his proposal seriously, then the best possible
model of rationality would be that one which resulted in the
Judgment that every decision ever made in the history of science
was rational.'*> This seemsa curious ideal to strive after, for just
as we are convinced that some scientific choices have been
rational ones, we are equally convinced (given ‘‘human nature’’)
that not all of them have been rational. Any model of rationality
which made the whole of science rational would be as suspect as
those models which make none of science rational. My
suggestion about the use of the set PI as a device for testing the
models of rationality is an effort to find a plausible middle
ground between such extremes.
164 HISTORY AND PHILOSOPHY OF SCIENCE

The Role of Normsin the History of Science

If the focus in the previous section was on therelations of
philosophy and HOS,, the concern here is with the connections,
if any, between HOS, and the philosophy of science.'* This
case is a more complex one, for the points at which valuational
elements enter HOS, are more subtle and more implicit than in
the reverse case. We will examine two very different points of
contact: in constructing a historical narrative and in offering
historical explanations.

Norms in Historical Narration

As Agassi pointed out in his classic study of the historiog-
raphy of science,'* every working historian of science must, in
sifting and arranging his data, make many assumptions about
the character of science. He must assume, amongother things,
that there were scientists, and he must be able to distinguish
those of their activities which were scientific (and therefore
relevant for inclusion in his narrative) from those which were
not. Even among theclass of scientific activities, the historian
must prune and select, since there are acute, practical limita-
tions on the completeness which HOS, can achieve. He must
decide, for instance, how much importanceto give to discussing
a scientist’s experiments, his theories, his laboratory journals,
his fecture notes, the books in his library and the like. In
principle, the historian could presumably make these decisions
by some randomizing device; but, in practice, what guides the
historian's choices are a set of assumptions about what is most
important to the doing of science. Philosophical and normative
elements inevitably enter at this stage. How much importance a
historian assigns to discussions of experiment will depend on
how important he believes experiments to be to scientific
development. The significance he assigns to the religious or
metaphysical background of a scientist will depend, again,
upon the historian’s conviction about how decisive those
elementsare in scientific deliberation.
Not surprisingly, historians with different ‘‘images”’ of science
will give radically divergent accounts of the same episodes (a
phenomenon probably exhibited most vividly in Galileo scholar-
 165
HISTORY AND PHILOSOPHY OF SCIENCE

ship—where we find Marxist, idealist, empiricist, instrumental-

ist, and pragmatic accounts of the “‘same”’ scientific achieve-
ment). There is nothing wrong in this; or perhaps we should say
that, wrong or not, it is inevitable that any historian’s account
of science is going to be colored by his views about how science
works. Such “coloring” only becomes invidious when the
motivating philosophy of science is implicit and uncritically
who
utilized, or when its existence is denied by the historian
imagines that he is free from any normative biases.
But we can go further than this. It is the historian’s
intellectual—even moral—obligation not only to be self-con-
scious about the kinds of norms he is applying, but also to see
to it that he is utilizing the best available set of norms. How can
he make that choice? By accepting that modelof rationality (or
perhaps those models if we can find more than onesatisfying
the appropriate conditions) which does the greatest justice to
our PIs about HOS,. It is with this step that we complete the
circle connecting the history and the philosophyof science. The
task of the historian of science, so conceived, is to write an
account (HOS,) of episodes in the history of science (HOS ,)
utilizing as his criteria of narrative selection and weighting those
norms contained in that philosophical model which is most
nearly adequate to representing PI. To do any less than this, to
utilize a half-conscious or less than adequate model of science
for writing the history of science, is as intellectually irrespon-
sible as deliberately ignoring the evidence.
it
Manyhistorians will doubtless agree that this is the ideal; if
is rarely achieved that is primarily because the models offered
up by philosophers seem to be even less adequate than the
historian’s own half-articulated views about the norms of
scientific evaluation. But, inductive evidence to the contrary
notwithstanding, the historian should not assume that every
philosophical model of rationality is incapable of illuminating
history.

Normsin Historical Explanation

Thus far, we have spoken only of the way in which philo-
sophical beliefs about science influence the historian’s decisions
about what factors to include in his narrative accounts. But
 166 HISTORY AND PHILOSOPHY OF SCIENC}

there is a second, deeper level at which philosophical o:

normative judgments ineluctably enter into HOS,—atthe leve
of historical understanding and historical explanations. Though
by no means the only goal of HOS,, one ofits primary functions
is to explain why various experiments, theories, and research
traditions were accepted, rejected or modified in the ways that
they were. Any serious study in the history of scientific ideas will
be replete with explanations of such factors. Normative evalua-
tions are crucially involved in all such explanations—not as
explicit premises, but as their ground. Consider a very typical
example:
Q,: Why did Newton reject Descartes’ vortex theory of
planetary motion?
A,: Because Newton correctly judged that Descartes’ theory
was grossly incompatible with data about the velocities
and positions ofthe planets.
Clearly, the answer is meant to explain Newton’s rejection of the
vorticular hypothesis. But suppose we go a step further and ask:
Q,: Why should Newton reject a theory that is grossly in-
compatible with the data?
The question itself seems peculiar; it does so because historia
ns
take it for granted that it was reasonable in Newton’s time to
insist that theories be compatible with the data, and thatif one
can show that someone’s action was reasonable (under the
circumstances), then there is nothing left to explain—our
explanatory task is finished. Questions like Q, seem super-
fluous. The history of science (HOS,) abounds with such cases:
the historian explains whya scientist accepted a certain idea by
showing that the scientist deduced it from a prior belief; he
explains why a scientist performed an experiment by showing
that it would test a theory the scientist was considering.
In all such cases, we are implicitly relying on a conception of,
“what it would be reasonable to do in the circumstances.” To
see that this is what is involved, consider a perverse ‘“‘explana-
tion” along the following lines:
Q,: Why did Jones accept the evolutionary hypothesis?
A,: Because all the evidence was against it.
Clearly, something is wrong here. In fact, the answer might be
true. If, for instance, we also know that Jones was a resolute
HISTORY AND PHILOSOPHY OF SCIENCE 167

iconoclast who always denied the evidence of his senses, then

the explanation would become convincing (although, of course,
we might still want to know what caused Jones’ iconoclasm).
But asit stands, A, carries no explanatory force. It fails to do so
because the reason it offers for accepting evolutionary theory
seems to be no legitimate reason at all. If, on the other hand,
our answer had been:
A}: Because all the evidence supported it, we would be
reasonably content with the answer (provided, of course, there
washistorical evidence for A}).
The point is that the historian’s explanations continually
invoke canons of rationality and plausibility, and thereby
presuppose a huge amount of normative machinery. And here,
as with norms ofselection, the historian should see to it that the
normsof rationality he invokes are the best ones available.
Other vital dimensions of historical! research similarly require
the use of norms about rational belief and rational action. To
an extent rarely appreciated by nonhistorians (who often
imagine that the historian is a mere reporter of events), studying
the history of ideas—scientific or otherwise—involves a great
deal of creative imagination. Scientists rarely leave full accounts
of how they came to make their discoveries; even when they do,
such accounts are often unreliable, because constructed long
after the fact. The task confronting the historian is often that of
conjecturally recreating lines of argument and influence which
lay behind the conclusions which a scientist explicitly pro-
pounds. This task of reconstrution is utterly impossible unless
the historian has a very subtle sense of what kinds of arguments
would be plausible in a given situation. Thus here, as with
narration and explanation, the historian’s task requires that he
possesses a theory (implicitly or explicitly) about rational belief
and rational action.

Rational Appraisal and “‘Rational Reconstruction”

What has prevented many historians from seeing the force of
these arguments is a fear that subscription by them to any
contemporary model of rationality will lead to the anachronistic
importation into the past of criteria of rational choice which are
 168 HISTORY AND PHILOSOPHY OF SCIENC
Y
not relevant to the historical circumstances.'
° Precisely becaus
he knows that the norms ofrational! evalu
ation change througt
time, the historian worries about the appro
priateness of trans.
posing our contemporary philosophical insig
hts—assuming thay
sound ones could be found—into an epoch
and a culture tc
which they are foreign. He has a right
to insist that any theory
of norms, if it is to be applied histor
ically, must take into
account the fact that previous scientists had
norms of their own
{often different from ours) which canno
t be ignored in explain-
ing their cognitive stances with respect to
the theories of the
day. Because no philosophical model of
rationality has made
any concessions to the norms of the past,
the historian has been
understandably loathe to utilize such model
s.
Indeed, perhaps the chief stumbling block
to the historian’s
admission of the relevance of philosophy
to HOS, has been the
flagrant disregard for HOS, exhibited
by many of those very
philosophers (especially Lakatos, Feyerabend
, and Agassi) who
have argued most loudly for the dependen
ce of HOS, upon
philosophy.'’ This disregard extends, not
merely to their misuse
of historical data, but is deeply grounded
in their convictions
about the aims of a philosophically based
history of science,
convictions which sometimes subordinate
historical veracity to
the desire to score philosophical points.
These issues probably emerge most clearl
y in Lakatos’
“theory of rational reconstruction,” itself
a theory about the
role of philosophy of science in writi
ng HOS,.'® Lakatos
purports “‘to explain how the historiography
of science should
learn from the philosophy of science.’
The rational recon-
structions of the past, which Lakatos urges
the philosopher to
undertake, bear a very curious and ambig
uous relation to the
actual episodes of which they are the purpo
rted reconstruction.
As Lakatos insists, the process of preparing
the ‘‘internal’’
history or rational reconstruction of a histor
ical episode is not
really an empirical task at all. One “inve
nts” or “radically
improves”’ on the actual historical record
in order to ‘‘rationally
reconstruct”’ it.2? In this rational reconstruction, one tells
history as it ought to have happened. The
actual beliefs of the
historical agents whose names figure in the
story are ignored or
often deliberately distorted. Lakatos here
is not making the
HISTORY AND PHILOSOPHY OF SCIENCE 169

point that the historian is inevitably selective in the data which

he mentions. He is rather making the very different claim that
the “‘rational historian’’ should construct a priori an account of
how a particular episode should have occurred. There need be
no resemblance whatever between the “‘internal’’ account so
constructed and the actual exigencies of the case under
examination.”?
If this sounds extreme, one of Lakatos’ examples makesclear
just how far he is prepared to move away from the historical
record. When discussing Bohr’s theory of the electron, for
instance, Lakatos points out that Bohr had not, by 1913, even
conceived the idea of electron spin. ‘‘Nonetheless,’’ insists
Lakatos, ‘“‘the historian, describing with hindsight the Bohrian
programme, should include electron spin in it, since electron
spin fits naturally in the original outline of the programme.
Bohr might have referred to it in 1913.’’?? On this criterion,
anything whatever that a historical figure might have said (i.e.,
presumably anything that is consistent with his ‘research
programme’’) can attributed by the historian to that figure. The
honest Lakatosian historian must, of course, “indicate in the
footnotes how actual history “misbehaved’,’’?? but the recon-
struction itself is by no means limited to the actual beliefs of
historical agents. Indeed, the liberties which the rational
reconstructionist is permitted go well beyond filling in beliefs
which are consistent with a thinker’s research programme. He
may often, too, ignore or even repudiate the standards of
rationality of a historical figure if he finds them uncongenial.
In discussing the work of the chemist Prout, for example,
Lakatos urges the historian to ignore one of Prout’s basic beliefs
about the experimental well-foundedness of his hypothesis
about elemental composition.”
Once an episode has been so re-cast by the rational
reconstructionist, he proceeds to appraise its rationality, accord-
ing to an appropriate model of rational choice. Whatever the
outcomeof that appraisal, however, the historical episode itself
remains untouched and unexplained—except to the extent of its
faithfulness to the a priori reconstruction (an isomorphism that
will, in the nature of the case, scarcely ever exist except in
limited fashion).?° .
 170 HISTORY AND PHILOSOPHY OF SCIENCE

Lakatos defends this theory of rational reconstruction by

arguing that “‘history without some bias is impossible.'?* There
is surely a difference, however, between having a theoretical
bias (i.e., selecting and interpreting historical events “in a
normative way’’’’) and consciously and deliberately falsifying
the historical record. Lakatos nowhere establishes the necessity
(or the desirability) of making a reconstruction of the past
which involves an intentional warping of the historical record.
Indeed, the fact that Lakatos assumes the possibility of
comparing a “reconstruction” of an episode with its “actual
history”’* shows that Lakatos himself believes that history does
not have to be ‘‘fabricated”’ to be understood.
I want to dissociate my own model of scientific rationality as
vigorously as possible from those of Lakatos and the other
rational reconstructionists. Like them, I believe that the
appraisal of the rationality of historical episodes is an essential
task for the historian of scientific ideas. But there the similarity
ends.’* Unlike the rational reconstructionist, I insist that it
must be actual episodes, not some figment of our imagination,
whose rationality we assess. Unlike them, I argue that the
actual beliefs of historical agents, and the canons of rational
belief of their epoch, must be scrupulously attended to. In
contrast to the reconstructionists, I object to the invention of
historical figures and the fabrication of historical beliefs in
order to score philosophical points or to teach philosophical
lessons.*° If the philosopher would learn something from
history, he must make himself a servant to it—at least to the
extent of dealing with actual cases and actual beliefs. And if the
historian is to find any philosophical model relevant for his
own work, that model must allow for the evolving character of
rationality itself. I have already claimed that the model
developed in Part One can succeed in doing just that.
 Chapter Six
The History of Ideas
Though the gap seems small, there is
no chasm that more needs bridging than
that between the historian of ideas and
the historian of science. T. S. KUHN (1968), p. 78

The work of too many professional

historians is diminished by an anti-
rational obsession—by an intense
prejudice against method, logic and
science. D. FISCHER (1970), p. xxi

The history of ideas or, as it is often called, intellectual history,

is among the oldest genres of historical writing. The presump-
tions which motivate it, namely, that what our ancestors
thought is as interesting as what they did, that their ideas were
as important as their wars and rulers, have their roots deep in
antiquity; indeed, many of the earliest extant historical writings
are concerned with what we would nowcall the history of ideas.
In recent times, particularly in the nineteenth century, studies
of the history of thought, cultural history, the evolution of ideas
and doctrines formed a large share of the historical literature.
In our own time, by contrast, the history of ideas is regarded in
many quarters as passé and irrelevant, as a discipline with
outmoded presuppositions and outrageous ambitions. Many

171
 172
HISTORY OF IDEAS
historians see intellectual history as an anach
ronistic excre-
scence on the scholarly and ideological
integrity of their field,
Because the bulk of this chapter (and
in certain respects the
entire essay) is an effort to stress the
importance of the history
of ideas—at least of a certain type
of history of ideas—it is
probably wise to begin by surveying
some of the reasons for its
present disrepute.
There are several, frequently cited
complaints directed against
intellectual history:
1. that it is “elitist’: not because most
people do not think,
but rather because we only have histor
ical records about the
“thoughts” of a tiny fraction of
the petsons in any society
(namely, those who were both literate
and fairly prolific).
2. that it assumes ideas have
an independent reality. it is, so
the critics stress, “people who have
ideas.” People live in
societies with certain economic, politi
cal and social character-
istics which condition or even cause
their ideas. Intellectual
history, to the extent that it abstr
acts ideas from their broader
social surroundings, distorts the
historical record.
3. that ideas are a far less potent
source of change than the
underlying socio-economic "realities":
on this view ideas (in
form of “ideologies’’) merely mirro
r the material condition of
the society and serve only as token
s for the class conflict
between the warring factions. To focus
on the evolution of ideas
is to misplace the genuine causes of
historical change.
4. that the history of ideas, becau
se it is “impressionistic ”’
and not readily quantifiable, is
out of Step with the move
towards “‘scientific”’ history.
I shall postpone any direct comm
ent on these well-known
quibbles with intellectual history.
It was important to state
them early, however, in order
to underscore the differences
between these standard criticisms
of the history of ideas and
those reservations which I shal]
be voicing. All the above are
objections in principle to any type
of intellectual history; they
seek to cast doubts on any effort
to study the evolution of ideas
(except within a broader socio-econ
omic context). My own
reservations, which I shall discuss
at length, are qualms about
the assumptions currently underlying
certain types of intellec-
tual history. In brief, I shall argue
that much intellectual
history, as currently practiced,
is too discipline-oriented in its

ee
HISTORY OF IDEAS 173

approach, too insensitive to the historical dynamics of intellec-

tual problems, and more preoccupied with chronology and
exegesis than with explanation—which should be its central
object. But all these defects are remediable. My claim will be
that there are ways of doing the history of ideas which are not
only intellectually well-founded, but highly relevant as well.
After describing what I believe to be, at least in its essentials,
an adequate model for a historiography of ideas, I shal! return
to points (1) through (4) above in order to see how cogent they
are in the face of a more complex conception of intellectual
history.

Disciplinary Autonomy and the History of Ideas

Without any doubt, one of the mostrestrictive features of
much intellectual history is its discipline-bound manner of
presentation. We have historians of philosophy, historians of
science, historians of theology, each generally assuming that the
ideas with which they are concerned have no crucial cross-
disciplinary dependencies. The tendency toward specialization
extends even into single disciptes. Philosophers write histories of
ethics, histories of epistemology, and histories of logic. Scien-
tists write histories of analytic chemistry, histories of physical
optics, even histories of x-ray crystallography. Theologians give
us histories of eschatology, histories of natural theology, and
histories of eucharistic doctrine. There is nothing surprising in
all this. Practitioners of a contemporary speciality have a
natural, perhaps inevitable, curiosity about their predecessors.
Noris there necessarily anything suspect about the high degree
of specialization which we see in much contemporary writing
about intellectual history. But in practice, if not in theory, this
manifold division of labor between the various disciplines has
exerted a deleterious effect on the writing of intellectual history,
because the assumption of (relative) disciplinary autonomy has
tended to blind many historians of ideas to the single most
striking fact about the history of thought, its integrative
character.
Up to, and even including, our own time, leading intellec-
tuals have been concerned simultaneously with a broad spec-
trum of problems and issues, ranging from the highly specific
 174 HISTORY OF IDEAS
and technical to the very general and abstract. As I showed in
Part One, rational appraisal has generally been construed by
our predecessors as a process of finding maximally adequate
solutions to a divergent range of compelling intellectual prob-
lems, problems which, moreover, occur in several diverse
disciplines.’ The evolution of ideas, and the problems to which
those ideas provide solutions, is necessarily an interdisciplinary
process. Historians of ideas, scientific or nonscientific, ignore
this integrative tendency at their peril.
Yet they do ignore it. The vast majority of contemporary
histories of science and histories of philosophy pay no more than
lip service to the mutual interpenetration of ‘‘scientific’ and
“philosophical” doctrines and problems. Equally, one is hard
pressed to find any history of social or political theory which is
i fully alive to the high degree of historical interaction between
the “soft’’ and the “hard”’ sciences.
If the nature of the interaction between the various disciplines
were just a kind of “‘spill-over”’ effect, whereby ideas from one
domain only occasionally penetrated into another, the tendency
to write disciplinary histories of ideas would be excusable. But
the fact of the matter is (if we extrapolate from the best recent
scholarship) that there is—or at least has been—a continuous
process of interpenetration and legitimation going on between
the intellectual structures of the various disciplines. Thus, the
problems of seventeenth- and eighteenth-century metaphysics
were posed by the new ‘“‘mechanistic science” and make no
sense except when seen against that background. The problems
of nineteenth-century social theory and aesthetics were the
by-product of a confluence of scientific, technological, and
epistemological developments which provided both the model
for, and the legitimation of, a succession of theories about social
structure and aesthetic perception.
What has led otherwise subtle and sophisticated scholars to
ignore so many of these interconnections? Why, more specific-
ally, has the chasm “‘between the historian of ideas and the
historian of science” (to which Kuhn refers in the passage at
the head of this chapter) developed? The core of the answer is
provided, ironically, in Kuhn's own work. Although bemoaning
the failure of historians to see the connections betweenscientific
HISTORY OF IDEAS 175

and nonscientific ideas, Kuhn himself articulates a now well-

known model of scientific development which, in its essentials,
denies the existence of any significant degree of interaction. It is
Kuhn, for instance, who writes that: “the practitioners of a
mature science are effectively insulated from the cultural
milieu in which they live their extra-professional lives.”? It is
Kuhn, again, who insists that: “the development of an indi-
vidual technical speciality can be understood without going
beyond the literature of that speciality and a few of its near
neighbors.’”?
Such tensions between the historian’s aspirations and his
convictions are so familiar as to be commonplace.‘ While
insisting that we ought to look for intellectual connections
between disciplines, when it comes to the discipline he knows
best, the historian often proceeds to write about its history as if
it were almost completely isolated from everything else! He does
not seem to realize that so long as we retain a model of strict
disciplinary autonomy, then the realization of an interdisci-
plinary history of ideas will forever elude us.

Ideas and Their Problem Contexts

A related and persistent failing of much scholarship in the
history of ideas is the tendency to ignore the problems which
have motivated the construction of great intellectual systems of
the past. Too often, the historian of ideas sees his function
primarily as that of setting out the systematic interconnections
between the beliefs of a thinker or group of thinkers on a
closely related family of issues; this is a subtle task, which
involves revealing the threads of reasoning whereby our prede-
cessors came to hold the beliefs they did. But this is to tell only
half the story, even when done well. Systems of thought are not
merely logical links between propositions. They are that, but
they are also attempts to resolve what are perceived as
important problems. To write about the history of conceptual
systems without ceaselessly identifying the problems which
motivated those systems is drastically to misconstrue the nature
of cognitive activity.° To give, say, a detailed exegesis of Locke’s
empiricism or Engel’s dialectical materialism without carefully
176 HISTORY OF IDEAS

identifying the empirical and conceptual problems that those

doctrines were designed to resolve is not unlike playing one of
those parlour games in which one is given an answer (often a
bizarre one), without knowing the question to which it is an
answer! One can only understand a system of ideas when one
knows, in detail, the problems to which it was addressed.
If it seems difficult to imagine that this commonplace is more
often ignored than observed, consider a pair of examples. For
several hundred years, historians of ideas have been writing
about Cartesian philosophy. Literally hundreds of books and
thousands of articles have been written about Cartesian dual-
ism, about the method of doubt tn Descartes, about the cogito
argument, and about Descartes’ borrowings from his prede-
cessors. Yet it is only in the last generation that scholars, such
as Gilson and Popkin,® have begun to shed any useful light on
Descartes’ problem situation and orientation. Only now can we
begin to see why Descartes’ philosophy sometimes takes those
curious twists and turns that madeso little sense when scholars
were insensitive to the actual problems with which that
philosophy grappled.
A second example is provided by the vast exegetical literature
dealing with John Stuart Mill’s influential views on epistemol-
ogy, logic, and political philosophy. As extensive asit is, westill
have almost no sense of Mill’s problem situation. Why, for
instance, did he devote so much energy to reviving the methods
of enumerative and eliminative induction? What were the
specific problems within the social sciences for which his well-
known “historical method’? was meant to provide a solution?
What was his motivation for classifying the sciences in the
manner he did? Much of the most careful scholarship on Mill
skirts these (and other similar) questions about the problems
Mill was tackling.
Even when intellectual historians recognize that systems of
thought have their roots in problems, they sometimes tend to
adopt an ossified and unilluminating notion of what a problem
is. Showing less sensitivity to historical process and conceptual
nuance than one might expect, many scholars write as if
problems have an unchanging identity through time, a peren-
nial character.’ How often does one see references in the history
 177
HISTORY OF IDEAS
the problem of
of philosophy to the problem of substance,
em of free will, the
induction, the mind-body problem, the probl
of science speak of
problem of universals? Similarly, historians
life, or the problem
the problem of combustion, the problem of
ems have not remained
of free fall. In each case, these probl
em of-in ducti on was very
static through time. Hume’s probl
are very differ ent from our
different from Mill’s and both
address the
version of it. There are times when twothinkers do
be shown rather
same problem or set of problems; but that must
To assum e probl em identi ty through
than idly presupposed.
of ideas, the first step on the road to
time is, for the historian
historical record,
what may be a most serious falsification of the
of a thinker’s
for when we misconceive the precise character
e of the
problems, we are apt to misunderstand the natur
solutions he proposes.
insistence that
Many historians are quite forthright in their
rd Nelson, for
intellectual problems are unchanging. Leona
be impos sible to
instance, goes so far as to claim that it would
assum ing the identi ty of
write the history of philosophy without
analys is, soluti ons may
problems through time. On Nelson’s
Nelso n’s appro ach borde rs on
change, but problems cannot.’
Nelso n would have it—th at
the perverse. To imagine—as
teent h-cen tury physic s, or the recent
medieval theology, or seven
new problems for
emergenceof the social sciences produced no
ve repudiation of
the philosophical tradition requires a massi
much ofthe best scholarship of the last 150 years.
of a problem-
The emphasis I have given to the importance
s Collingwood's
oriented approach to intellectual history echoe
consta ntly be aware
insistence that the historian of ideas must
ical figure s sough t to
of the problems and questions which histor
d’s appro ach makes
solve.'® Unfortunately, however, Collingwoo
aphy because of his
nonsense of a problem-solving historiogr
ems and solutions. For in-
idiosyncratic conception of probl
to the view that the only way
stance, Collingwood is committed
what probl ems a think er was trying
the historian can determine
probl ems that think er actual ly solved.
to solve is by seeing what
the same passage
As Collingwood says of Leibniz: “One and
of what the problem
states his solution and serves as evidence
em is proof that he
was. The fact that we can identify his probl
 178
HISTORY OF IDEAS

has solved it; for we only know what the problem is by arguing
back from the solution.’"! On such an analysis,
we could never
say that a thinker failed to solve a problem because the
only
criterion Collingwood allows for attributing a proble
m to a
thinker is that he has solved it. Such a Panglos
ian view of
intellectual activity—entailing as it does that the only
problems
we ever attempt to solve are those which we actuall
y solve—
makes it impossible for the historian either to criticiz
e the past
or to explain its vicissitudes (at least in so far as the
latter
depend upon thefailure of certain intellectual systems
to solve
the problems which they address). Collingwood
failed to
" recognize that the historian can often find strong
evidential
grounds forattributing a problem to a thinker, even
when that
thinker fails to solve the problem which he has set
for himself.

The Aims and Tools of Intellectual! History

Chronology, exegesis, and explanation. Another
of the core
problems bedeviling the historiography of ideas
is a crucial
unclarity about the very aims of the enterprise. As
construed by
many of its practitioners, the aim of intellectual
history is
neither more norless than exegesis, and its basic
method is the
classical one of explication des textes. Seen in
this light, the
ptimary task of the historian of ideas is to get clear
about what
people in the past have said and (in so far as he
can get at it)
what they have thought. One examines, for exampl
e, Newton’s
views on time or the Marxist theory of aliena
tion and, in
essence, tries to set out the appropriate doctrine in
a clearer and
more perspicacious way than its original propon
ent(s) did.
Intellectual history, done this way, amoun
ts to an elaborate
form of paraphrase and précis. The historian
sees his task as
that of rehearsing the arguments he finds in the
classical texts,
perhapsfilling in from time to time certain
presuppositions
which were not fully or carefully formulated
in the original
sources.
This sort of intellectual history I shall call exeget
ical history,
precisely because its aim is straightforwardly explica
tive. Exe-
getical history aimsat providing a natural histor
y of the mind as
it evolves through time. Like every other form
ofnaturalhistory,
HISTORY OF IDEAS 179

it is primarily descriptive in its ambitions. It seeks to record the

temporal sequence of beliefs, in much the same fashion that
descriptive geology aspires to record the sequence of changes on
the face of the earth. But there is a very different type of
intellectual history to which we might aspire, namely, explana-
tory history.
Our aim here would be not merely to rehearse what ‘‘great
minds’ have said but also to explain why they have said it.
Clearly, exegetical history of ideas stands to explanatory history
in the samerelation as chronology stands to general history, or
as any descriptive science stands to its explanatory counterpart.
The explanatory scientist must be clear about the temporal
succession of events, but he aspires to more than mere
chronology. Indeed, he seeks to exhibit the reasons and causes
that lie behind and explain the temporal sequences. In exactly
the same way, the historian of ideas—if he intends to be more
than a chronologist—must be ready to go beyond exegetical
history. He must be prepared to ask, and to answer, questions
like: Why did a certain thinker at a certain time espouse certain
beliefs? Why was a given system of ideas modified when and
where it was? How did one intellectual tradition or movement
grow out of another?”
Unfortunately (and this may do much to explain the uneasi-
ness many people have about it), scholarship in the field of
intellectual history is still largely exegetical and not yet
explanatory, neither in fact nor even in aspiration. The
historiography of philosophy, almost certainly the most back-
ward in this respect, provides some vivid examples:
For instance, scholars are widely agreed that the emergence of
a hypothetico-deductive model of science was a very important
characteristic of nineteenth-century logic and epistemology.
Numerous exegetical studies have been written about the views of
Kant, Whewell, Mill, Peirce, and others on that new philosophi-
cal model of science. Yet virtually no one has asked why it is the
case that most nineteenth-century philosophers, unlike their
eighteenth-century predecessors, thought it appropriate or
important to stress the speculative nature of science. We have,
as yet, not even the outlines of an explanatory history of
epistemology and inductive logic for this period.*?

ee
 180
HISTORY OF IDEAs
Enlightenment historians of idea
s have long agreed that the
shadows of Bacon and Newton
loom very large over eighteenth.
century thought. Countless
books and articles have been
devoted to tracing out the infl
uence of their ideas in France,
Britain, and Germany during
the period. Yet if one asks why
Bacon and Newton were so muc
h more influent ial than, Say,
Hobbes or Boyle or Malebranc
he, one finds that answers
neither frequently offered nor, are
when offered, cogently formu-
lated. The fact of the domina
nce of Newton and Bacon
eighteenth-century thought has in
been documented ad nauseam;
we have yet to make it a reasoned, or an
explained, fact.
With individuals, as tmauch
as with broader movements
intellectual , most
history remains exegetical and
nonexpla
natory. It is
now well known, for inst
ance, that Newton and Leib
heav ily influenced by Cartesian niz were
philosophy in their formative
years. Yet both, for differen
t reasons, came to repudiat
Cartesian conceptions in thei e
r philosophical maturity. The
chronology ofthis process has
been well documented for som
time. Yet if one asks for a conv e
incing explanation of Newton’s
or of Leibniz’s change of min
d, contemporary scholarship
scarcely taken us beyond the has
sketchy explanations offered
Leibniz and Newton themselves. by
The pervasive explanatory pauc
ity of intellectual history, as
illustrated by these few exam
ples, is presumably more than
accidental. There must be,
one is inclined to conjecture,
something about the current
methods and presuppositions
the history of ideas which acco of
unts for jts explanatory bank-
tuptcy. There are at least two
areas where I am inclined to
locate the difficulties: in the
basic units of analysis hitherto
utilized by historians of ideas;
and in the difficulties that atte
any effort to explain the beli nd
efs of human agents. I shall
with these in turn. deal

Concepts, “unit ideas,"’ and

research traditions. Until re-
cently, the dominant mode of
approach in the history of idea
has involved tracing one or mor s
e related ideas as they evolved
through long stretch oftime.
The concept of space, the idea
the great chain of being, the doct of
rine of habeas corpus, entities
such as these have long bee
n the stock-in-trade, the pri
mary

en
HISTORY OF IDEAS 181

units of analysis, in intellectual history. This is hardly surpris-

ing; what else besides ideas would one expect the history of
ideas to be concerned with? For all its initia] plausibility,
however, there is something profoundly deficient about focus-
sing primarily on the concept, or (as Lovejoy called it) the
“unit idea.”’
For one thing, such an approach tends to ignore the fact that
ideas are interrelated and interconnected. If we want to
understand what someone means by an idea, we must see how
he uses it, how it functions for him in a broader framework of
convictions about the world. In manycases, the very determin-
ation of the meaning of a concept or idea requires us to pene-
trate deeply the belief network of the thinker using the concept.
As recent scholarship has shown, pinning down the meaning of
{to cite a few examples) Newton's conception of matter,
Faraday's conception of force, or Hobbes’ conception of the
state, requires an unpacking of the entire Weétbild of the
thinker in question.
But there are other, even more serious ways in which the
focus on single ideas puts acute obstacles in the way of historical
analysis. As we know, ideas change and evolve. Accounting for
such changes must be oneof the central tasks of the history of
ideas. Such changes can only be explained by looking at the
shifting position of an idea within a broader conceptual network
which is undergoing continuous modification. Hence, to explain
the changes wrought on a particular and specific concept, we
generally must look to a larger unit than the concept itself.
Recent studies have shown, for instance, that the concept of
“natural regularity,’ whose pre-history can be traced to
antiquity, underwent a major alteration during the seventeenth
century. We can begin to understand this change when we see
that it was closely linked to the emergence of voluntarist
theology and that the notion of a law of nature which comports
with a freely acting deity is very different from the kind of
natural regularity which one contemplates for a deterministically
ordered, teleologically structured universe. To attempt to tell
the history of the metaphysical concept of natural order,
without at the same time telling the history of the larger systems
or traditions of thought (embracing both science and theology)

he
182 HISTORY OF IDEAS

within which that concept was embedded, is evidently doomed

to failure.
At a deeper level, the central danger with this approach to
intellectual history is its tendency to blind historians to the
changes wrought upon an idea or concept in the course of its
evolution. To suggest, as Lovejoy does, that both Plato and
Leibniz subscribe to the idea of the great chain of being is to
gloss over the fact that “the chain of being’? means something
radically different for the two thinkers. To assert, as Holton
does, that the “theme’’ of discontinuity is a recurrent one in
human thought probably obscures more issues than it clarifies
since (to take two extremes) Democritean discontinuities are
very unlike the discontinuities postulated by Bohr or Planck.'*
What do we gain by seeing the history of thought as a
counterpoint between such polarities as being and becoming,
activity and passivity, or quantity and quality? Does it really
explain what a thinker is doing to represent him as constructing
a system by dipping into the familiar well of primordial
concepts? My belief is that concepts evolve every bit as much as
problems do and that the presumption of stasis at either level
is tantamount to accepting an outmoded Platonic conception of
the nature of intellectual history.
Recent philosophical, as well as historical, scholarship under-
scores the need to abandon the traditional “vertical” or unit
idea approach to intellectual history. Thinkers like Duhem,
Quine, Hanson, and Feyerabend have cogently argued that it
is entire systems of thought which confront experience.'*
Individual concepts, particular propositions, which are com-
ponents of these larger complexes, do not—indeed cannot—
stand alone, and as a result we generally should not appraise or
evaluate concepts on a piecemeal basis. Because these larger
systems (which I have called “research traditions’’) function at
any given timeas the effective units of acceptance (or rejection),
it follows that the intellectual historian—in so far as he wants to
explain the evolving vicissitudes of belief—must take such
traditions as his fundamental units for historical analysis.'* This,
in turn, requires a more horizontal, less vertical, approach than
one customarily sees in historical scholarship. We must focus
more on smaller time slices, wherein we examine the systematic
HISTORY OF IDEAS 183

interconnections between the concepts in several contemporan-

eous research traditions. If we would learn why Newton
introduces the concept of absolute time, or why Locke modi-
fies the traditional concept of monarchy, we must examine
in detail their own traditions and the traditions of their rivals.
We must be prepared to show how, for ifstance, the introduc-
tion of certain conceptual variations improved the overall
problem-solving capacity of one or another system which
embodies the variation.
There is another dimension to the contrast between “‘vertical”
and “horizontal” history. It is sometimes argued, and even
more widely assumed, that the central function of the history of
ideas is to clarify what some “great thinker’ meant in a given
text. Dilthey’s quest for Verstehen, Collingwood’s preoccupa-
tion with “re-thinking another man’s thoughts’ and “getting
inside his head,’’ as well as Skinner’s concern with identifying
the ‘‘intentions” of great thinkers,'® all speak of the preoccupa-
tion of theorists of intellectual history with faithful exegesis.
Doubtless important, this process is, however, by no means the
most important task of the intellectual historian. The historian
must be at least as concerned with how ideas are received (what
the Germans call Rezeptionsgeschichte) as he is with how they
reached maturity in the head of the thinker who produced them
in the first place. The intentions or the internal thought
processes of a man who generates an idea are largely (and often
completely) irrelevant to explaining how that idea is received in
the appropriate intellectual community. Putting the point
differently, if our primary focus is on the evolution of research
traditions, then we need to pay relatively more attention to the
ways in which the tradition is interpreted and modified by its
defenders and relatively less attention than we have previously
to the ratiocinative processes that produced the tradition in the
first place.
In arguing that research traditions rather than individual
concepts should be the basic unit of historical analysis, | am far
from claiming that the intellectual historian should eschew
ideas and concepts. Myclaim is, rather, that even if (perhaps
especially if) we are interested in single concepts, we must begin
with an analysis of research traditions, for it is the changing
 184
HISTORY OF IDEAs
fortunes of the latter that generall
y serve to explain both the
specific changes in, and the fort
unes of, the former. We must
not be misled by the fact that most
physicists talk about space,
or most political theorists talk abou
t the state, into thinking that
concepts like ‘‘space’” and “the
state” have an_ historical
autonomy about them whic
h allows one to explain their
historical transformations independ
ently of the broader patterns
of belief of which the particular
concept forms only one strand.

Explanations in intellectual hist

ory. If the failure to focus
consistently on the most useful
unit of analysis has caused som
mischief, an even more serious e
problem has been an ambiguity
about the explanatory ambition
s of the history of ideas. In mos
of the explanatory sciences, the t
object of explanation, what is to
be explained, is either an event
(the falling of a stone), a process
(the growth of a plant), or
an action (the bombing of
shima). By and large, the Hiro-
history of ideas is not prim
concerned with explaining any arily
of the above. Its basic data are
beliefs, and the changes and modi
fications of belief. If inteilec-
tual history is to be properly expl
anatory, its aim must be the
explanation of the vicissitudes
of belief and conviction on the
part of historical agents. Merely
documenting what those beliefs
were and how they changed,
which is the aim of exegetical
history, clearly does not give
us explanations. To secure the
latter, we must offer cogent hist
orical arguments which show
why a given belief was form
ulated, accepted, modified,
rejected . It is here that the rub come or
s, however, for there is still
much debate about what can legi
timately count as “explaining a
belief.”
What should an explanans of the
appropriate type look like?
If we subscribe to the usual mod
el of explanation, and apply it
the history of ideas, we migh to
t suggest that any adequate
explanans must contain both som
e universal statements (“‘laws’’)
and some statements ofinitial
conditions. The two sets should
jointly entail a Proposition enun
ciating a belief Situation that we
wish to explain. Accepting this
model for the moment, our
question about explanation with
in intellectual history reduces to
this: What sort s of things, if any, count as
appropriate laws and
initial conditions for the explanat
ion of beliefs?

ee
HISTORY OF IDEAS 185

There are at least two different ways in which we might seek

to answer this question. The first: if we were committed social
(or psychological) determinists, convinced that all beliefs are
caused by (and thus to be explained in terms of) the socio-
economic position or psychological state of believers, then we
should require laws relating a specific type of social situation,
x, to a specific type of belief, a@ (namely, those occurring in the
explanandum). Our initial conditions would (hopefully) assert
that a particular believer z was in the relevant situation x. We
could then deduce that (and thereby explain why) z accepted
belief a. This type of explanation is rarely offered by historians
of ideas; not surprisingly, for most of them do not subscribe to
situational determinism of belief and thus are not willing to
accept the truth of the “laws” invoked by this type of explana-
tion. Because social accounts of belief are not a widely accepted
mode of explanation, and because much of chapter seven is
concerned with the ramifications of such an approach, I shail
not discussit further here.
The second: far more frequently invoked than the above are
what we might call rational explanations of belief. Here we
assume, implicitly or explicitly, certain rules or laws of rational
belief and then apply them to particular belief situations. A
historian might say (for instance) that Bacon rejected a belief in
superstitious magic because he could see no evidence for it
(assuming as his general explanatory law that ‘rational agents
only accept beliefs when they have positive evidence for them”).
Because this mode of explanation is so crucial for the very
possibility of an explanatory history of ideas, it is worth examin-
ing its structure in more detail. Consider the following schema:
All rational agents in situation type, a, will accept (or
reject, or modify) belief type, b. (i)
Smith was a rational agent. (2)
Smith wasin situation a, (i.e., an a type situation). (3)
Smith accepted (or rejected, or modified) belief b,. (4)
Statements (3) and (4) of schema (1) are presumably unprob-
lematic; the evidence should establish their truth status unam-
biguously. Statement (2) is only slightly more difficult; suffi-
ciently careful biographical studies can establish with high
likelihood whether a given historical figure generally was or was

Da
 186
HISTORY OF IDEAS
not rational in his appraisal of beliefs in this field.
By contrast,
(1) is the problem case, for how do we discover
laws or principles
of type 1?
The question cannot be avoided or postponed,
for a plausible
answer to it is a necessary prerequisite to any histor
y (as opposed
to a chronology) of ideas. The general laws sought will
belong, of
course, to the theory of rational belief; forit is only
such theories
that can provide general principles of the type repres
ented by
statement (1). The applicability of such
theories of rational
belief, in turn, crucially depends on what we
pack into our
characterization of the believer’s “situation type.””
As I pointed
i out in Part One, most theories of ration
al belief fail to be
of much use to the historian because they deal
with a very
impoverished rangeof situation types.
On an inductivist theory of rationality, for examp
le, the only
situation types discussed would be those in which
a belief was
assigned very high (or a very low) probability on
the strength of
the known empirical evidence. But, as we have
seen, this is of
little aid to either the historian of science or
the general
intellectual historian, because virtually no actual
historical cases
of belief exemplify the strict conditions demanded
by inductivist
models. In deductivist theories of rationality, on the
other hand,
the only allowable situation types would be
those in which
relations of entailment obtained between the
belief to be ex-
plained and other beliefs of the agent. While
such cases do
certainly occur in the history of thought (and
to this extent
deductivist models of rationality have more to offer
the intellec-
tual historian than inductivist ones), they
still constitute only a
tiny proportionof the belief situations which he
seeks to explain.
A variant of the deductivist model often invok
ed by intellec-
tual historians is Collingwood’s theory of presup
position. The
idea hereis to get at those core concepts which
lie behind, asit
were, the explicit beliefs to which a thinker
subscribes. The
problem is that presuppositional analysis (at least
in its Colling-
woodian form) is, at its core, purely deductivist.
It can explain
those beliefs of a historical figure which follow
strictly from his
alleged presuppositions; but it can explain neithe
r the presuppo-
Sitions themselves nor any beliefs which fail
to be deductive
consequences of those presuppositions. Still worse,
presupposi-
tional history offers no machineryfor discussing
why historical
HISTORY OF IDEAS 187

agents accepted one set of presuppositions rather than another.

Thus it leaves unexplained that very aspect of history which it
takes to be the most important.
Apart from the limitations already discussed, these models of
rational belief suffer from yet another liability when applied to
the history of ideas: namely, their insensitivity to (tantamount to
a denial of) the degree to which specific canons of rationality are
time-dependent. A mode of argument which one epoch, or
“school of thought,”’ views as entirely legitimate and reasonable
may be viewed by another era or anotherintellectual tradition as
ill-founded and obscurantist. Neither inductivist nor deductivist
theories of rationality leaves the historian any scope for attend-
ing to those subtle, temporal shifts in standards of argumenta-
tion which continuously confront him in his research.
Whatintellectual history stands most in need of, in my view, is
a theory of rational belief which goes beyond the restrictive
limitations of the inductivist and deductivist models.
The problem-solving model of rationality discussed earlier
represents a step in that direction. It is sensitive to shifting,
local canons of rational belief; it does allow for a comparative,
rational appraisal of presuppositions; it does not limit rational
belief to those cases in which there is a rigid deductive or
inductive propositional link.
Such grandiose claims may sound fine in the abstract, but
how in practice does the problem-solving model shed any light
on specific cases? The method of application of the model is
relatively straightforward. One begins by identifying the pool of
available explanatory systems (i.e., research traditions) in any
given epoch and intellectual community. One then determines
for each of these research traditions how progressive they were
(i.e., how effective they were at maximizing solved problems and
at minimizing anomalous and conceptual ones). This analysis
will allow the historian to construct a profile on the progress of
each of the available options.'’ Independently of these profiles,
one has a number of laws or general principles of rationality.
Among them would be principles such as: (1) all rational agents
will prefer a more effective research tradition to a less effective
one; and (2) all rational agents, in modifying a research tradi-
tion, will prefer more progressive to less progressive modifica-
tions ofit.
me
A PF

188
HISTORY OF IDEAS
Such principles, when conjoine
d with the progress orrationa]-
ity profiles of each of the
available research traditions, wil|
allow one to explain many deve
lopments within the history of
thought which have thus far elud
ed explanation. Such, at any
rate, is the claim being made for
the problem-solving model.
It might be held that, in orde
r to give historical explanations
of the kind Proposed here,
we have no need for rational
normative appraisals whatever ,
. It could be said thatit is not
historian’s task to determine the
whether some belief was rational
only to show that some thinker ,
thought it to be so. Suppose, for
instance, we wish to explain why New
ton advocate d action-at-a-
distance forces to explain grav
itation. Is it not enough mere
recite Newton’s stated reasons ly to
for introducing the concept,
adding perhaps that he regarded
these as sufficient reasons for
using the concept? On this anal
ysis, there is no place for the
historian to ask the normative
question whether, by the then
appropriate canons ofscientific
belief, Newton was right in
judging action-at-a-distance
to be well conceived.
To locate the flaw in this approach
, we mayconsider a second,
parallel example. Suppose we
wish to explain why a certain
“special creationist” believes ther
e was a universal deluge during
the time of Noah. Suppose, furt
her, that we can show that his
only reason forthis belief is that
it accords with Scriptures and
since he takes Scriptural conc
ordance as a sign of truth, thin
his belief is well-founded. Conf ks
ronted with such an “explana
tion,”” we would feel that the -
historian’s job was only half
completed, for we now want
to know why this creationist
subscribed to such a peculiar
theory of truth. Our curiosity
only whetted, not Satisfied, by is
being told that someone accepted
a belief for which there were
only ‘‘bad”’ reasons, not “good’’
ones.
By contrast, if we can show that
a thinker accepted a certain
belief which was really the best
available in the Situation, then
we feel that our explanatory task
is over. Implicit in this way of
looking at the matter is the assu
mption that when a thinker does
what it is rational to do, we
need inquire no further into
causes of his the
action;whereas, when he does wha
irrational—evenif he believes t is in fact
it to be rational—we require som
e

|
HISTORY OF IDEAS 189

further explanation. This assumption thus functions in the

realm of human behaviorvery like the principle of inertia within
mechanics. In both cases, the principles provide a characteriza-
tion of what we regard as “‘normal behavior.” A body movingat
constant velocity and a man behaving rationally are both
“expected states,’’ which require no further causal analysis. It is
only when bodies change velocity or when men actirrationally
that we require an explanation of these deviations from the
expected pattern. Of course, this proposal—that rational beha-
vior is the rule rather than the exception—is open to debate, but
as we shall see in chapter seven, it is preferable to the
alternatives. Precisely becauseit is preferable, normative evalua-
tions—as opposed to purely descriptive ones—mustplay a role in
historical explanations, for those evaluations tell us when our
explanatory task is at an end.

Problem Solving and Nonscientific Research Traditions

It might be thought that the problem-solving model articu-
lated in Part One, although applicable to the history of scientific
ideas, is of only very limited use in those areas of intellectual
history which deal with nonscientific domains. While conceptual
problems obviously occur in all fields of inquiry, empirical
problems seem to occur far less globally. It has, after all, been
argued at length by many scholars that the sciences alone are
empirical disciplines, from which it would follow that science
alone has what I have called empirical problems and that there
would be no counterpart of empirical problem solving in the
nonscientific disciplines. If it were true that the natural and
social sciences exhausted the range of empirical problems(as,
for instance, the positivists maintained) then one would have
grave doubts aboutthe suitability of a problem-solving model for
dealing with general intellectual history. But to imagine that
“‘nonscientific”’ disciplines traditionally have had no significant
empirical element is a gross historical travesty. Consider only a
few examples:
1. Metaphysics is frequently cited (especially by professional
anti-metaphysicians) as an ideal example of a discipline with no
ar?

190 HISTORY OF IDEAS

empirical content. But there are, and classically have been, a

host of empirical problems which metaphysical systems have
sought to resolve. For instance, most objects are seen in our
daily experience to endure through time. One of the core
empirical problems of metaphysics has been explaining what
properties of being can explain the seeming endurance of
objects. Similarly, most changes which we experience in the
world seem to be causally linked to other changes. An explora-
tion of the causal nexus has been a persistent problem for
metaphysics. Even those metaphysical systems (like occasional-
ism) which deny the ultimate reality of a causal connection
between events, still must explain an empirical problem; to wit,
whyit is that the world seemsto be causally interconnected. It is
doubtlessly true that the specificity of the empirical problems
which confront, say, the chemist and the ontologist are very
different; but the difference is one of degree, not kind. The
metaphysician and the historian of metaphysics, every bit as
much as the chemist and his historian, must attend to the
empirical problems ofthefield.
2. Theology, like metaphysics, is often alleged to be empiric-
ally transcendent and thus devoid of empirical problems. But
few traditional theologians or historians of theology would
subscribe to such a view. For instance, the “problem of evil” is
at its core an empirical problem par excellence: how can one
maintain one’s belief in a benevolent, omnipotent deity in the
face of all of the death, disease, and natural disasters which are
a daily element of our experience? Many theological doctrines
have been devised largely to deal with this seeming empirical
anomaly. Judeo-Christian theology is, more than most, replete
with a large body of similar empirical problems. On onelevel,
that theology makescertain historical claims about the existence
of persons and the occurrence of events. At another level,
Judeo-Christian theology makes claims about the experiential
effects of “‘true belief’? on believers. These claims are, in
principle, testable within the realm of experience.’ If false,
these claims are confronted by a large base of empirical
anomalies which any adequate (i.e., progressive) theology must
either solve, or suffer the cognitive consequences of failing to
solve. If true, then they constitute solved empirical problems.
 191
HISTORY OF IDEAS
ence of empir-
Similar remarks could be made about the exist
n inquiry. Even in
ical problems in every other branch of huma
logic and mathematics,
the so-called formal sciences, such as
ical probl ems, they exist
where one might least anticipate empir
natin g studi es on the history
in large numbers—as Lakatos’ fasci
of mathematics amply demonstr ate. "
l to nonscientific
The applicability of a problem-solving mode
not only for writi ng the history of
disciplines has implications
appra ising their cogni tive status. It
such disciples, but also for
are progressive and
is frequently claimed that the sciences alone
ry exhibit changes of
cumulative, while other areas of inqui
ngfully described as
fashion and style which cannot be meani
put differently; it is
progressive.’*® The contrast is sometimes their
discover when
sometimes said that the sciences can
disciplines cannot; it
assumptions are wrong, but the humanistic
ces are ‘‘self-corrective,’’ but
is frequently alleged that the scien
characteristic. However
that the nonsciences lack that crucial
non-p rogressive, rational v.
the distinction is put (progressive v.
cal, falsifiable v. non-
non-rational, empirical v. non-empiri
detai led scrut iny. Disciplines
falsifiable), it will not hold up to
litera ry criti cism exhibit all the
like metaphysics, theology, even
ng ratio nal appra isals of the relative
features we require for maki
them. The nonsciences,
merits of competing ideologies within
empirical and conceptual
every bit as muchas the sciences, have
sing the adequacy of
problems; both have criteria for asses
solutions to problems; both can be shown to have made
historical evolution.
significant progress at certain stages of their
n of the cognitive
What has stood in the way of a recognitio
s has been a simplistic
parity of the sciences and the nonscience
y with experimental control
identification of(scientific) rationalit
‘“‘humanistic’’ theories
and quantitative precision. Because
for some thinkers to dismiss
usually lack both, it has been easy
we have seen, the essence of
their rational credentials. But, as
such characteristics.
rationality in science does not depend on
ough t not jump to the opposite
That much said, however, we
and appro priat e to talk of progress
extreme. That it is possible
as established, at least
and rationality in the nonsciences I take
w, however, that the
programatically; it clearly does not follo
fact been as progressive
various humanistic disciplines have in
 ioe

192
HISTORY OF IDEAS
and as rational as the sciences, Progress,
as we said in Part One,
is a matter of degree; two systems
of thought can each be
progressive, while one may show a highe
r rate of progress than
another.
If there is any truth at all in the (posit
ivistic) claim about the
differences between the Sciences
and the nonsciences, and |
suspect there is sometruth in it,
it will be found, not in the
exclusive exhibition of Progress by the scie
nces, but rather in the
higherrate of progress exhibited by them
. But even this claim is
still a matter of vague intuition, and
will remain so until
historians of nonscientific ideas begin
re-writing history with a
view toward appraising the relative progr
ess and rationality of
competing researchtraditions in the human
ities.
There is one final aspect of the human
ities-sciences contrast
which requires comment. It is frequ
ently alleged that the
adoption of doctrines in the nonscience
s can only be a subjective
matter of taste and fashion. If one
becomes an empiricist, or
ideali st, or a trinitarian, or a socialist,
the decision (so itis
claimed) is entirely arbitrary. None
of the positions can be
“proven”’ true or false, and there are alway
s arguments pro and
con. As a piece of descriptive socia
l psychology, there is
doubtless much to be said for this
view. Many persons do
indeed see (and make) the choice betw
een competing ideologies
as an intrinsically nonrational affair. But
there is no reason in
principle why this need be the case. The
choice between atheism
and theism, between phenomenalism and reali
sm, betw
een
intuitionism and formalism, between
capitalism and socialism
(to cite only a few examples) could be
made by appraising the
relative progressiveness (and thereby
the relative rationality) of
these competing research traditions.
lf we could show (as |
suspe ct we can forall the pairs cited above
) that, at the moment,
one tradition has been a more progressiv
e problem solver than
its competitors, then we would havel
egitimate, rational grou nds
for preferring it. If and onlyif the comp
eting traditions emerged
from the analysis with equally progressiv
e evaluations would we
then be entitled to argue that the choic
e between them was
necessarily arbitrary and conventional.
The presumption that
the acceptance or rejection of ideologies
can never in principle be
rationally justified (a presumption at
the core of sociology of
knowledge) is, on this analysis, entirely
unfounded.

',
HISTORY OF IDEAS 193

The Indispensability of History for Theory Appraisal

Thus far in this chapter, our concern has been to examine a

few of the foundational problemsin the historiography of ideas.
By way of conclusion, however, 1 want to shift the discussion
towards a consideration of the relevance of intellectual history to
contemporary cases of theory evaluation. It has often been
argued that any attemptto utilize the historical evolution of a
system of ideas as a vehicle for criticizing or appraising its
current status is a category mistake. Logicians teach us thatit is
a special version of the so-called genetic fallacy to imagine that
the origin or historical career of a doctrine has anything what-
ever to do with its cognitive well-foundedness. Offering us a
sophisticated version of Henry Ford’s dictum that “‘history is
bunk,” most moderntheories of rational appraisal insist that the
temporal career of a doctrine or research tradition is absolutely
irrelevant to its rational acceptability.*' I want to take exception
with this view, even to turn it on its head, by arguing that no
sensible rational appraisal can be made of any doctrine without
a rich knowledge of tts historical development (and of the history
of its rivals).
What leads to these very divergent perspectives on the rele-
vance of intellectual history is a deep disagreement about the
aims and nature of rational appraisal itself. If one takes the
traditional view that, in appraising any doctrine, we should
identify rational belief with truth presumptiveness, then the
history of any doctrine is indeed largely irrelevant to its rational
status. A doctrine could be imagined to have virtually any prior
history whatever and yetstill be true; similarly, a false doctrine
could conceivably exhibit any historical pattern one likes. The
crucial trouble here, of course, is that—for reasons already
discussed—we have no way of determining whether a (consis-
tent) system or theory is true or false, or even presumptively one
or the other. As a result, appraisals of the rationality of accept-
ing any doctrine must be basedon factors other than their truth
status. I suggested earlier that the most promising factor on
whichto pin our acceptanceis “progress at problem solution.”
If, however, we once accept the proposal that the appraisal of
any doctrine should be based on the problem-solving progres-
siveness and effectiveness of the research tradition with which it
 194 HISTORY OF IDEAS

is associated, then we are inevitably committed to the view that

intellectual history must be an ineliminable ingredient in every
rational choice situation. For until we know how a research
tradition has fared through time (especially relative to its extant
rivals) we are in no position to appraise its rational credentials.
To a certain degree, such an approach as that proposed here is
already widely used. The claim that “‘logical positivism has run
out of steam,”’ the observation that ‘‘the New Criticism is no
longer a promising device for literary analysis,” the charge that
“psychoanalysis is becoming increasingly ad hoc and doctrin-
aire’; these, and similar, familiar characterizations already
exploit the insight that a tradition’s history is relevant to an
appraisal of its current cognitive status.
But this mode of analysis is still sufficiently undeveloped that
it has been assumed that a very superficial historical “intuition”’
about how tradition has evolved is enough. If, however, we
take this view as seriously as it deserves then we need far more
than vague impressions about the temporal dimensionsof a re-
search tradition. What we need, if our appraisals are to be atall
reliable, is serious historical scholarship devoted to the various
research traditions in any given field of inquiry. Without the
information which such studies can generate, it is impossible to
make an informed and rational choice between competing
ideologies within any domain. In this sense, and to this degree,
all contemporary disciplines are, or ought to be, parasitic upon
their intellectual ancestry—not only genetically but cognitively.
This last point puts us in a position to return to the objections
voiced by general historians about intellectual history as an
enterprise. Those objections, in so far as they imply that general
history can dispense with the history of ideas, must fall wide of
the mark if the arguments of this chapter are cogent. For history
itself is a theoretical discipline with rival ideologies, alternative
methodologies, and competing traditions; sensible choice be-
tween those traditions hinges, as we have seen, on an awareness
of the intellectual history of those ideologies. Hence, for all its
alleged ‘‘elitism’’ and “idealism,” intellectual history, far from
being at the periphery of the concernsof the general historian, is
directly at the core of any historical research, and is presupposed
by every other form of history—at least to the extent that the

|
 195
HISTORY OF IDEAS

general historian’s problems and methodologies do themselves

have an intellectual history of which the historian must be aware
if he is to write sound history.
Butto say as muchis only to insist that the social or economic
historian must be aware of the intellectual history of history
the
itself. We have not yet challenged the common claim that
history of ideas must be replaced by a broader form of socio-
of
economic history whose function would be the identification
the “‘real,” nonintellectual causes of changing patterns ofbelief.
It is that specific issue which we must now confront.
 Chapter Seven
Rationality and the
Sociology of Knowledge
A man always has two reasons
for doing anything—a 00d reason
and the real reason. 3. PIERPONT MORG
AN

Anyone who wants to drag in

the irrational where the
lucidity and acuity of reason
| still must rule by right
merely shows that he is
afraid to face the mystery
at its legitimate place. KARL MANN
HEIM (1952), p. 229

One of the most important controve

rsies within the community
of scholars who study the evolution
of science has concerned the
role of sociological and psychologica
l factors in the development
of scientific thought. It is at this
nexus between the “internal”
and the “external’’ that intellectual
historians of science cross
swords with social historians of
science, and where thos e who
favor a rational analysis of scie
nce quarrel with historical
sociologists and psycho-historians
of science. Recently, this
controversy has generated more
heat than light, which is
unfortunate becauseit is a genuine
controversy, the outcome of
which may do much to shape our gene
rat conception of science
196

,
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 197

itself. There is, of course, a huge and burgeoning literature in

the sociology of science. The object of this chapter is not
primarily to discuss the detailed conclusions now emerging
within that field; its aim, rather, is to examine the explanatory
scope and range of the sociology of scientific knowledge in
particular, and the sociology of knowledge (of which the former
is a part) in general.' I shal! try to show in particular that the
model of rationality outlined in Part One has many ramifications
for an understanding of the nature and limits of sociology of
knowledge.
We must begin, however, with some preliminary distinctions,
for much of the confusion in this area has arisen from failure
to bear in mind some elementary differentia. It is vital to
distinguish, at the outset, between two very different sorts of
sociologies of science: (1) suppose someonewishes to explain why
a particular scientific society or institution was founded, why a
scientist’s reputation waned, why a particular laboratory was
established when and where it was, or why the number of
Germanscientists rose dramatically between 1820 and 1860. I
propose to call the investigation of such problems the non-
cognitive sociology of science. Such studies are non-cognitive
precisely because their primary concern is not to explain the
beliefs of scientists about the natural world, but rather to
explain their modes of organization and their institutional
structures. ({t is true, of course, that scientists’ beliefs may
condition their modes of institutional organization,’ but what
makes this form of sociology non-cognitive is that the problems
it sets itself to solve are not themselves beliefs about the natural
world.) (2) By contrast, a sociologist may seek to explain why a
certain theory was discovered (or, after discovery, accepted or
rejected) by pointing to the social or economic factors that
predisposed scientists to be sympathetic or hostile to it. Alterna-
tively, he may seek to show that certain socia! structures were
influential on the genesis of the concepts of a theory. Such
efforts fall within the scope of what I shall call the cognitive
sociology of science. Clearly, these two modes of approach, the
cognitive and the non-cognitive, could be applied to any intellec-
tual discipline, ranging from the specific sciences to theology,
metaphysics, or sociology itself. As a result, we can speak more
198 RATIONALITY AND SOCLOLOGY OF KNOWLEDGE

generally of the non-cognitive and the cognitive sociology of

knowledge.
From what was said in chapter six, it should be clear that
there is neither overlap nor conflict between the intellectual
historian of science (or knowledge) and the non-cognitive sociol-
ogist, because they are addressing themselves to radically dif-
ferent problem situations. The intellectual historian is trying to
explain why scientists or other thinkers in the past adopted the
beliefs or the solutions (theories) which they did; the non-cogni-
tive sociologist does not, by definition, have beliefs about the
world among his class of problems to be solved. Quite the
reverse is the case, however, when we compare cognitive sociol-
ogy of science with the intellectual or rational historiography of
science. For here, there is the possibility for enormous (and
potentially fruitful) conflict. The intellectual historian of knowl-
edge will generally seek to explain why some agent believed some
theory by talking about the arguments and the evidence for and
against the theory and its competitors. The cognitive sociologist
of knowledge, on the other hand, will generally try to explain
why the agent believed the theory in terms of the social,
economic, psychological, and institutional circumstances in
whichthe agent found himself. Both are trying to solve the same
problem (namely, the belief of some historical agent), yet their
modesof solution are so different as to be almost incommensur-
able. Is there any way, given these conflicting explanatory
strategies, of determining whois right, the intellectual histor-
ian or the cognitive sociologist? Or could they both be?
The possibility of answering this important question hinges
upon whether we can articulate any fair criteria for deciding
between the seemingly competing historical accounts given by
the cognitive sociologist and the intellectual historian. The
articulation of such criteria is one of the central aims of this
chapter.

The Domain of Cognitive Sociology

Before we proceed to that task, however, we must get clearer
about the character of cognitive sociology, for some of its ablest
 199
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE
even incon-
practitioners seem at times to have made confusing,
ogical theory and about
sistent, claims about the scope of sociol
the nature of sociological explanations.

The Nature of Cognitive Sociology

re of cognitive
As we have already seen, one important featu
s to be its empir ical problems. But
sociologyis that it takes belief
to disti nguis h it from many other non-
that is clearly not enough
instance, the
sociological modes of explaining belief (such as, for
nguishes cognitive
rational history of science). What further disti
makes it sociological,
sociology from these other fields, what
to be explained in
must be an assumption that beliefs are
So we can say that an
terms of the social situations of believers.
must be that of
essential task of any cognitive sociologist
in, its social roots
exhibiting, for any belief he wishes to expla
chara cterize what a
and origins. To say as much is only to
we have formulated
sociological explanation will look like once
is some wayof identify-
it. But presumably what we need as well
likely to be amena bie to a
ing those belief situations which are
sociological analysis.
virtually every shift
There are those who would maintain that
y of think ers is expli cable in terms of
in belief in any communit
and would thus make the problem
the social substructures,
co-extensive with the
domain of the sociology of knowledge
the other extreme are
entire history of human thought.’ At
who have claimed
certain critics of the sociology of knowledge
in the history of ideas
that there are virtually no transformations
or functions of, changes in
which are in any way indebted to,
determinists (e.g.,
social structure. The uncompromising social
lf) and the intransi-
certain Marxists—though not Marx himse
exemplify these two
gent idealists (e-.g., Hegel) respectively
view makes much sense
poles.* Unfortunately, neither point of
an enor mous amount of
of the historical record. There is
doctr ines and ideas bear no
evidence which shows that certain
ncies of social circu mstance:
straightforward relation to the exige
ple that ‘2 + 2 = 4” or the
to cite but two examples, the princi
down ward s when relea sed” are
idea that ‘“‘most heavy bodies fall
a wide variet y of cultur al and
beliefs to which persons from
 200 RATIONALITY AND SOCIOLOGY OF KNOW
LEDGE
social situations subscribe. Anyone who
would suggest that such
beliefs were socially determined or cond
itioned would betray a
remarkable ignorance of the ways in
which such beliefs were
generated and established. Similarly,
there clearly are ideas and
beliefs which do have tangible socia
l roots and origins. To
imagine, for instance, that a nine
teenth-century white slave -
holder espouses a belief in the racial
inferiority of blacks for
purely intellectual reasons requires a feat
of moral charity which
few of us could make. To suggest that
most nineteenth-century
Germanfactory workers who favored
socialism did so because of
the rational well-foundednessofits doctr
ines is again a view that
demands an enormous degree ofcreduli
ty.
But if we grant that the truth of the
matter lies somewhere
between rigid social determinatio
n and insular idealism, we are
immediately confronted with a central
problem, to wit, whaz
sorts of beliefs are candidates for a
sociological analysis and
which are not? Put differently, and in
the language of earlier
chapters, what kinds of belief situations
can function as legiti-
mate empirical problems for sociology?
One might think that
this is a purely empirical question whic
h cannot be settled in
advance a priori, but which can only
be decided by looking at
matters case-by-case. The troubles with
this seemingly innocu-
ous answer to our question are both
practical and theoretical.
On thepractical side, we run up again
st the fact that thereare,
within the extant record, literally
millions of beliefs. If the
sociologist has no regulative Principles
which will guide his
initial selection of potentially promisin
g problems, he could
make scarcely any headway. For example,
one might ask about
each ofthe truths of arithmetic whether
they have social origins.
Weshould begin with, say, “1 + 1 =
2” and work up the scale.
Because of such practical difficulties with
a purely empirical
approach to specifying the problem-set
for cognitive sociology,
virtually all researchers in the field have
sought to delimit the
domain of possible sociological problems
by adopting certain
regulative or methodological principles
, whose function is to
provide a useful initial sorting mech
anism that will draw
attention to those types of beliefs whic
h are most likely to be
susceptible to sociological analysis,

ee
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 201

But there are theoretical, as well as practical, reasons for

laying down in advance some wayof deciding on the boundaries
of the potential problems for the cognitive sociologist. If it were
true that a// beliefs were not the result of rational deliberation or
enlightened evaluation, but rather were simply determined by
the social situation of the believer, then the whole enterprise of
cognitive sociology would be self-indicting; for if a// beliefs are
socially caused, rather than rationally well-founded, then the
beliefs of the cognitive sociologist himself have no relevant
rational credentials and hence no special claims to accepta-
bility.° Ernst Griinwald put the point tellingly when he re-
marked: ‘‘For the thesis that all thought is existentially [i.e.,
socially] determined and thus cannot claim to be true doesitself
claim to be true.’® Thus, the cognitive sociologist, to avoid
being hoisted with his own petard, is committed to the view
that some beliefs are rationally well-founded rather than socially
determined.
There are three different methodological principles most often
cited (or used implicitly) in this connection by cognitive sociolo-
gists of knowledge. I shall call them the arationality assumption,
the historico-social assumption, and the interdisciplinary as-
sumption. Althoughnotstrictly compatible, these conditions are
widely (and often simultaneously} used in most works on the
sociology of knowledge. I want to discuss them in some detail,
for the model of science and knowledge developed in Part One of
this essay impinges upon each of them and thereby upon the
whole of cognitive sociology of knowledge.

The Arationality Assumption. Many sociologists of knowledge,

following Karl Mannheim, distinguish between “immanent”
and “non-immanent” (or “existentially determined’’) ideas.’
Immanent ideas (or concepts or propositions or beliefs—these
are all run together by most writers) are those which can be
shown to be naturally and rationally linked to other ideas to
which a believer adheres. An archetypal example would be the
theorems of Euclid’s geometry. Once one accepts the axioms,
oneis logically or rationally constrained to accept their theore-
matic consequences. No thinking person who understood the
202 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

one, could deny the other. Non-immanent (existential) ideas, on

the other hand, are those which do not carry their rational
credentials with them. They are ideas which people may accept,
but which are not intrinsically more rational than many other
alternative ideas which they might have accepted.
Mostsociologists of knowledge agree with Mannheim thatit is
only non-immanent ideas, only those which are not the most
rationally well-founded in a given situation, which it is appro-
priate for sociology to attempt to explain. {t is easy to see the
plausibility of this stipulation. If the acceptance of somebelief,
x, seems to follow naturally and rationally from the prior
acceptance of beliefs y and z, then there seems no point in
maintaining that the espousal of x is directly caused by social or
economic circumstances.® If, on the other hand, someone
accepts a belief a, which is not rationally related to his other
beliefs b, c, ..., i, then it seems as if the only natural way of
explaining his espousal of a will be in terms of factors that are
extra-rational, such as the social (or psychological) situation of
the believer in question.
I propose to call this demarcation criterion the arationality
assumption; basically, it amounts to the claim that the sociology
of knowledge may step in to explain beliefs if and only if those
beliefs cannot be explained in terms of their rational merits. As
Robert Merton points out, this view is widely accepted by
working sociologists: ‘‘A central point of agreement in all
approaches to the sociology of knowledge is the thesis that
thought has an existential [i.e., social] basis in so far as it is not
immanently [i.e., rationally] determined.’”® Essentially, the ara-
tionality assumption establishes a division of labor between the
historian of ideas and the sociologist of knowledge; saying, in
effect, that the historian of ideas, using the machinery available
to him, can explain the history of thought insofar as it is
rationally well-founded and that the sociologist of knowledge
steps in at precisely those points where a rational analysis of the
acceptance (or rejection) of an idea fails to square with the
actual situation.
The arationality assumption, we must stress, is a methodo-
logical principle, not a metaphysical doctrine. It does not assert
that “whenever a belief can be explained by adequate reasons,
 203
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

then it could not have been socially caused”; it makes the

weaker, programmatic proposal that “whenever a belief can be
explained by adequate reasons, there is no need for, andlittle
promise in, seeking out an alternative explanation in terms of
social causes.”’
Although the arationality assumption is widely subscribed to
by cognitive sociologists, there are few arguments ever cited for
its cogency. Because it has recently come under attack by
historical sociologists and becauseit is so crucial as a demarca-
tion criterion between rational explanations of belief and extra-
rational explanations thereof, it is worth exploring briefly the
grounds for it. In order to do so, let us suppose the following
imaginary situation: there is some person, x, whobelieves A. His
belief patterns are being investigated by two researchers, y and
z, Suppose y is an intellectual historian who takes the arational-
of
ity assumption seriously; he looks for, and finds, a way
showing that x's belief is rationally well-founded, given x's other
beliefs B, C,..., L So far as y is concerned, he now has as full
an explanation of x's belief in A as seems possible. Suppose,
however, that z is a maverick sociologist who refuses to accept
the arationality assumption. While granting that y has found a
“rational” explanation of x's belief, z is convinced that there
maystill be scope for sociological work on belief A (perhaps
because z suspects that y has mistaken a “rationalization” of
x’s for the ‘real’ cause for x's espousing A). After some
biographical research on x, z discovers that x was a memberof a
lower middle class and had an Oedipalfixation on his mother.
Suppose, further, that z would argue that persons in x's
situation generally tend to hold beliefs such as A. While not
denying that y has offered an alternative explanation of x's
belief, the sociologist z nonetheless insists that his own explana-
tion still stands; thatit is, if anything, “‘more fundamental” than
y's explanation. How, if at all, cany convince z that his explana-
tion is bogus becauseit violates the arationality assumption?
One might, of course, simply postulate the arationality as-
sumption as a matter of faith; as a postulate without which it
would perhaps never be possible to choose between conflicting
accounts of human belief. But such pious hand waving will
may
scarcely convince our resolute social determinist, z. What

aE
 204 RATIONALITY AND SOCIOLOG
Y OF KNOWLEDGE
help is an analysis of z's own intellectual
orientation. Z, and
like-minded types, are attempting to expla
= amcor nee,weedSitSORE, aye lage ate.

in beliefs. Any ex.

planation, if it is cogent, is an argument,
a reasoning process
that moves from adequate premises
to plausible conclusions.
The whole point of offering an explanatio
n, unless it be an idle
speech act, is to demonstrate that the concl
usion follows ration-
ally from the premises. So that z, insofar as he
offers sociological
explanations, is presuming that at least some
persons (notably
himself) accept certain beliefs because they
have good reasons
for doing so. (One presumesherethat z woul
d not take kindly to
the suggestion that the only cause of his belief
in a certain
sociological explanation washis location in
the social network!)
But if z insists that certain agents’ beliefs
(namely, his own) are
rationally well-founded and not merely a funct
ion of their social
position, then the onus is on him to show whyit
is appropriate to
regard his own beliefs as situationally
transcendent, while the
beliefs of the persons he studies—even
when they can be
rationally explained—ought not be viewe
d as independent of
their social situation.
There is a very different manner in which
one might seek to
adjudicate this controversy between y and z,
namely, by viewing
their theoretical systems (in the langu
age of Part One) as
competing research traditions. Approach
ed in that way, we
could ask which has solved more important
empirical problems.
| There can be no doubt whatever that,
at least at this point in
time, the rational historiography of ideas has
gone much further
toward explaining a large number of impor
tant historical cases
of belief than historical sociology has. Indee
d, the “success
ratio”’ of intellectual history is several order
s of magnitude
greater than that of cognitive sociology.'°
At the level of
conceptual problems, too, the traditions
of the intellectual
historian are generally acknowledged tobein
less acute difficulty
than those of the cognitive sociologist."
Under such circum-
stances it would be entirely appropriate to
point out to z that,
whenever we have competing rational and socio
logical explana-
tions of the same belief, good sensedictates
that the “rational”
explanation should be given priority over
the sociological one
precisely because the former has shown itse
lf to be the most
fruitful. (Which is not to say, of course, that sociological
 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 205

explanations are inappropriate where rational reconstructions

fail to apply.)
Whether it is for reasons such as these that most cognitive
sociologists adhere to the arationality assumption, 1 do not
know. But whatever its grounds might be, it is taken as axio-
matic by most practitioners, and it is important for us here to
examine someofits important consequences.
Despite its pervasiveness, it has been little noticed that the
arationality assumption is a great deal more problematic than
most of its proponents recognize. In order to apply it, we
obviously need a theory about whatrational belief is. Without
such a theory, the arationality condition is meaningless. But as
we have seen in Part One, and as should have been clear all
along, there is more than one conceivable theory of rationality.
Because different theories of rationality will classify beliefs
differently (certain theories making a particular belief rational
while other theories make the samebelief irrational), we can see
that an essential prolegomenon to any adequate cognitive sociol-
ogy of knowledge is the choice as to a theory of rationality."? If
we accept, as some working sociologists are prone to, a simple-
minded theory of rationality which puts excessive constraints on
what is to count as rational belief, then the domain of the
arational—and thus the domain of the soctological—is going to
loom very large. If, on the other hand, we subscribe to a richer
theory of rationality, far more beliefs are going to seem “‘imma-
nent” and thus not susceptible to sociological analysis.
Insensitivity to the variety of theories about rational belief has
been the source of much mischief and confusion in the writings
of many prominent sociologists. Assuming that the “textbook,
inductivist’’ theory of rationality which they inherited from
philosophers of science was sacrosanct and final, sociologists
have been proneto regard asirrational (and thus as sociological)
many episodes in the history of thought which are, by other
standards of rationality, entirely rational. This, in turn, has led
sociologists to seek social causes for processes which can be
explained entirely in immanentist terms.
If, for instance, we subscribe to a crude ‘empiricist’? model of
rationality according to which the empirical success of a theory
is the only relevant determinantof its rational acceptability, we

i
 206 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE
are going to look askance on those episodes in the history
of
thought where (to use the language of Part One)
conceptual
problemsplay a majorrole in determining which theorie
s will be
acceptedorrejected. If a theory in the past has been
objected to
on the grounds ofits incompatibility with a certain
metaphysica]
or epistemological or theologicalbelief structure,
proponents of
this limited, empirical model of rationality will see
the episode as
intrinsically irrational, as one in which certain
ill-founded
prejudices were allowed to controvert the rationa
l judgments of
the agents in question. This, in turn, will lead
to the conclusion
that social factors must have had something to
do with the
outcome of the decision, for the rational canons
of preference
were seemingly ignored.
What vitiates this approach to history, of course,
is the
existence of other models of rational belief which
would makeit
perfectly reasonable, under certain circumstance
s, for factors of
a philosophical or theological kind to enter into
the rational
appraisalof a particular theory. Viewed through
the perspective
of such models, developments which might previously
have been
regarded as prejudiced, obscurantist and irrational
acquire a
rationallegitimacy, which obviates the need to look
to the social
milieu for an explanation of whatis going on. The
moral should
be clear: before we classify an episode as aration
al, before we
begin the search for social causes to explain the ‘‘deviat
ions”’
from the rational norm, we must be quite sure that
our notion of
rationality is an adequate one. To my knowledge,
few if any
sociologists have seen the force of this point, and their
work is
the worse for their failure to seeit. Unfortunatel
y, the error is
doubly confounded; in addition to their failure to recogn
ize that
there may be a broad spectrum of theories of rationa
lity, they
have also generally chosen to subscribe to that
model of
rationality which is the most limited ofall.
To see just how pervasive this error is, it may be useful
to look
at a few prominent examples. In his influential
Structure of
Scientific Revolutions, Thomas Kuhn considers
several of the
better-known “empiricist” models of scientific rationa
lity which
philosophers have espoused. He finds both the confir
mational
and the falsificational models inadequate, but goes
on from
there to enunciate his own model ofscientific rationa
lity. In its
 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 207

essential features, this model is a purely empiricist one which

shares with the other models the conviction that it is only the
empirical problem-solving ability of a theory which can be
relevant to its rational appraisal. Kuhn then points out, quite
rightly, that there are many episodes in the history of science
which seem to involve decisions about theories where factors
other than the empirical credentials of the theories under
examination were prominent.'? Kuhn argues, or rather alleges
without argument, that in such cases there must be important
social and institutional pressures at work. In making this slide,
Kuhn is obviously, if implicitly, invoking the arationality condi-
tion. There can be no objection to that, but one wishes he might
have worried more deeply about what rationality amounts to
before jumping to the conclusion that his empirical model of
rationality was subtle enough to provide the careful discrimina-
tions between the immanent andarational.
Similar precipitate leaps to the presumption of arationality
occur repeatedly in Maurice Richter’s recent Science as a Cul-
tural Process. Richter argues, for instance, that Darwin’s theory
of evolution: “was challenged in the nineteenth century not only on
the basis of reasonable scientific arguments . . . but also on the
basis of dogmatic theological assumptions.”’* Richter may, of
course, be correct in his historical claim; but the image of
scientific rationality which forms the backdrop to his notion of
“reasonable scientific arguments” is suspect, at best. He main-
tains, for example, that “‘the contents of scientific knowledge . . .
are to be determined by observations of nature.’’'* This highly
empiricist notion of what constitutes proper science leads Richter,
not surprisingly, to see many historical episodes as irrational (be-
cause not reconstructible on a naive empiricist model of ration-
ality) and thus as sociological.
One of the most striking instances of the hyper-positivistic
current in the cognitive sociology of science is provided by the
work of the well-known sociologist Bernard Barber. In a widely
cited article in Science in 1961,'° Barber explored the various
factors which dispose scientists to refuse to accept new ideas and
new discoveries. In this latter-day version of Bacon’s “‘idols,”’
Barberidentifies methodology and theology as two of the major
| sources of ‘‘cultural resistance to new ideas.”’ There is clearly

a
 208 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

nothing wrong with Barber’s hunch that philosophy and theol-

ogy have played an important role in scientific debate. His
positivism only comes out when, having noted the interaction, he
goes on to bemoan it and to urge that we should seek to reduce
their pernicious influence.'’ Barber has recognized neither that
it is often perfectly reasonable, and not just prejudice, to attend
to the broader methodological and philosophical implications of
a newscientific theory, nor that methodology and theology have
3 historically been involved as often to legitimate new theories as
they have been called upon to discredit them. Barber yearns for
what he calls the “‘open-minded” scientist, who restricts himself
entirely to the straightforward, “‘scientific’”’ merits of a new idea.
Barber’s purely empiricist model of theory appraisal allows no
room for anythingelse.
In these, as tn numerous other cases that could be cited from
the recent literature, scholars seem to have leapt prematurely to
the conclusion that the inapplicability of one or other standard
? model of rationality to any particular case establishes the
7 arationality (and thereby the social character) of the case at
hand. It should be clear that if we accept a different model of
rationality, one perhaps built along the lines sketched earlier in
this essay, then the domain of would-be sociological cases will be
far smaller than if we accept one of the more traditional
empirical theories of rationality. (My own proposal would be
that the need for sociological analysis of a case only arises when
we can show that the actual evaluation of a particular theory in
the past was radically at odds with the appraisal it should be
accorded by the lights of the problem-solving model of ra-
tionality.)
I have dwelt at length on the parasitic dependence of cognitive
sociology of knowledge on theories of rationality not only to draw
attention to the need for sociologists to be more self-critical
about their judgments concerning the rationality of particular
cases, but also to emphasize the fact that the application of
cognitive sociology to historical cases must await the prior results
of the application of the methods of intellectual history to those
cases. The cognitive sociologist must look to the intellectual
historian for cues and clues about which cases it is appropriate
for him to analyze. Until the rational history of any episode has

,
 EDGE 209
RATIONALITY AND SOCIOLOGY OF KNOWL
ble theory of
been written (and that, by using the best availa
must simpl y bide his time;
rationality), the cognitive sociologist
the aratio nality assum ption which
to do otherwise is to abrogate
sociol ogical though t. (Some thing
is at the heart of contemporary
ized by Mannh eim,' ® but his latter-
akin to this point was recogn
could do socio-
day disciples have tended to assume that one
blissfu l ignor ance of the ration al history of
logical history in
ideas!)
assumption has
We thus see that accepting the arationality
n ofpossible belief
three important consequences: (1) the domai
to those in which
situations for sociological analysis is restricted
incompatible
agents accept beliefs or weight problems in ways
t; (2) the sociologist
with what rational appraisals would sugges
of rationality
of knowledge must be able to show that the theory
what cases might
to which he subscribes (in order to determine
(3) the historical
be sociological) is the best available one;
episode he
sociologist of knowledge must show, for any given
ble of being expla ined in terms
wishes to explain, thatit is incapa
of rational, intellectual history.
socially ex-
In distinguishing between the rational and the
do not mean to sugges t that there is
plicable as I have, [
ration ality or nothi ng ration al about social
nothing social about
the revers e is the case. The flouri shing of
structures. Quite
inevitably upon
rational patterns of choice and belief depends
e of certai n social struct ures and social norms.
the pre-existenc
choice would be
(To take an extreme example, rational theory
ively suppressed
impossible in a society whose institutions effect
y, the efficient
the open discussion of alternative theories.) Equall
the system oftrial
functioning of most social institutions (e.g.,
such instit utions can,
by jury) presupposes that the agents within
more often than not, makerational decisions.
and “‘so-
But this continuous interpenetration of “rational”
to invoke the
cial” factors should not hinder our capacity
John Stuart Mill pointe d out more
arationality assumption. As
offeri ng any explan ation of some event or
than a century ago, in
to comple teness . To give a “full”
belief, we must not aspire
situat ion, S, would presu mably requir e a
explanation of any
have happened in
complete enumeration ofall the events which
are links in the
the universe prior to S, since all those events
 210 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE
causal chain culminating in §. Rather than aspiring to such full
explanations, Mill argued that when we explain anysituation, S,
we should select from among the antecedents of S, those
particular circumstances, c, which seem to be most crucial
and
relevant to the occurrence of S. If we take Mill’s analysis
seriously (and a failure to do so would result in explanatory
anarchy) it provides grounds for avoiding that muddle-headed
eclecticism which argues thatintellectual and social factors can
never be usefully distinguished.
Following Mill’s lead, we can grant that certain social factors
may well be preconditions for rational belief and yet still
legitimately exclude those social factors from an explanation
of
a certain belief, provided we can show that the most crucial
and
relevant antecedent to the acceptance of the belief was a well-
founded reasoning process on the part of the believing agent. In
thus arguing (as the arationality assumption suggests) for the
priority of rational over social explanations for a belief—where
both are available—one is not implying that rational decision
making has no social dimensions; rather oneis stressing
that, in
those cases where agents have sound reasons for their beliefs,
those reasons are the most appropriate items to invoke in an
explanation of the beliefs which those reasons watrant.

The Historico-Social Assumption. If the failure to recognize

the dependenceof cognitive sociology on theories of rationality
has been onepersistent feature of the sociology of knowledge,
another major source of ambiguity is to be found in the tendency
to equivocate between, and sometimes to identify, the “‘histor-
ical”’ with the “social.” Kar! Mannheim’s writings provide ample
illustrations of this equivocation. As Mannheim points out, there
are two very different sorts of beliefs which we find espoused in
the past: those whose very formulation and whose presupposi-
tions are clearly traceable to a particular time and place, and
those which betray virtually nothing about their historical
or
social origins. Putting the distinction bit differently, one might
say that certain propositions inherited from their past carry
something of their history with them, while others give no
clues
as to when, and under what circumstances, they
werefirst
enunciated. If, for instance, we encounter the stateme
nt ‘‘the
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 211

’
heart is like a pump,” we know perfectly well that such a
statement must have been madeafter the invention of pumps,
and presumably after some detailed anatomical investigations
investigations of the circulatory system. It is simply not a
statement which a Greek of the third century B.c., nor a
Polynesian of the eighteenth century Could have made. At the
other extreme, certain beliefs (e.g., “2 + 2 = 4’’) tell us very
little indeed about either the time or the place where they first
emerged.
We might call beliefs which do carry their history with them,
contextual, for they do provide important clues as to the cultural
context which generated them. Other beliefs we might cail
non-contextual,'® Clearly, these two extremes are ideal cases; for
the working historian, almost every case will be a matter of
greater or lesser contextuality. (Even in the extreme case of
beliefs such as ‘‘2 + 2 = 4,” we can derive reliable conclusions
about certain intellectual characteristics of the cultures in which
such beliefs could arise.)
What is important here is not the distinction itself, but rather
what cognitive sociologists of knowledge seek to do with it.
Mannheim, for example, argues that a belief which is contextual
(in the sense just sketched above) is a belief which is “historically
and socially determined.’’ Given a sufficiently foose sense of
‘determination,’ this argument is doubtless sound, indeed
vacuously so. But Mannheim’s next step is to argue that any
contextual belief—that is, any belief which can be definitely
located in history—is thereby open to sociological analysis. If we
can pin a belief down to ‘‘a particular historical setting,” then
we have, Mannheim claims, presumptive evidence ‘‘of an ‘infil-
tration of the social position’ of the investigator into the results
of his study.’’?°
This argument is entirely specious, precisely because in
making it, Mannheim (like others who follow him) trades on an
equivocation between the ‘‘historical” and the “‘social.”’ If, for
instance, we encounter a statement such as “electricity is caused
by a fluid whose particles mutually repel one another,’’ anyone
familiar with the history of the physical sciences can readily date
it approximately and make somereliable conjectures about the
intellectual context in which the statement first arose. Similarly,
 212 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

if we meet a statement like “the Absolute is pure Becoming,”

anyone familiar with the history of philosophy can readily make
sound conjectures about when, where, and by whom that
statement became an object of belief. But the fact that these
statements are contextual, that they have only been believed at
certain times and places, does not establish any interesting sense
in which they are necessarily social, or open to sociological
analysis. What makes Mannheim’s argument seem initially
plausible is his constant conjoining of the terms ‘‘historical’”’ and
“social,"” as when he regularly speaks of “historically and
socially determined beliefs."*! He spends great effort quite
rightly establishing that certain beliefs are of an historical
character. It is then by a purely rhetorical slide that Mannheim
is able to pretend he has thereby shown such beliefs to be also
socially determinedin character.
No less a thinker than Emile Durkheim exhibits a similar
tendency to assumethat any belief which arises in a particular
culture or at a particular time must of necessity be socially
produced. For instance, in Durkheim’s influential Elementary
Forms of the Religious Life, he claims that certain cultural
differences in the laws of logic “prove that they depend upon
factors that are historical and consequently social.’?* The last
italicized portion of this passage gives the game away. If the
establishment of the historical contextuality of a belief is
tantamount to rendering that belief socially determined, then
the cognitive sociologist has an easy task. He need only look to
the history of ideas to find those beliefs which are contextual and
he has—hey presto—a whole setof “‘sociological’” desiderata.
As we have said before, however, the slide from the historic-
ally to the socially determined is nothing more than intellectual
legerdemain. The ‘“‘consequently social” in the Durkheim
passage quoted above is completely gratuitous; if one is to
establish that any belief is socially determined, one must—at a
minimum—establish some connection between the social situa-
tion of the believer and the belief he espouses. The fact that he
espoused the belief in 1890 rather than 1870—which is enough
to establish the historical character of the belief—leaves the
questionof its social character entirely open.
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 213

There are many other cognitive sociologists, besides Mann-

heim and Durkheim, who seem to believe that if a belief
emerged within a particular historical context, then that belief is
a fortiori susceptible of sociological explanation.”? But this
assumption involves a confusion of the intellectual culture with
the social culture. As Part One has madeclear, it is very often
the case that certain beliefs tend to emerge under specific
intellectual circumstances which are a function of both the
empirical problems recognized in the period and of the dom-
inant research traditions characteristic of the period. But there
may be nothing of social or sociological interest about this
process of intellectual assimilation of ideas within a pre-estab-
lished intellectual context or framework.

The Interdisciplinary Assumption. Thus far, we have looked

at some of the ambiguities implicit in the historico-social
assumption and at someof the difficulties posed by the aration-
ality assumption. There is yet another pervasive assumption
about the scope of cognitive sociology which we might call *‘the
interdisciplinary assumption.’” In its most general form, it
assumes that whenever thinkers in one branch of inquiry or
discipline draw on, or react to, ideas in other disciplines, then we
have grounds for presuming that sociological factors are at work.
The morespecific version of this postulate, when applied to the
history of science, amounts to the claim that whenever ‘“‘scien-
tists’’ are influenced by the “nonscientific’’ (e.g., moral, reli-
gious, epistemological, metaphysical) consequences of scientific
theory, then this indicates the intrusion of extra-rational, social
factors into the scientific situation.
The interdisciplinary postulate arises, I believe, from an
idiosyncratic interpretation of the arationality assumption. If
one assumes that science is rational only insofar as it is self-
contained, and if one also assumes that whateveris arational is
thereby socially caused, then the interdisciplinary assumption
follows without difficulty. It is the first premise which makes the
inference untrustworthy. As Part One of this work makesclear,
it is not necessarily arational for scientists to be concerned about
the conceptual relationships between their scientific work (in the
214 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

narrow sense of that phrase) and the broader intellectual com-

ponents of contemporary culture. We have already discussed the
merits of this claim earlier. What should be pointed out here is
that there are whole ‘‘schools”’ of cognitive sociology (one thinks
especially of Sorokin, Scheler, Durkheim?‘ and Richter, for ex-
ample) which see the céntral aim of sociology to be a study of the
waysin which the different ideological elements in a culture are
integrated. If the arguments of this essay have any cogency,
studies of ‘ideological integration,” in so far as such integration
is rationaily well-founded, belong to intellectual history and fall
completely outside the reaim of cognitive sociology.
It might be thought that these abstract considerations have
little bearing on the actual research done byhistorically oriented
sociologists and that these foundational confusions pose no
problems when applied to particular cases. Such a view would be
quite misleading, as we can see by looking in detail at two of the
best-known recent historical studies on the sociology of scientific
ideas; namely, the work of Theodore Brown and Paul Forman.
These two historical studies, although concerned with differ-
ent epochs and different sciences, both seek to show how the
reception of certain scientific theories was crucially dependent
upon social and institutional circumstances. It is worth analys-
ing these investigations in some detail, for they highlight some of
the confused assumptions lying behind even the most sophisti-
cated studies in the historical sociology of science.
Brown’s aim is to explain why certain prominent English phy-
sicians and natural philosophers enthusiastically accepted the
mechanistic approachto life in the middle of the seventeenth
century. In brief, his answer is that these thinkers were asso-
ciated with the Royal College of Physicians, an organization
whosesocial prestige and monopoly overthe licensing of medical
practitioners were acutely threatened—in part because the Col-
lege was associated with a moribund and old-fashioned type of
Galenic-Aristotelian physiology. The mechanical philosophy, by
contrast, was perceived as an up-to-date, ‘‘trendy’’ approach
with which the physicians could combat their traditional adver-
saries—the apothecaries. Brown suggests that the endorsement
of the new mechanistic approach to physiology by members of
the College was a direct consequence of the institutional and
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 215

social crisis confronting the College. In Brown’s own words: “‘the

collegiate physicians . . . borrowed ideas from the mechanical
philosophy . . . because they were engagedin political struggles
with their professional prestige seriously lowered and because by
borrowing they hoped to raise their prestige again, thereby
improving their political position.’’?°
Forman, on the other hand, seeks to explain why the indeter-
minacy principle was so readily and quickly accepted by German
theoretical physicists in the late 1920s. Forman’s hypothesis is
that those physicists were predisposed to be sympathetic to
assaults on the causal principle because there was in the German
intellectual milieu a strong current (deriving especially from
Spengler) which argued that science was overly rationalistic,
overly mechanical, overly deterministic—that, in short, it left no
room for human values nor for the frailty of the human mind.
On Forman’s account, this neo-romantic, anti-mechanistic
movement threatened the prestige of physical scientists to such a
degree that they were actively seeking ways to improve their
image by repudiating that deterministic materialism of which
they stood accused.’® The uncertainty relation (when naively
interpreted) offered them a splendid riposte to their detractors
since the physicists could use it to prove that they were not
weddedto a fully mechanistic world picture.
Whatlies behind the analyses of both Brown and Forman are
a set of historiographical assumptions about the character of
science, assumptions which allow them to set up their problems
in the ways they do. Chief among these assumptions are the
Kuhnian convictions that: (1) disciplines generally possess an
autonomy which makes them immunefrom ‘‘outside pressures”
coming from the broader social and cultural environment;?’
(2) every scientific discipline is fundamentally conservative, and
resists any re-orientation of its conceptual commitments except
in times of acute crisis; (3) these rare periods of intellectual crisis
(and here Brown and Forman depart from Kuhn) are generated
not from within a discipline, but rather by some external threat
to the prestige, funding or intellectual standing of the disci-
pline’s practitioners;”* and (4) the re-ordering of the beliefs of a
community of scientists is caused by these outside social pres-
sures rather than any process of rational appraisal within the
216 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

discipline itself. Forman himself makes many of these presuppo-

sitions explicit when hewrites:
We may suppose that when scientists and their enterprise are enjoying high
prestige in their immediate (or, otherwise most important) social environment,
they are also relatively free to ignore the specific doctrines, sympathies, and
antipathies which constitute the corresponding intellectual milieu. With appro-
bation assured, they are free of external pressure, free to follow the internal
pressure of the discipline—which usually means free to hold fast to traditional
ideology and conceptual predispositions. When, however, scientists and their
enterprise are experiencing a loss of prestige, they are impelled to take
measures to counter that decline . . . [which] may even affect the doctrinal
foundations of the discipline . . .?°

It is worth noting at the outset that neither Forman nor Brown

explores whether the emergence of acausal theories in German
physics or mechanistic theories in British physiology might have
been an entirely appropriate and rational response to the
empirical and conceptual criticisms of previously dominant
theories. They evidently jump immediately to the assumption
that social forces were at work because of their commitment to
the thesis that disciplines only allow for the intrusion of non-
disciplinary considerations (e.g., of a philosophical, cultural, or
political nature) when the discipline in question is under acute
social pressure. Equally, their conviction that disciplines are
reactionary and resistant to change makes it almost inevitable
that, when profound conceptual changes occur within a disci-
pline, they as historians will look to outside social and insti-
tutional factors for an explanation of what must seem (on their
model of change) to be uncharacteristic, even “‘unscientific,”’
behavior. *°
In crucial respects, therefore, the Forman-Brown investiga-
tions hang upon the adequacyof their historiographical assump-
tions (1) to (4). To the extent that the latter are dubious (as I
have shown them to be in Part One), the historical researches
done under their aegis must remain unconvincing.
Because their (Kuhnian) image of science precludes them
from believing that scientists can ever have good scientific
reasons for changing their minds, or for worrying about broader
intellectual issues, both Forman and Brown systematically ig-
nore the scientific and rational merits of the ideas they discuss.
After all, it might just be the case that Heisenberg enunciated
the indeterminary principle because, as he tells us, he thought
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 217

that the weight of argument favored it. It might just be that

Walter Charleton accepted the mechanical philosophy because
—as he explains in 400 turgid pages—that theory wasrationally
preferable to its alternatives. The invocation by Forman and
Brown of social and institutional explanations takes place in a
curious intellectual vacuum. They do not ask themselves whe-
ther their “social” accounts of theory reception do or do not
succeed at explaining dimensions of the historical situation
which might be explained in terms of sound cognitive reasons.
They produce no evidence for their core historical conviction
that science is intrinsically conservative and, in usual circum-
stances, entirely autonomous.*!

The Theoretical Foundations of Cognitive Sociology

The social causation of ideas. Up to this point, we have been
preoccupied with preliminaries, important ones to be sure, but
we have yet to say anything about the content of sociological
theories. If our aim thus far has been to geta little clearer about
the problem situations with which a cognitive sociologist ought,
in principle, to be concerned, we must now turn our attention
toward the character of sociological theory itself. Although this
is not the place for anything like a detailed treatment of the
substantive commitments of cognitive sociology, a few general
observations are perhaps in order, particularly about the cogni-
tive sociology of science.
As we have already noted, any cognitive sociological explana-
tion must, at the very least, assert a causal relationship between
some belief, x, of a thinker, y, and y’s social situation, z. It will
(if the explanations of sociology are in any sense “scientific’’) do
so by invoking a general jaw which asserts that all (or most)
believers in situation type z adoptbeliefs of type x.
Hence, the viability of cognitive sociology depends upon our
ability to discover general causal (or functional) relationships
between social structures and beliefs. More specifically, cogni-
tive sociology of science is predicated on the existence of
determinable correlations between the social! background of a
scientist and the specific beliefs about the physical world which
he espouses. Despite decades of research on this issue, cognitive
sociologists have yet to produce a single general law which they

———
a ee

218 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

are willing to invoke to explain the cognitive fortunes of any

scientific theory, from any past period. The acceptance of
Boyle’s law, the rejection of Lamarck’s theory of heredity, the
reception of Lyell’s geology, the genesis of Newton’s ideas, the
repudiation of Galenic physiology, the historical career of the
theory of relativity—these are but a tiny sample of the cases
where contemporary sociological theory has failed to provide any
historically significant aids to the understanding. When socio-
logical explanations of specific cases are offered, the reader is
generally left to guess for himself what principles they pre-
suppose.*?
Nor should one be surprised at the exegetical bankruptcy of
contemporary, cognitive sociology of science, for its current
explanatory repertoire is far too crude to permit the kinds of
discriminations that are called for. Whether we talk about social
classes, economic backgrounds, systems of kinship, occupa-
tional roles, psychological types or patterns of ethnic affiliation,
we find that these generally bear no close relation to the belief
systems of majorscientists. Sons of working class men as well as
of noblemen are found among both defenders and detractors of
the Newtonian theory in the eighteenth century; politically
conservative as well as politically radical scientists accept Dar-
winism in the 1870s and 1880s. Followers of Copernican astron-
omy in the seventeenth century represent the entire spectrum of
occupational roles from university don (Galileo) to gentlemen-
soldier (Descartes) to priest (Mersenne), and of psychological
types.
A judicious examination of the historical record seemingly
undercuts the effort to link major scientific theories to any
particular socio-economic group. The Marxists are simply wrong
in speaking of a specifically bourgeois mathematics; the fol-
lowers of Weber have presented no convincing evidence for the
existence of a specifically Puritan natural philosophy; contrary
to fascist ideology, there is no distinctly Jewish physics; against
the claims of many Leninists, we have no evidence that there is
a specifically proletarian version of the special theory of
relativity.
The chief reason for the sociologists’ failure to find a correla-
tion between scientific belief and social class is that the vast
majority of scientific beliefs (though by no meansall) seem to be
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 219

of no social significance whatever. That gravity obeys an inverse

square law, that mechanical energy can be converted into heat,
that atoms have nuclei; such beliefs seem (and I stress the verb)
to have no conceivable social roots or social consequences. Given
the evident conceptuai distance between most scientific beliefs
and the vagaries of social change, it is very different to imagine
how social pressures can have been responsible for the genera-
tion or reception of such ideas. To make matters worse, contem-
porary sociology does little to clarify, even in the abstract, the
mechanisms wherebysocial factors might influence the adoption
of specific scientific ideas. Whether we look to Marx, Mann-
heim, Merton or to any of the other leading sociological
theorists, we are left completely in the dark when it comes to the
specification of a general mechanism for explaining the connec-
tion between social situation and ideological commitment in the
scientific or philosophical sphere. Why should (to take some
standard examples) living in a mercantile society incline one to
favor empiricism? Why should living in a feudal society dispose
one to a geocentric theory of the universe? Why should—to use a
notorious example from Hessen—the fact that Newtonlived in a
seafaring nation cause him to interpret Boyle’s law in the way he
did?*? What evidence we do have suggests that patterns of
scientific belief, both rational and arational, cut across ali the
usual categories of sociological analysis. It is presumably for just
such reasons as these that many contemporary sociologists of
science (such as Ben-David and, in certain moods, even Merton
and Mannheim) hold out little hope for the cognitive sociology
of science. As Ben-David puts it, ‘“The possibilities for . . . [a]
sociology of the conceptual and theoretical contents of science
are extremely limited.’’**
Confronted by the widely acknowledged failure of contem-
porary cognitive sociology to explain any interesting scientific
episodes, we could draw one of two conclusions:
(a) we might conclude that the failure of cognitive sociology of
science is due to the fact that belief-determination in the natural
sciences is instrinsically immune to sociological influences, and
thus to sociological analysis.
Alternatively, we might more charitably suggest that,
(b) there is no reason in principle why arational scientific
beliefs cannot be explained sociologically, provided we can
220 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

develop more subtle theories than we presently have about

the social causation of scientific belief. Many leading sociolo-
gists of science argue for (a), seeing the role of sociology
as entirely non-cognitive, at least so far as the natural sciences
are concerned.
Robert Merton, for instance, in his classic Science, Technol-
ogy and Society in 17th-Century England, specifically disclaims
any ambition to explain the content of seventeenth-century
science in sociological terms, remarking that “specific discov-
eries and inventions belong to the internal history of science and
are largely independent of factors other than the purely scien-
tific.”** Karl Mannheim goes so far as to conclude that historical
developments in ‘‘mathematics and natural science”’ are ‘‘deter-
mined to a large extent by immanent factors.’’** Their argu-
ments for this view are unconvincing, however, because they rest
on that same naive empiricist conception of science and of
scientific rationality which we have discussed before. On the
whole, those cognitive sociologists who exclude science from
their domain do so because of two related convictions, both of
which are seriously wrongheaded:
1. a conviction that scientific theories are dictated by the
data, leaving no room for subjective, nonfactual determinants of
knowledge; as Maurice Richter puts it, “‘society cannot, in
principle, determine the contents of scientific knowledge, be-
cause these are to be determined by observations of nature.’’*’
2. a belief that proper scientific knowledge is self-contained
and insulated from other strands of human belief (e.g., religion,
philosophy, values) which are, in part, socially determined.
It is the conjunction of beliefs (1) and (2) which leads many
thinkers to deny the possibility of a cognitive sociology of
science. To the extent that both images of science are wrong, as
I claim them to be, then there is little warrant for asserting
(a) above. Since it has been established that science interacts
with other disciplines, then, if we could establish that beliefs
in those disciplines were “‘existentially’’ determined, it would
follow as a matter of course that science too, at least to the
extent of its interaction, is (at least indirectly) socially deter-
mined. But even if the denial of (1) and (2) allows for the
possibility of a cognitive sociology of arational scientific beliefs
RATIONALITY AND SOCIOLOGY OF KNOWLEDGE 221

(i.e., (b) above), it must be stressed that far more theoretical

work within sociology itself is required before we can get any
mileage from cognitive socio-history.
If it is true that many sociologists are pessimistic about the
prospects for a cognitive sociology of science, they are generally
far more sanguine about a cognitive sociology of such disciplines
as theology and philosophy. Unfortunately, their record in these
other areas is almost as discouragingasit is in scienceitself. For
instance, in his provocative discussion of the history of episte-
mology, Mannheim observes, with muchjustice, that theories of
knowledge in the seventeenth century were powerfully influenced
by the newly emergingscientific theories of the period. General-
izing this result, he claims that, ‘every theory of knowledge is
itself influenced by the form which science takes at the time and
from which alone it can obtain its conception of the nature of
knowledge.’’** Mannheim then immediately asserts that this
dependence of epistemology on science proves that theories of
knowledge are “socially” determined.*? The only way to make
Mannheim’s inference even seem cogent is by assuming thatit is
not immanent or rational for epistemologies to reflect shifts in
scientific belief. But if we adopt an alternative model of rational-
ity, we can see that it is often entirely reasonable and natural for
a symbiotic interconnection to exist between science and phil-
osophy. The existence of such an interdependenceitself entails
nothing about whetherit is socially caused.
It was argued in the first part of this chapter that the
application of sociological analyses to the history of scientific
ideas must await the prior development of a rational or intellec-
tual history of science; it should be equally clear that the
emergence of a cognitive sociology of knowledge in general
history must also await the articulation of some radically new
tools and concepts of sociological analysis.*° Until both these
logically prior tasks are well under way, pious assertions about
the social determination of scientific belief remain merely gratui-
tous articles of faith.

Conclusion
Through much of this chapter, I have been highly critical of
much work, both theoretical and applied, in the sociology of
222 RATIONALITY AND SOCIOLOGY OF KNOWLEDGE

knowledge. It is crucial to stress, however, that these are

objections to the subject as it is normally practiced. Nothing I
have said here raises doubts about the possibility of the sociology
of knowledge (provided that it works within the framework of
the arationality assumption). To the contrary, large scope is
given in my account for cognitive sociological research. When-
ever, for instance, a scientist accepts a research tradition which
is less adequate than a rival, whenever a scientist pursues a
theory which is non-progressive, whenever a scientist gives
greater or lesser weight to a problem or an anomaly than it
cognitively deserves, whenever a scientist chooses between two
equally adequate or equally progressive research traditions; in
all these cases, we must look to the sociologist (or the psycholo-
gist) for understanding, since there is no possibility of a rational
account of the action in question. We stand badly in need of
sociological theories that can illuminate such cases, which are
undoubtedly frequent in the history of thought. Particularly
promising here would be an exploration of the social determi-
nants of problem weighting, since that phenomenon—probably
more than the others—seems intuitively to be subject to the
pressures of class, nationality, finance and other social in-
fluences.
Equally, we need further exploration into the kinds of social
structures which make it possible for science to function ration-
ally (when it does so). Although nosocial system is sufficient to
guarantee progress and rational scientific choice, certain socio-
political institutions are presumably more conducive to achiev-
ing those ends than others. Once again, however, we must
understand whatscientific rationality is before we can study its
social background.
 Epilogue:
Beyond Veritas and Praxis

1
Of the many questions which this excursion has left unanswered,
at least two require further discussion:
1. even if we grant that the aim of science is problem solving,
and if we grant further that science has been effective at
such problem solving, we are entitled to ask whether an
inquiry system like science—with the techniques it has at
at its disposal—is the most effective possible mechanism
for the solution of problems:
2. we are also entitled to ask whether the investigation of
intellectual problems of the type science studies can be
justified, given the other compelling demands on our
limited mental, physical, and financial resources.
Definitive answers to these questions are not readily within
reach, but one can at least sketch in what directions we need to
move to answer them.
Muchhas beenwritten of the methods of science, yet with the
notable exception of pragmatists like Peirce and certain recent
“systems analysts,’’ no one has seriously investigated whether
the methods utilized by science are the most conducive to
generating solutions to problems. The preoccupationof classical
philosophers of science has been with showing that the methods
of science are efficient instruments for producing truth, high
probability, or ever closer approximations to the truth. In this
enterprise, they have failed dismally. What we now need to ask

223
 924 EPILOGUE

is whether the methods of science—even if they fail as good

ee

“truth machines’’—are the best available tools for the solution of

problems.
That science has solved problems is undoubted; the question
is whether any emendations of the traditional tools of empirical
and logical appraisal would be likely to increase the problem-
solving efficiency of science.
re

This is not the place to propose answers to these global

queries. But we are entitled to urge that the questions them-
selves are serious ones which should no longer be ignored. Until
and unless we can show why science can be an effective
2:
.
instrument for the solution of problems, then its past success at
: problem solving can always be viewed as an accidental piece of
;

good fortune which may, at any time, simply dry up.

But this, in turn, raises the still larger question we mentioned
above: even if science could be shown to be the best tool for the
solution of cognitive problems, how can wejustify devoting such
ample resources to the satisfaction of a peculiar feature of
animal evolution, namely, man’s sense of curiosity?
Classically, the justification for scientific research was two-
fold. It was stressed, on the one hand, that man’s quest for truth
about the world (‘“‘knowledge for its own sake’’) was the driving
force behind scientific inquiry. On the other hand, it was urged
that science has enormous practical, utilitarian value in improv-
ing the physical conditions of life. Both these approaches to the
matter have worn thin. Science does not, so far as we know,
produce theories which are true or even highly probable.
Equally, it is time to acknowledge publicly that the optimistic
Baconian identification of knowledge with power is as ill-
founded in our time as it was when the Lord Chancellor of
England first promoted it some 350 years ago. Much theoretical
activity in the sciences, and most of the best of it, is not directed
at the solution of practical or socially redeeming problems. Even
in those cases where deep-level theorizing has eventually had
practical spin-off, this has been largely accidental; such fortui-
tous applications have been neither the motivation for the
research nor the general rule. Were we to take seriously the
utilitarian approach to science, then a vast re-ordering of
priorities would have to follow, since the present allocation of
 EPILOGUE
225
talent and resources within science manifestly does not reflect
likely practical priorities.
If a sound justification for most scientific activity is going to
be found, it will eventually come perhaps from a recognition that
man’s sense of curiosity about the world and himself is every bit
as confpelling as his need for clothing and food. Everything we
know about cultural anthropology points to the ubiquity, even
among ‘‘primitive’’ cultures barely surviving at subsistence
levels, of elaborate doctrines about how and why the universe
works. The universality of this phenomenon suggests that
making sense of the world and one’s place in that world has
roots deep within the human psyche. By recognizing that solving
an intellectual problem is every bit as fundamental a require-
ment of life as food and drink, we can drop the dangerous
pretense that science is legitimate only in so far asit contributes
to our material well-being or to our store of perennial truths.
Viewed in this light, the repudiation of theoretical scientific
inquiry is tantamount to a denial of what may be our most
characteristically human trait.
To say as much is not to suggest that the expenditure of
resources on all theoretical problems in science is justified tout
court. Far too much scientific research today is devoted to
problems which are as cognitively trivial as they are socially
irrelevant. If the ‘“‘pure’’ scientist is to deserve the generous
support presently being lavished on him, he must be able to
show that his problems are genuinely significant ones and that
his program of research is sufficiently progressive to be worth
gambling our precious and limited resources on it.
 Notes

Prologue
1. Rudolf Carnap, for instance, readily concedes his system of inductive
logic and his confirmation theory to be totally inadequate for dealing with the
more important episodes in the history of science: ‘‘For instance, we cannot
expect to apply inductive logic to Einstein’s general theory of relativity, to find
a numerical] value for the degree of confirmation of this theory. . . . The same
holds for the other steps in the revolutionary transformation of modern physics
. an application of inductive logic in these cases is out of the question”
(my italics; [1962], p. 243). Most other proponents of inductive theories of
rationality have madesimilar disclaimers about their models.
2. Carnap, again, finds himself being forced to the view that the degree of
confirmation (Carnap’s basic measurefor rational acceptability) of all universal
scientific theories is zero, which is precisely the confirmation they would haveif
they had never been confirmed at all! Carnap, in a classic piece of understate-
ment, agrees that this “result may seem surprising; it seems not in accord with
woe
the fact that scientists often say of a law that it is ‘well-confirmed’ .. .”:
([bid., p. 571.)
3. Whether these episodes are genuinely irrational or whether they only seem
to be is an issue to which I shali return in chapter seven.
4. See especially Kuhn (1962), and Feyerabend (1975).
S. For a detailed discussion of Kuhn's views on this question, see below,
pp. 148ff.
6. Cf. Lakatos (1968b), where he struggles valiantly to make the Popperian
theory of rationality germane, and to fit his own interesting ideas into a
Popperian context (where they do not really belong).
7. Although Hintikka has avoided some of the difficulties which Carnap
encountered, he, like Carnap, retains the view that degrees of confirmation are
generally language dependent. This failing is as troublesome and as counter-
intuitive as any of Carnap’s earlier results.

227

he
 228 NOTES TO PAGES 11-21

Chapter1
1. The two apparent exceptions to this claim are Kuhn and Popper, who
both insist that their models of science are based on a problem solving
approach to scientific growth. Unfortunately, such overtures to problems are
only rhetorical. Popper never convincingly shows how the logic of problem
solving relates to any of the technical elements of his philosophy of science
(such as “‘falsifiability’’ or “‘empirica) content’’); Kuhn, for his part, denies that
“the ability to solve problemsis either the unique or an unequivocal basis for
paradigm [i.e., theory] choice’? (Kuhn [1962], p. 168). Both thus take away
with one hand whatthey give with the other.
2. This is not to claim, of course, that philosophers of science have ignored
the fact that science is empirical. But, as we shall see below, there are vast
differences between “explaining empirical data” and “solving empirical prob-
lems.’ Philosophers of science have said too much about the former and
virtually nothing about thelatter.
3. Cf. Oresme (1968), p. 244. ( I am grateful to Dr. A. G. Molland of the
E. R. Institute for this reference.) A fascinating account of some of the ‘‘non-
factual’ phenomena which have been treated as empirical problems by
, scientists is in Martin (1880).
! 4. There are other important technical differences between empirical prob-
lems and facts (such as that a theory always explains an infinite number of
factual propositions but only solves a finite number of problems) which will be
discussed later.
5. My category of unsolved empirical problems corresponds approximately
to Kuhn's notion of ‘“‘puzzle."’ It is important to stress that Kuhn's puzzle
solving view of science embraces nothing other than this class of unsolved
problems.
6. It should be emphasized that this conception of an anomaly is signifi-
cantly different from the conventional one. (See the next sections for a full
discussion of the details.)
7. Once solved by any theory, however, they generally remain as problems
which subsequent theories are expected to solve (at least until they can
convincingly be shown to be pseudo-problems).
8. A propos the problem of Brownian motion, John Conybeare—a contem-
porary of Brown’s—wrote: “I don’t believe a word on't . . . [Biot] states it to be
possible that solid bodies may be compared of [sic] systems of moving
moiecules, representing in small what the planetary systems do in large. I
would only add one supposition more; that these molecules are inhabited, and
have philosophers among their population who . . . believe they have developed
the system of the universe.” This quotation is taken from Mary Jo Nye’s
excellent history of the reception of Brownian motion (1972), pp. 21-22. For
further discussions of this episode, see Brush (1968).
9. See Vartanian (1957).
10. It is worth pointing out that Lakatos’ theory of ‘‘research programmes”
(for ail its stress on competition between theories) cannot explain cases such as

aS
 NOTES TO PAGES 24-47
229
these because materialistic biology did not predic
t the polyp in advance of its
discovery, and thus (on his view) can take no credit for
being able to explain it.
(lL. See chapterfour below, especially pp.
125-27.
: 12. Cf. especially Duhem (1954), Neura
th (1935), and Quine (1953)
.
; 13. Especially Kuhn and Lakatos.
: 14. Popper has comecloseto grasping this
point, (b'), with his requirement
: that any acceptable new theory must be
able to explain everyt
hing which its
i predecessors and competitors can. Unfortunatel
y, however, Popper goes too
far, because in his adherence to (a),
he makes any loss in explanatory conten
i a fatal blow to any theory which exhibi t
ts it. By contrast, I am claiming that
: loss of explanatory content by virtue of a non-re the
futing anomaly counts against
a theory, but not necessarily decisively. For
a fuller criticism of Popper’s (and
Lakatos’) cumulative theory of science, see
below, pp. 147-50 and Laudan (1976b).
i 15. It is important to stress the converse
of this point: if a problem has not
been previously solved by any predecessor
of a theory, then it simply constitutes
an unsolved, not an anomalous, problem
for that theory (with the proviso that
at some fater point the problem may cease
to be a problemaltogether; in which
; case, of course, it would no longer be
anomalous).
16. Indeed, it would probably not be far
wrong to identify the histori cal
emergenceofa science from a proto-scientific state
as that point at whichall its
problems cease to be of the same weight.
17. Home (1972-73) shows convincingly
that Franklin’s treatment of the
Leyden jar effectively diverted attention
away from what had previously been
regardedas the central problems of electri
cal theory. (Cf. especially ibid., pp-
150-51.)
18. Cf. especially Kuhn (1962),
19. See Duhem (1954) and L. Laudan (1965S)
.
20. Many of these claims have been disput
ed by Griinbaum; see especially
(1960), (1969), and (1973).
21. The only way in which some T, which
was a member of the compl ex C
can remove a from among its class of anomalous
instances is by the develop-
ment of an alternative complex C’,
including T,, which can turn the anom-
alous a into a solved problem.
22. And in showing when it is rational to
preserve the entire complex and
ignore the anomaly.

Chapter 2
1. Fora criticism of Kuhn's views on this matter
, see below pp. 150-51, 73-75.
2. Karl Popper, for instance, has often
insisted that the use of metaphysical
or theological beliefs to criticize scientific theori
es is only of “sociologica l”
interest and is in no way germane to the
understanding of rational! evaluation.
In one of his most recent essays, for exampl
e, Popper writes: “the historical and
sociological fact that the theories of both
Copernicus and Darwin clashed with
religion is completely irrelevant for the
rational evaluation of the scientific
theories proposed by them” ({1975], p. 88).
Ina slightly different vein, Philip

he
 230 NOTES TO PAGES 47.59

Frank—confronted by the failure of Renaissance astronomers to. accept

Copernicanism—argues that they made the choice by asking “whetherthe life
of man would become happier or unhappier by the acceptance of the Coperni-
can system” ({(196L], p. 17). Frank allows for no middle ground between a
purely “‘scientific” (i.e., empirical) evaluation, on the one hand, and hedonistic
value judgments, on the other.
3. The most interesting recent exception is Gerd Buchdahl who has
discussed at length (see especially [1970]) the role of controversies about
nonempirical issues in the history of science. My account of conceptual
problems, although different from Buchdahl’s, owes a great deal to his sensitive
treatment of these issues.
4. Heimann (1969-70) utilizes the search for internal consistency as a means
of explaining the evolution of Maxwell's views on electricity and magnetism.
5. It should be noted, however, that the refusal to accept an inconsistent
theory need not require that one cease working on such a theory. (See below,
pp. 180ff.) On the role of internal conceptual! problems in the development
of Thomas Young’s work, see Cantor (1970-71).
6. See Hare (1840), and Faraday’s perplexed response (1840) to this
conceptualcriticism.
7. See especially Stallo (1960).
8. See Whewell (1840), part I]. For an excellent account of Whewell’s
analysis see Butts, forthcoming.
9. The most common form of mutual reinforcement between theories is that
relation usually known as ‘analogy.’ (For an interesting demonstration of how
crucial this sort of analogical problem was in nineteenth century chemistry, see
Brooke [1970-71.)
10. Viner (1928) offers a convincing argument that one of the central
conceptual! problems for Adam Smith’s economic theory wasits incompatibility
with the Newtonian thesis of a balance of forces in nature. The issue was
particularly acute since Smith’s economic theory relied on a general (New-
tonian) balance of nature and yet postulated forces of economic motivation
(e.g., self-interest) which were seemingly incompatible with such a balanced
system. It has been argued that Smith wrote his treatise on moral philosophy in
order to resolve this tension.
11. For instance, any astronomical claim based on telescopic observation
presupposes the acceptability of certain optical theories. The best general
discussion of the conceptual and experimental interdependence of the physical
sciencesis still Duhem (1954).
12. Koyré put the point this way, “‘abstract methodology hasrelatively little
to do with the concrete developmentofscientific thought” ([1956], p. 13).
13. To mention only a few examples: Buchdahi (1969) and Sabra (1967)
have examined the role of methodology in seventeenth century mechanistic
science; Cantor (1971), Olson (1975) and Laudan (1970) have studied the
ee

impact of the epistemology of the Scottish school on reception of physical

theories in the late eighteenth century; McEvoy and McGuire (1975) have
 NOTES TO PAGES 60-75 231
explored the relations between Priestley's methodology and phlogistic chem-
istry; Brooke (1970-71) has analyzed the impact of Comtean positivism
on
nineteenth century French chemistry and physics; Hooykaas (1963) and R.
Laudan (forthcoming) have studied the impact of methodology on geology in the
Lyellian period; Buchdahi (1959), Knight (1970) and L. Laudan (1976a) have
analyzed the methodology of the atomic debates; Hull (1973), Ellegard (1957),
Ghiselin (1969) and Hodge, (forthcoming), have documented the impact
of
methodological ideas on Darwin and hiscritics.
14. See Cantor (1971) and L. Laudan (1970).
15. See L. Laudan (1973b) and (1977).
i
i 16. Buchdahl (1970).
17. McGuire and Heimann (1971).
{ 18. See especially Cotes’ preface to the second edition of Newton's Principia.
| 19. This point is cogently argued in McGuire and Heimann (1971),
i
: 20. For a brilliant study of the role of epistemological and metaphysic
al
issues in eighteenth century embryology, see Roger (1963). Roger’s treatment
of Buffon provides an ideal working model of the type of conceptual historical
analysis for which this chapter seeks to provide a rationale.
21. A contemporaneous example of world-view difficulties can be found in
Culotta’s suggestive study (1974)of nineteenth century biophysics.
22. Some members of this group flatly deny that the evolution of science
owes anything to the broader backdrop of philosophical convictions; others
(such as Duhem) recognize the impactof philosophy on science, but bemoan
it.
23. See above pp. 36-40.

Chapter3
1. Cf. especially Shapere’s excellent critique (1964), and Masterman (1970).
The ambiguity of Kuhn’s analysis has been multiplied as a result of Kuhn’s
later retractions of many of the basic ideasof the first edition of his Structure
of
Scientific Revolutions (1962). Unable to follow the logic of his later changes of
mind, I have been forced to characterize Kuhn’s views in their original form.
2. For a criticism of Kuhn’s theory of “mature” science, see below pp.
150-51.
3. It should be stressed that Kuhn’s notion of “anomaly” is the traditional
one (anomaly=refuting instance) rather than the one I have sketched above,
pp. 26ff.
4. “If any and every failure to fit {the facts] were ground for theory rejection,
all theories ought to be rejectedat all times” (Kuhn [1962], p. 145).
5. As Kuhn originally put it: ‘there are losses as well as gains in scientific
revolutions” ([1962], p. 66). Kuhn, however, is not altogether consistent on
this issue (see below p. 237 n. 18).
6. Shapere (1964).
7. See especially Feyerabend (1970c).
8. Cf. the post-script to the secondedition of Kuhn’s (1962).
232 NOTES TO PAGES 76-83

9. Kuhn (1962), p. 42.

10. Cf. Lakatos (1970), pp. 133-34.
li. [bid., p. 135.
12. [bid., p. 118.
13. See below, pp. 140ff.
14. In spite of the universally acknowledged—and prima facie insoluble—
difficulties facing anyone who would make comparisons of the logical or
empirical content of actual scientific theories, virtually all the recent discus-
sions of scientific growth to have come out of the Popperian tradition—includ-
ing those of Popper himseif, Watkins, Lakatos, Musgrave, Zahar and Koertge
—still assume that the touchstone for scientific progress is increasing content.
1S. Cf. especially Griinbaum (1976a).
16. Despite much special pleading and handwaving, neither Lakatos’ study
of Bohr (1970), nor Zahar’s study of Lorentz (1973), nor the Lakatos-Zahar
study of Copernicus (1975) utilizes Lakatos’ “‘official’’ theory of progress. At no
point do they show those relations of content-inclusion which are crucial to
progress (in Lakatos’ sense).
17. Nor can Lakatos use those assessments for explaining the actions of
scientists since he denies that anything except retrospective autopsies of long-
dead scientific controversies can producea reliable assessment.
18. I may be unfair to Lakatos here, for he wildly equivocates on this issue.
On the one hand, he insists that the unfalsifiable hard core of a theory is one of
the central features of a research programmefrom its inception. On the other
hand, hetells us that ‘‘the actual hard core of a programme does not actually
emerge fully armed . . . [it] develops slowly’’ ({1970], p. 133n). If, in fact, the
hard core cannot be identified during much of the history of a research
programme, then how doscientists know what to hold dear when confronted
by an anomaly?
19. In her study of eighteenth century mechanics, Iitis (1972-73) seems to
find it strange that scientists who accepted Newton’s or Leibniz’s mechanics
also tended to accept the ontology, the methodology, and even the theology,
associated with these theories. The doctrine of research traditions makes this
surprising phenomenon completely natural and unsurprising.
20. For a useful account of seventeenth-century optics, see Sabra (1967).
21. McKie and Partington (1937-39).
22. For a discussion of what is involved in abandoning truth and falsity as
determinable characteristics of research traditions, see below pp. 124ff.
23. Historians who focus on specific theories, rather than the larger
traditions of which they are a part, often find themselves perplexed by, and
unable to explain, the reception of such theories. Such perplexity often
dissolves if these theories are seen in a larger context. For instance, Alan
Shapiro's excellent study of seventeenth-century wave optics (1973) ends with a
“paradox”; Huygens’ theory of light, as Shapiro rightly argues, was the only
theory available at the time which could explain the double refraction in
Iceland spar. Why, then, asks Shapiro, was Huygens’ approach so totally
NOTES TO PAGES 84-97 233
ignored for the next century and why did scientists remain committed to a
Newtonian approach (which could not do justice to the problems posed by
double refraction)? Shapiro offers no answer. Surely a part of the answeris to
be found in the fact that Huygens’ theory—although it could (even this was
sometimes doubted) handle Iceland spar—was found wanting because it
refused to addressitself to, or to offer any solutions for, most of the important
problems in late seventeenth-century optics. (For instance, it did nothing to
solve the problems of colors or Newton’s rings.) Equally, it was thought to
suffer from some serious anomalies (e.g., its inability to explain the sharp lines
around shadows). If we add to this the fact that Huygens’ optical work was
allied to a broader Cartesian tradition in optics—a tradition farless progressive
than Newton’s—it is not surprising that Huygen's Traité de la lumiére “rapidly
sank into oblivion” (Shapiro [1973], p. 252). One could even go so far as to say
that Huygens’ theory was not taken seriously because, given the flaws
mentioned above, it did not deserve to be taken seriously.
24. Cf. Brown (1968).
25. As we have already indicated, the methodology of the uniformitarian
tradition in geology (as developed by Hutton, Playfair, and Lyell) decreed that
all the problems of cosmogony—which had previously been regarded as
geological problems—would no longer haveto be solved by geologists.
26. For an interesting account of the fortunes of aether theories in the late
nineteenth century, see Schaffner (1972). For a discussion of empirical
problemswhich ‘‘vanish,” see Griinbaum (1976a).
27. See especially Cantor (1971).
28. Cf. L. Laudan (1970), (1973b), and (1977).
29. Lakatos has been misled by this feature of research traditions into
thinking that empirical anomalies are of virtually no significance to the
development of science. Quite the reverse is the case, for at least two reasons:
a. It sometimes happensthat the heuristic capacity of a research tradition is
too weak to allow it to accommodate certain anomalies, and its failure to deal
with them convincingly weighs heavily against it.
b. Even when a researchtradition is sufficiently fecund to provide guidelines
for transforming some anomatous problem into a solved one, the existence of
the anomaly is historically crucial if we wish to understand why the theories
within a research tradition exhibit the sequential character they do. Contrary to
Lakatos’ a priorism, the order of theories constituting a research tradition will
mirror, at least partially, the order in which different anomalies emerged.
30. There are unmistakable ambiguities in Lakatos’ treatment of this
question. On the one hand, Lakates characterizes a research programme
chiefly in terms of its so-called hard core, i.e., those doctrines which are so
crucial to the programmethat noscientist within the programmewill abandon
them. On the other hand, Lakatosinsists that ‘the actual hard core does not ac-
tually emerge fully armed . . . it develops slowly, by a long, preliminary process of
trial and error’ ({1970], p. 133n). This latter approach suggests that research
programmes have no “hard core” in their early stages; but if that is true, then
234 NOTES TO PAGES 98-113

how can Lakatos identify research programmes in their infancy, since that
identification depends upon specification of the contents of the hard core? (Cf.
note 18 above.)
31. For an illuminating analysis of the manner in which the core assump-
tions of a research tradition can undergo radical transformation, see Brown's
study (1969) of theories of the electric current in the early nineteenth century.
32. As Hull has cogently argued “no degree of similarity between earlier and
later stages” in the developmentof an historical ‘‘object’’ such as a research
tradition is necessary in order for it “to remain the same entity” ([1975},
p. 256).
33. Despite Lakatos’ contempt for the method of trial and error, his only
explanation for the emergence of the core of a research tradition is that it
results from ‘‘a long, preliminary process of trial and error’’ ({1970], p. 133n).
34. Indeed, if Forman (1971) is right, the abandonment of strict determin-
ism in modern quantum mechanics was prompted by the irreconcilability of
classical physics with the general world-view.
35. (1961), p. 191.
36. Schofield (1970).
37. See above, pp. 66-69.
38. Myanalysis here owes muchto discussions with Adolf Griinbaum.
39. I find it very difficult to pin down precisely what Kuhn’s views on this
issue actually are. Consider, for instance, the following remark: ‘Though the
historian can always find men—Priestley for instance—who were unreasonable
to resist [a new paradigm] for as long as they did, he will not find a point at
which resistance becomesillogical or unscientific.”” (Kuhn [1962], p. 158). The
first half of the passage suggests that there are criteria for determining
whether the acceptanceor rejection of a paradigm is rational; whereas thefinal
clause denies that there is any point at which that acceptance becomesrational
(assuming, as I think weare entitled to, that Kuhn is here using ‘‘unreason-
able,” “‘illogical,”’ and ‘unscientific’ as approximate synonyms). Butif there is
no point at which the acceptance (or rejection) of paradigm becomes reason-
able, how could we decide—as Kuhn has—that Priestley was ‘‘unreasonable”’
in rejecting Lavoisier’s paradigm?
40. Like Feyerabend, Kuhn recognizes that there is a context of pursuit and
denies that there are usually any rational grounds for pursuing a new theory
which has not yet been well confirmed: ‘‘the man who embraces a new
paradigm at an early stage must often do so in defiance of the evidence
provided by problem-solving [success] . . . A decision of that kind can only be
madeon faith’ (Kuhn [1962], p. 157; myitalics).
41. In a famous paper published in 1813, the Swedish chemist Berzelius
discussed many of the anomalies for Daltonian atomism. However, precisely
because ‘“‘it would be rash to conclude that we [atomists] shall not be able
hereafter to explain these apparent anomalies in a satisfactory manner”’
({1813], p. 450), Berzelius did not urge non-pursuit of the atomic theory even
though, within the context of acceptance, ‘‘the hypothesis of atoms can neither
be adopted nor considered as true” (Ibid.) Cf. also Berzelius (1815).
NOTES TO PAGES 114-127 235

42. See, for instance, A. Griinbaum (1973), pp. 715-25, 837-39; 1. Lakatos
(1970); E. Zahar (1973), especially 100ff.; K. Schaffner (1974), especially
78-79; and J. Leplin (1975), A thorough historical analysis of the evolution of
the notion of adhocness would probably show that the idea originated at a time
when scientists and philosophers believed: (1) that the constituent parts of a
theory could be tested in isolation; and (2) that only directly observable entities
could be legitimately postulated within a theory. Most philosophers and
scientists have now abandoned both (1) and (2), yet continueto believe that the
requirement of independent testability is still legitimate. Whether the con-
tinued demand for the latter makes any sense given the repudiation of the
simple-minded philosophy of science which originally motivated it is an open
question. [Grtinbaum’s 1976b appearedtoo late for meto discuss it here. —Au.]
43. See the writings of Lakatos and Griinbaum cited above, as well as the
relevant sections of Karl Popper (1959) and (1963).
44. Cf. Griinbaum (1973), p. 718. (Although this useful clarification is due
to Griinbaum, it does not represent his own approach to the problem.)
45. Cf. above pp. 40-44.
46. A fuller treatment of this problem is in L. Laudan (1976b).
47. Such a context—and comparison—dependent sense of ad hoc is dis-
cussed sympathetically in Griinbaum (1973).
48. Utilizing the machinery outlined above, pp. 66-69.
49. Zahar, for instance, speaks of a theory being ad hoc ‘“‘if it is obtained
from its predecessor through a modification of the auxiliary hypotheses which
does not accord with the spirit of the heuristic of the [research] programme”
({1973], p. 101; myitalics). On anotheroccasion, he suggests that a theory is ad
hocin this sense if it “destroys the organic unity of the whole nexus” (Ibid., p.
10S). Zahar may have clear criteria for these processes, but he never unpacks
what it would mean to be out of ‘accord’ with ‘‘the spirit of a programme’s
heuristic’ or to destroy its ‘‘organic unity.”” Schaffner is slightly more specific,
suggesting that theories can encounter “‘trans-empirical” difficulties such as
“complexity’’ or “theoretical discord’’; but until these notions are further
developed one cannot be sure whether Schaffner has in mind the samesort of
analysis for which I have argued here.

Chapter 4
1. For a discussion of some of the weaknesses in classical theories of
self-correction and truth-approximation, see L. Laudan (1973a). A devastating
critique of Popper's theory of verisimilitude is Griinbaum (1976c).
2. Maxwell has attempted to defend the view thatit is rational to seek a goal
(such as truth) ‘“‘even though we have no rational assurance whatsoever that the
aim will meet with success” ({1972], p. 151). It is just such an argument as this
which lies behind beliefs in immortality, the philosopher's stone and El
Dorado. It argues that quixotic quests are always rational until we can prove
them to be otherwise. Surely the burden of proof is precisely reversed; hunting
236 NOTES TO PAGES 129-146

the snark does not become rational just because we have not yet proved its
nonexistence!
3. Scheffler (1967), pp. 9-10.
4. Evidently in fear and trembling lest the incorporation of these evolving
standards into a model of rationality might deprive it of its supra-temporal
(“third-world”) status, they have deliberately repudiated the use of such
notions, taking refuge, rather,in what they imagine to be non-‘‘time-depen-
dent propertfies}"” (Zahar [1973], p. 242n.; see also Lakatos [1970], p. 137)
such as “‘mathematical coherence.’’ Leaving aside the dubious contention that
conceptions of mathematical coherence themselves have not evolved, one
wonders what point there is in maintaining that all the significant meta-level
characterizations of science have been static since the garden of Eden.
5S. This model thus allows us to have best of both worlds; we can
acknowledge that specific standards of rationality have evolved, without
surrendering our capacity to make normative judgments about the past. It is
not uncommon to find in the sociological literature a distinction (similar to the
one I have sketched) between rationality within a given context of belief and
what is frequently called ‘‘transcendent rationality” (See, for instance, Winch
{1964} and Lukes {1967].) What has not been suggested before, so far as I can
determine, is that there is a third, hybrid sense of rationality which allows us to
make transcendent judgments about the rationality of beliefs without ignoring
the crucial particularities of context.
6. “It is precisely the abandonment of critical discourse that marks the
transition to a science. . . [thereafter] critical discourse recurs only at moments
of crisis when the bases of the field are again in jeopardy” (Kuhn [1970], pp.
6-7).
7, Truesdell (1968), the well-known historian of eighteenth-century mechan-
ics, does his best to play down manyof these issues, particularly those which
are not mathematical in character. Costabel (1973) and Aiton (1972) give much
more sensitive accounts of some of the philosophical issues at stake in the
mechanicsof the Enlightenment.
8. Concerning ontology, cf. especially McGuire and Heimann (1971) and
Schofield (1970). On methodology, L. Laudan (1973b) and (1976). See also
above, pp. 57-61.
9. Cf. Kuhn (1962), p. 10.
10. Kuhn’s cynical view is that scientific revolutions are regarded as progres-
sive because the “‘victors’”’ write the history and they would hardly view their
own successes as anything but progressive. (Cf. especially his [1962], pp. 159ff.)
Here, as elsewhere, Kuhn slides too readily between political and cognitive
characterizations of science.
11. See below, pp. 147-50.
12. There are excellent summary discussions of the difficulties of the implicit
definition theory of meaning in Suppe (1974), pp. 199ff, and in Shapere (1966).
13. Kuhn (1970), p. 266.
14. If the theoretical assumptions are inconsistent with the theory under
analysis, then the problem will become a “pseudo-problem.”’
15. In making these determinations, we would have to restrict ourselves, of
NOTES TO PAGES 146-149 237

course, to those problems and anomalies capable of being expressed within the
framework of the research tradition under scrutiny and would have to ignore
rival and (by hypothesis) incommensurable research traditions. The possibility
of assessing these variables does depend upon translations being possible
between the theories that constitute a research tradition.
16. My approach to the problem of incommensurability resembles that of
Kordig (1971), insofar as we both argue that there are methodologicalcriteria
for theory comparison, even when substantive translation between different
theories is inappropriate. Kordig and I differ drastically, however, about what
these methodological criteria should be. Following Margenau, Kordigstresses a
comparison of theories with respect to their empirical confirmation, their
“extensibility,”’ their ‘‘multiple connection,” their simplicity and their ‘“‘causal-
ity”; unfortunately, most of these remain entirely intuitive notions in Kordig’s
discussion, and it is to be hoped that he will refine them into the sensitive
instruments of analysis needed for the comparative appraisal of theories.
17. The cogency of this argument does not rest on the acceptance of the
model outlined in this essay. Any model of rationality which offers a method of
determining an appraisal measure of scientific theories without inter-theoretic
translation can avoid the difficulties of incommensurability.
18. Here, as elsewhere, Kuhn is ambivalent. On the one hand, he stresses
the non-cumulative character of science by insisting that there are always
problem losses as well as gains in every case of paradigm replacement. (See
above, p. 231 n. 5.) Yet, on the other hand, he claims that: ‘‘A scientific
community will seldom or never embrace a new theory unless it solves all or
almost all the quantitative, numerical puzzles that have been treated by its
predecessor" ({1970}, p. 20).
19. Collingwood (1956), p. 329; my italics. Elsewhere, Collingwood reiter-
ates this claim: ‘Progress in science would consist in the supersession of one
theory by another which served both to explain all that the first theory
explained, and also to explain . . . ‘phenomena’ which the first ought to have
explained but could not. . . . Philosophy progresses in so far as one stageofits
developmentsolves the problems which defeatedit in the last, without losing its
hold on the solutions already achieved” ([1956], p. 332).
20. Popper (1963). As he puts it elsewhere: ‘“A new theory, however
revolutionary, must always be able to explain fully the success of its predeces-
sor. In all those cases in which its predecessor was successful, it must yield
results at least as good . . .”” ({1975]}, p. 83).
21. Cf. Lakatos (1970), p. 118.
22. Post (1971), p. 229. Cf. also Koertge (1973). Phenomenological theories
of progress, every bit as much as positivistic and idealistic ones, are committed
to the cumulativity postulate. For a detailed example, see Harris (1970),
especially pp. 352-69.
23. Cf. especially Kuhn (1962), p. 169.
24. This was pointed out, amongothers, by Berzelius (1815).
25. Home’s study (1972-73) makes it quite clear that Franklin realized this
failure of his theory, but did not regardit as sufficient grounds for rejecting it.
One might add that Franklin’s theory also failed to give any solution at al! for
238 NOTES TO PAGES 150-159
the fact—widely observed and explained before his time—that there was
generally a correlation between the density of a substance and its capacity to
act as an electrical conductor.
26. We can illustrate what is involved by an example. Suppose that our
scientific aim is to understand the embryology of birds. We have one theory,
T,, which offers a detailed account of the embryological developmentof eagles
and egrets. We have another, 7,, which explains the embryological develop-
ments of ali birds smaller than eagles, including egrets, but does not work for
eagies. In such a circumstance, we would certainly view T, as preferable to
(i.e., a progressive improvement on) T,, even if T, was not able to solve the
problem about embryonic development for eagles. Such a plausible judgment
would be disallowed on almost all the standard (cumulative) theories of
scientific progress. (For a fuller treatment of these issues, cf. L. Laudan
[1976b].)
27. For the relevant discussions see Lakatos (1970), pp. 137, 175-77, and
Kuhn (1962), pp. 11ff. and (1968).
28. Lakatos—at best—has shown how a programme could conceivably be
progressive, while ignoring many anomalies; but that is a far cry from the
stronger claim—required by his theory of mature science—that such anomaly-
ignoring programmesare ipso facto more progressive than programmes which
pay serious attention to their anomalies.
29. Given Lakatos’ (Kuhnian induced) aversion to anomalies, he would
probably have regarded this very feature of the dichotomy as a bonus. For those
of us who do not share his views on the irrelevance of anomalies and criticism,
however, such untestability must countas a serious liability.
30. It is worth pondering what motivates the search for a distinction between
immature and mature science. My guess is that the quest harkens back to the
old inductivist-positivist conviciton that “proper’’ science only began with
Galileo, Newton, and the other classic heroes of the seventeenth century.
Although eschewing inductivism, both Kuhn and Lakatos propose a demarca-
tion criterion between mature and immature science which resurrects the
inductivists’ search for a definite point in time at which science became
genuinely ‘‘scientific.” (For a lengthy illustration of a historian’s efforts to
write about the history of science by utilizing such a demarcation criterion, see
Gillispie’s whiggish [1960].)

Chapter 5
1. Agassi (1963).
2. Griinbaum (1963).
3. For a guide to muchofthis literature, see Suppe (1974).
4. With the exception of Lakatos who is committed to this thesis. (See
below, p. 165.)
5. Giere (1973).
6. Ibid., p. 292.
NOTES TO PAGES 159-169 239

7. Ibid., p. 293.
8. Ibid.
9. Ibid., p. 290.
10. Most philosophers of science ultimately do fall back on such a class of
“privileged intuitions’ about specific episodes as the final arbiter. Popper, for
instance, writes: “‘It is only from the consequences of mydefinition of empirical
science, and from the methodological decisions which depend upon this
definition, that the scientist will be able to see how far it conforms to his
intuitive idea of the goal of his endeavors’ (my italics; Popper [1959],
p. 55).
11. For a detailed exploration of those issues, see McMullin’s valuable
discussion (1970).
12. Cf. Lakatos’ claims that: (1) “All methodologies . . . can be criticised by
criticising the rational historical recontructions to which they lead" ({1971],
p. 109); (2) “A rationality theory. . . is to be rejectedif it is inconsistent with an
accepted ‘basic value judgment’ of the scientific élite’ ({1971], p. 110); (3)
“|. . better rational reconstructions . . . can always reconstruct more of actual
great science as rational” ({1971], p. 117); and, more explicitly, (4) ‘‘Thus
progress in the theory of rationality is marked... by the reconstruction of a
growing bulk of value-impregnated history as rational” ({1971}, p. 118).
13. Although Lakatostries to avoid this dilemma (saying that no theory of
rationality ‘“‘can or should explain ail history of science as rational” [1971], p.
118), it follows inevitably from his method of ranking theories of rationality
that the best such theory is that which ‘‘rationalizes” the largest part of the
history of science.
14. The bulk of this section is concerned with the role of norms in the history
of scientific ideas. The other main branch of the subject, the social history of
science, likewise utilizes normsof rationality, but in different ways than the
history of ideas. These issues are discussed below, pp. 184ff, 201 ff.
15. Agassi (1963).
16. For a candid voicing of such anxieties, see Cohen (1974).
17. The “ahistoricity” of these philosophers is pointed out by McMillin
(1970), Machamer (1973), McEvoy (1975), and Beckman (1971).
18. Although implicit in much of his work, this doctrine is most explicitly
formulated in Lakatos (1971). The method of rational reconstruction began
initially as philosophical technique for shedding light on the nature of rational
deliberation and decision making. In its original conception, it involved
postulating contrived and artificial cases of choice, which were deliberately
simplified in order to get a handle on the case; these over simplified cases were
then to be rendered more applicable to the actual situation by the gradual
addition of complicating factors.
19. Ibid., p. 91.
20. [bid., p. 106.
21. Similarly, Tornebohm claims, in his ‘‘rational reconstruction’ of seven-
teenth century astronomy, that “historical accidents [sic] which affected the
growth of this knowledge are not of interest . . . 1 will therefore take the
240 NOTES TO PAGES 169-176

liberty of making a reconstruction of the historical development. The cast

consists of two people whom I have invented . . .”’ ({1970], p. 79).
22. Lakatos, (1971), p. 107.
23. Ibid.
24. Ibid., p. 106.
25. A similar example of the dubious historical relevance of the techniques
of rational reconstruction can be found in Watson's book-length study (1966) of
the downfall of Cartesianism. Watson’s procedureis to define ‘‘a model of late
seventeenth century Cartesian metaphysical system,’’ whose weaknesses he
proceeds to explore. Watson attributes the downfall of Cartesianism to the
failure of this ‘‘model’’ system to come to terms with serious weaknesses it
exhibited. What is curious is that Watson freely acknowledges that “none of
the Cartesians . . . professed a system of exactly the sort’’ defined by his model
({1966], p. 29). Given that no actual Cartesian accepted the Watsonian
reconstruction, Watson’s lengthy analysis cannot explain why the genuine
Cartesian philosophy was abandoned. Watson’s discussion of the logical flaws
in his imaginatively ersatz version of Cartesianism, for all its suggestiveness,
never becomes authentic history.
26. Lakatos (1971), p. 107.
27. Ibid., p. 108.
28. Ibid., p. 107.
29. In fact, of course, there is not even strong similarity here, for the
reconstructionist is not appraising the rationality of historical episodes, but
feigned ones.
30. As already observed, it is probably the predilection of many “historically
oriented” philosophers (from Hegel to Lakatos) for the cavalier method of
rational reconstruction which makes most historians so suspicious of philo-
sophical attempts to deal with the history of thought.

Chapter 6
1. See especially the discussion of conceptual problems in chapter two.
2. Kuhn (1968), p. 8&1.
3. Ibid.
4. Kuhn’s beliefs about disciplinary autonomy are widely shared among
historians, of both the ‘‘old” inductivist and the ‘“‘new” socially oriented
school. For references to some of the relevant literature, see below, pp.
213-17.
5. Hodge’s study of the evolution of Lamarck’s ideas (1970-71) exhibits
vividly how important it is to attend to the problems a scientist is trying to
resolve. Hodge points out that a widespread misconstrual of Lamarck’s
problem situation has fed many historians to misinterpret the whole thrust of
his theoretical research. (For a similar analysis of Chambers’ work, cf. Hodge
[1972).)
6. Cf. especially Gilson (1951) and Popkin (1960).
amet, thalll

NOTES TO PAGES 176-179 241

7. Compare Karl Jaspers: “The great philosophers . . . are best approached

as contemporaries. . . . We shall understand them best by questioning them,
side by side, without regard for history and their placein it’ ({1962], p. xi).
8. This is not to say, of course, that there is nothing in common between the
three. But historical understanding very often depends upon our ability to
recognize that, in the course of time, problems undergo subtle, and sometimes
profound, changes of both formulation and substance. As Quentin Skinner
: aptly remarks: ‘It is this essential belief that each of the classic writers may be
j expected to consider and explicate some determinable set of ‘fundamental
j concepts” of “perennial interest’ which seems to be the basic source of the
confusions engendered by this approach to studying the history of either
literary or philosophical ideas”’ ([1969], p. 5).
9. Nelson writes: “The history of philosophy itself is the succession of
increasingly successful! solutions of these [unchanging] problems” ({1962],
p. 22).
10. See especially Collingwood (1939).
11. Ibid., p. 70. -
12. The very possibility of answering such questions is denied by one of the
more banal movements now popular in intellectual history, specifically, that
i form of structuralism associated with the work of Michel Foucault, especially
i (1970). For our purposes, the two chief flaws in Foucaultian historiographyare:
i (a) its completely stochastic character. The ‘‘archaeology of ideas” (Foucault's
version of intellectual history) offers no means, indeed denies the possibility, of
one
; ever giving a coherent acount of how world views (‘‘epistemes’’) give way to
the
another, or of their mutual interconnections. Because Foucault insists that
i emergence of new conceptual systems are the result of “ruptures of human
| consciousness,” there can be no explanation—neither intellectual nor socio-
i economic—of the processes whereby new epistemes displace older ones. The
allegedly
next flaw is (b) its vague invocation of the Zeitgeist. Although
for the
eschewingtraditional categories of historical analysis, Foucault's search
thought
: common structures and metaphors which (on his view) permeate the
of any epoch harkens backto the old, oft discredited belief that ideas ‘‘in the
i air’ and “the collective consciousness” are the appropriate causal modalities
for the historian. To understanda classic text, for Foucault, is neither to relate
it to the biography of its author nor to examine the arguments within it; rather
the
the historian studies such texts in order to find out what theytell us about
ss of an era. With its twin emphases on the mystery and
(linguistic) consciousne
the opacity of human thought, with its stress on “history as poetry,’ Foucau!-
tian structuralism must rank as one of the most obscurantist historiographical
fashions of the twentieth century. It says something about the state of mind of
manyintellectual historians that they are prepared to pay obeisance to a work
like Foucault’s which they generally concede to be unintelligible. Like Bergson
Anglo-
and Teilhard before him, Foucault has benefited from that curious
a profundity
American view that if a Frenchman talks nonsense it must rest on
whichis too deep for a speaker of English to comprehend.

i
242 NOTES TO PAGES 179-201

13. I have tried to give some preliminary answers to these questions in L.

Laudan (1973a), and (1977).
14. Cf. Holton (1973), especially chapters one and three. Holton claims to
have identified most of the core concepts (“‘themata”’) ever to have occurred in
the history of science and “‘suspect[s] the total will turn out to be less than 100”
({1975], p. 331).
15. See above, chapter three.
16. Skinner (1969).
17. These profiles correspond roughly to exegetical or descriptive history.
18. And they have been massively discredited by experience.
19. Cf. Lakatos (1963).
20. Popperis entirely typical in arguing that “‘in science (and only in science)
can we say that we have made genuine progress: that we know more than we
did before”’ (Popper [1970], p. 57).
21. It was one of Lakatos’ genuine insights that this classical sacred cow of
the philosophers had to be abandoned before one could develop an adequate
theory of rationality. For examples of this approach, see Lakatos (1968) and
L. Laudan (1973a).

Chapter 7
1. Although the bulk of this chapter will focus specifically on the sociology of
knowledge, most of its conclusions also apply, mutatis mutandis, to the
psycho-history of ideas.
2. For instance, unless a scientist believes in subatomic particles, he is
hardlylikely to join a laboratory doing research on the structure of the nucleus!
3. See, for instance, Scheler who asserts that ‘‘the sociological character of
all knowledge, all forms of thought, intuition and cognition is unquestion-
able” (Quoted in Merton [1949], p. 231).
4. A propos these two extremes, it is more than a little ironic that
Mannheim, who chastizes the “‘older’’ intellectual historians for making the a
priori assumption ‘‘that changes in ideas were to be understood on thelevel of
ideas’ ((1936], p. 268), is himself committed—in what can only be described
as an equally a priori fashion—to the view that virtually a// changes in ideas are
“boundup with social existence”’ (Ibid., p. 278).
5. Mannheim grappled with this problem (unsuccessfully) through most of
his career. On the one side, he wanted to insist that sociology had shown the
social origins of virtually all systems of belief, including sociology itself: “Once
we have familiarized ourselves with the conception that the ideologies of our
opponents are, after all, just the function of their position in the world, we
cannot refrain from concluding that our own ideas, too, are functions of a social
position” ({1952], p. 145). On the other hand, as Mannheim gradually
realized that such a view would vitiate the claims of sociology to possess
NOTES TO PAGES 201-209 243

objective validity (and perhaps under pressure from the arguments of Alfred
Weber), he began to argue that the thinkers—such as himself—were often
immune from social influences and he developed the notion of ‘‘the relatively
socially unattached intelligentsia’ (Ibid., pp. 252ff.). But if the intelligentsia
can transcend social determination, and if the history of ideas is mainly
concerned with the intelligentsia, what scope—even on Mannheim’s account—
is left to cognitive sociology?
6. Grtinwald (1934), p. 229.
7. For an articulation of this distinction, see especially Mannheim (1936),
chapterfive.
8. It may be, of course, that the acceptance of beliefs y and z is a function of
social factors, in which case we might say that the acceptance of x (rationally
dictated by y and z) is indirectly the result of the social situation. But this does
not controvert the claim that the most direct and most fundamental explana-
tion for the acceptance of x by some thinkeris that it follows rationally from
y and z,
9. Merton (1949), pp. 516, 558. For Mannheim’s formulation of this
assumption, cf. (1936), p. 267.
10. It is as true today as it was when Mannheim pointed it out in 1931, that
“the most important task of the sociology of knowledge . . . is to demonstrate its
[explanatory] capacity in actual research in the historico-social realm’ (Ibid.,
p. 306).
11. For a discussion of some of these conceptual problems see below pp.
217ff.
12. A point similar to this was made by Imre Lakatos when he wrote,
“internal history [of science] is primary, external history [of science] only
secondary, since the important problems of external history are defined by
internal history” ([1971], p. 105). What handicaps Lakatos’ analysis is a failure
to recognize the difference between cognitive and non-cognitive attempts to
deal with the history of science. Although we are entitled to say that the
“important problems” of cognitive sociology are, as it were, defined by the
rationa! history of science, it is manifestly untrue to believe that the ‘important
problems” of non-cognitive sociology are, to any significant degree, defined by
the so-called internal (or rational) history of science.
13. See, for instance, Kuhn’s remark quoted above p. 234 n. 40.
14. Richter (1973), p. 81; myitalics.
15. Ibid., p. 6.
16. Barber (1962).
17. Barber, for instance, speaks of Kelvin’s ‘‘blindness’’ in opposing
Maxwell’s theory of light because the latter was not sufficiently mechanistic
(Ibid., p. 540). One may, with the advantage of hindsight, quibble with
Kelvin’s search for mechanical models; but in the historical circumstance there
was nothing blind or irrational about Kelvin’s initial reaction to Maxwell's
work.
18. Mannheim effectively concedes this point in (1952), pp. 181f.
244 NOTES TO PAGES 211-216

19. What I am calling contextual beliefs are more commonly called

“existentially’’ or “‘situationally determined beliefs.’ I have avoided the latter
terminology because it needlessly conjures up images of nineteenth-century
German academic philosophy which areirrelevant to the case at hand.
20. Mannheim (1936), p. 272. Also, see pp. 265, 266, 271f.
21. Cf. Mannheim (1936), pp. 264-299 passim.
22. Myitalics; quoted by Merton (1949), p. 232.
23. For some examples, see below, pp. 220, 221.
24. If my inclusion of Durkheim seems peculiar, one need only recall his
argument that anytime the acceptanceor rejection of concepts is determined by
their compatibility with prevailing beliefs, then we must be dealing with a
“sociological process.””
25. Brown (1970), p. 29.
26. Forman: “it was only as and when this romantic reaction against exact
science had achieved sufficient popularity inside and outside the university to
seriously undermine the social standing of the physicists and mathematicians
that they were impelled to cometo terms with it” ([1971], p. 110).
27. Compare, for instance, Kuhn’s views about “the unparalleled insulation
of mature scientific communities from the demands of the laity and of everyday
life’’ ([1962], p. 163). See also my discussion of Kuhn’s views on disciplinary
autonomyabove, pp. 173-75.
28. This belief that all intellectual conflicts and debates are in essence, a
sublimated form of social conflict, permeates the work of many historians of
science. As the social historian Steven Shapin puts it, the “‘good’’ historian
must “‘try to assimilate conflict in ideas to conflict among competing groups in
society” ([1975], p. 221). It is hard to regard this belief (or related ones such as
“scientific disciplines are reactionary,” ‘‘scientists only worry about philosophy
when their prestige is threatened,” “influences of the cultural environment on
science must be caused by social factors,’’ ad nauseam) as anything other than
purely a@ priori prejudices, since none of the historians who subscribe to them
ever offers even the pretense ofa justification for them. (For a detailed critique
of some of Shapin’s views, see Cantor [1975b].)
29. Forman (1971), p. 6. Confronted by bald assertions of this kind, it is
difficult to resist the ad hominem hypothesis that social historians are massively
engaged in projecting their own disciplinary insecurities onto the history of
science, convinced that scientists are as sensitive to questions of prestige as
these historians evidently are.
This criticism is more than purely rhetorical. As Mannheim concedes, the
entire discipline of the sociology of knowledge emerged as a generalization from
the features of sociology itself. Early twentieth-century sociologists, examining
the history of their own discipline, came to the conclusion that it was full of
doctrines which owed more to the social background of their defenders than to
their intrinsic rational merits. The general thesis of the sociology of knowledge
(to wit, that ideas in most disciplines are socially determined) was founded on
the hope that all other forms of knowledge might prove to be as subjective
as sociology clearly was.
 NOTES TO PAGES 216-221 245
One sees this phenomenon in microcosm as well as in
macrocosm by
examining some of the more candid statements of working social
historians of
! science. Steven Shapin, for instance, seeks to justify the reductio
n of scientific
theory-choice to straightforward cases of social conflict by arguing that
we
usually seek in “everyday” life to explain “people’s behavior and
motives”
({1975], pp. 220-224) by reducing them to social causes, rather than paying
attention to the reasons people give for their actions and beliefs. Can Shapin
really believe that in ‘‘everyday” life, we never conceive that people believe
things because they have good, nonsocial reasons for doing so? Can he be
serious when he maintains that the social motivations of belief are “relative
ly
familiar and known’ when contrasted with the intellectual motivations for
belief? In a different vein, Thackray (1970) urges that history of science
must
become more sociological and less intellectual in order to gain esteem
in the
eyes of general historians, sociologists, and campus radicals!
Virtually every conceivable reason for doing sociology of science has been
rehearsed in the recent literature, except the argument that sociology might be
able to offer some convincing explanationsof important historical situations.
30. For all their opposition to “‘whiggish history,” and to looking at the past
through the spectacles of the present, Kuhn, Forman, and Brown areall guilty
of projecting into the past a conception of disciplinary autonomy andinsularity
which derives from generalizations about present-day science. No conscien
tious
survey of seventeenth, eighteenth, or nineteenth century science could have
produced the Kuhn-Brown-Forman view that, as Forman puts it, “when
scientists and their enterprise are enjoying high prestige . . . they are also
relatively free to ignore the specific doctrines . . . which constitute the
i
correspondingintellectual milieu’ ({1971], p. 6).
3
i 31. It is revealing that when Forman’s sociological modelfails to explain the
beliefs of scientists (as he admits it does in certain cases), he insists that
3
we
must look for some “psychological” explanation for whya scientist resisted
the
social forces upon him, rather than looking for some rational account of the
scientist's belief. (Cf. especially Forman [1971], pp. 114-115.)
32. Consider, for instance, Elkana’s recent claim that “the law of conserva
-
tion could not be born either within the institutional framework of France or
that of England” ([1974], p. 155). What are the general rules or laws of
sociology which would warrant such a sweeping assertion? Where
are the
detailed case studies of the relation between institutional frameworks and
scientific discoveries which might make us teasonably confident that we
understood enough about the circumstances in which theories emerge to be
warranted in asserting claims as strong as Elkana's?
33. Cf. Hessen (1971).
34. Ben-David (1971), pp. 13-14.
35. Merton (1970), p. 75.
36. Mannheim (1952), p. 135.
37. Richter (1973), p. 6.
38. Mannheim (1936), p. 288.
39. Ibid.

SS
 cient Ramwenmtasturthants [Wedeal
246 NOTES TO PAGE 221

40. Very similar conclusions apply to the psycho-history of scientific

-
knowledge, which is probably even further from possessing a psycho-dynam
ical model which can correlate beliefs about the natural world with psycho-
logical (or psychiatric) dispositions. Questions about whether, say, manic-
i
depressives tend to favorfield theories are on about the same level as whether
:
gentlemen prefer blondes!
 Bibliography

Agassi, J. ‘“Towards an Historiography of Science." History and Theory

Beiheft 2 (1963).
. “Scientific Problems and their Roots in Metaphysics.’’ In The Critical
Approach to Science and Philosophy, edited by M. Bunge, pp. 189-211,
1964.
Aiton, E. The Vortex Theory of Planetary Motions. London, 1972.
Barber, B. “Resistance by Scientists to Scientific Discovery.’’ Science 134
(1961): 596ff. (My references are to this paper as reprinted in Barber, B.,
and Hirsch, W., eds. Sociology of Science. New York, pp. 539ff., 1962.)
Bartley, W. “Theories of Demarcation between Science and Metaphysics.” In
Problems in the Philosophy of Science, edited by Lakatos and Musgrave,
pp. 46-64. Amsterdam, 1968.
Beckman, T. “On the Use of Historical Examples in Agassi’s ‘Sensational-
ism’.”’ Stud. Hist. Phil. Sci. 1 (1971): 293ff.
Ben-David, J. The Scientist's Role in Society. Englewood Cliffs, New Jersey,
1971.
Berzelius, J. “Essay on the Cause of Chemical Proportions.” Ann. Phil. 2
(1813):443ff.
. “An Address to those Chemists Who Wish to Examine the Laws of
Chemical Proportions.”” Ann. Phil. 5 (1815): 122ff.
Boring, E. “The Dual Role of the Zeitgeist in Scientific Creativity.” In The
Validation of Scientific Theories, edited by P. Frank, pp. 187ff. New York,
1961.
Brooke, J. ‘Organic Synthesis and the Unification of Chemistry—a Reapprais-
al.” Brit. J. Hist. Sci. 3(1970-71): 363ff.
Brown, T. The Mechanical Philosophy and the Animal Oeconomy. Unpub-
lished dissertation, Princeton University, 1968.
. “The Electric Current in Early 19th-century French Physics.” Hist.
Stud. in the Phy. Sci. 1 (1969): 61ff.
. “The College of Physicians and the Acceptance of latro-Mechanism in
England, 1665-1695." Bull. of the History of Medicine 44 (1970): 12ff.

247

:
248 BIBLIOGRAPHY

Brush, S. ‘A History of Random Process. I. Brownian Movement from Brown

to Perrin.”’ Archive for History of Exact Sciences 5 (1968-69): 1-36.
Buchdahl, G. ‘Sources of Skepticism in Atomic Theory.’ Brit. J. Phil. Sci.
10 (1959): 120-34.
——. Metaphysics and Philosophy of Science. London, 1969.
. “History of Science and Criteria of Choice.” In Historical and Philo-
sophical Perspectives of Science, edited by R. Stuewer, pp. 204ff. Min-
neapolis, 1970.
. ‘Methodological Aspects of Kepler’s Theory of Refraction."” Stud.
Hist. Phil. Sci. 3 (1972): 265ff.
Bunge, M. Scientific Research. 2 v. Berlin, 1967.
Butts, R. “‘Consilience of Inductions and the Problem of Conceptual Change
in Science,”’ In Pittsburgh Series in Philosophy of Science, edited by R.
Colodny, forthcoming.
Cantor, G. “The Changing Role of Young’s Ether.” Brit. J. Hist. Sct. 5
(1970-71): 44ff.
. “Henry Brougham and the Scottish Methodological Tradition.” Stud.
Hist. Phil. Sci. 2 (1971): 68ff.
—. “The Edinburgh Phrenology Debate: 1803-1828.” Annals of Science 32
(1975a): 195ff.
——. “A Critique of Shapin’s Social Interpretation of the Edinburgh Phren-
ology Debate.” Annals of Science 32 (1975b):245ff.
Carnap, R. Logical Foundations of Probability. 2nd ed. Chicago, 1962.
Cohen, [. B. “History and the Philosopher of Science.” In The Structure of
Scientific Theories, edited by F. Suppe, pp. 308ff. Urbana, 1974.
Collingwood, R. G. Autobiography. Oxford, 1939.
. The Idea of History. New York, 1956.
Costabel, P. Leibniz and Dynamics; the Texts of 1692. Ithaca, New York,
1973.
Culotta, C. “German Biophysics, Objective Knowledge, and Romanticism.”
Historical Studies in the Physical Sciences 4 (1974): 3ff.
Duhem, P. The Aim and Structure of Physical Theory. Princeton, 1954.
Durkheim, E. Elementary Forms of the Religious Life. Glencoe, Illinois, 1947.
Elkana, Y. The Discovery of the Conservation of Energy. London, 1974.
Ellegard, A. “The Darwinian Theory and 19th-Century Philosophies of
Science,” J. Hist. Ideas 18 (1957): 360ff.
Eriksson, B. Problems of an Empirical Sociology of Knowledge. Uppsala, 1975.
Faraday, M. “‘An Answer to Dr. Hare’s Letter on Certain Theoretical
Opinions.” Phil. Mag. 17 (1840): 54-65.
Farley, J. ‘‘The Spontaneous Generation Controversy, | & H."’ J. Hist. Bio.
5 (1972): 95ff., 285ff.
Feyerabend, P. ‘‘Problems of Empiricism.”’ In Beyond the Edge of Certainty,
edited by R. Colodny, pp. 145-260. Englewood Cliffs, New Jersey, 1965.
. “Problems of Empiricism, II.” In The Nature and Function of
Scientific Theory, edited by R. Colodny. Pittsburgh, 1970a.
 BIBLIOGRAPHY
249
. “Against Method.” In Minnesota Studies in the Philosophy of Science,
vol. 4. Minneapolis, 1970b.
- “Consolations for the Specialist,” in Criticism and the Growth of
Knowledge, edited by Lakatos and Musgrave, pp. 197ff. Cambridge, 1970c.
. Against Method. London, 1975.
Fischer, D. Historians’ Fallacies: Toward a Logic of Historical
Thought.
New York, 1970.
Forman, P. “Weimar Culture, Causality, and Quantum Theory, 1918-
1927: Adaptation by German Physicists and Mathematicians to a Hostile
Intellectual Environment.” Historical Studies in the Physical Sciences
3
(1971): Uf.
Foucault, M. The Order of Things. New York, 1970.
Fox, R. “The Rise and Fall of Laplacian Physics.” Historical
Studies in the
Physical Sciences 4 (1974): 89ff.
Frank, P. ‘‘The Variety of Reasons for the Acceptance of Scientific
Theories.”’
in The Validation of Scientific Theories, edited by P. Frank,
pp. 13ff. New
York, 1961.
Ghiselin, M. The Triumph of the Darwinian Method. Berkeley,
1969.
Giere, R. “History and Philosophy of Science: Intimate Relationsh
ip or
Marriage of Convenience?” Brit. J. Phil. Sci. 24 (1973): 282-97.
Gillispie, C. The Edge of Objectivity. Princeton, 1960.
Gilson, E. Etudes sur le réle de la pensée médiévale. Paris, 1951.
Goldberg, S. ‘‘Poincaré’s Silence and Einstein's Relativity.
” Brit. J. Hist. Sci.
5 (1970-71): 73ff.
Griinbaum, A. “The Duhemian Argument,” Phil. of Sci.
11 (1960): 75-87.
- “The Special Theory of Relativity as a Case Study of the Importanc
e of
Philosophy of Science for the History of Science." In Philosoph
y of Science,
vol. 1, edited by B. Baumrin. New York, 1963.
- “Can We Ascertain the Falsity of a Scientific Hypothesi
s,” Studium
Generale 22 (1969): 1061-93,
. Philosophical Problems of Space and Time. 2nd ed.
Dordrecht, 1973.
- “Cana Theory Answer More Questions than One ofIts Rivals?”
Brit. J.
Phil. Sci. 27 (1976a): Lff.
. “Ad Hoc Auxiliary Hypotheses and Falsiticationism.”
Brit. J. Phil.
Set. 27 (1976b).
Griinwald, E. Das Problem einer Soziologie des Wissens. Wien, 1934.
Hare, R. “A Letter to Prof. Faraday on Certain Theoretical Opinions.
” Phil.
Mag. 17 (1840): 44-54,
Harris, E. Hypothesis and Perception. London, 1970.
Heimann, P. ‘Maxwell and the Modes of Consistent Representation.””
Archive
Sor History of Exact Sciences 6 (1969-70): 171ff.
Hessen, B. The Social and Economic Roots of Newton's
“Principia."’ New
York, 1971.
Hodge, M.J.S.P.H.D. “Lamarck’s Science of Living Bodies."” Brit. J. Hist.
Sci. 5 (1970-71): 323¢f.
250 BIBLIOGRAPHY

. “The Universal Gestation of Nature: Chambers’ Vestiges and Explana-

tions. J. Hist. Bio. 5 (1972): 127ff.
. “Methodological Issues in the Darwinian Controversy.’ Forthcoming.
Holton, G. Thematic Origins of Scientific Thought. Cambridge, Mass., 1973.
. “On the Role of Themata in Scientific Thought.” Science 188 (1975):
328ff.
Home, R. ‘‘Francis Hauksbee’s Theory of Electricity.”” Archive for History of
Exact Sciences 4 (1967-68): 203ff.
. “Franklin’s Electrical Atmospheres.” Brit. J. Hist. Sci. 6 (1972-73):
343ff.
Hooykaas, R. The Principle of Uniformity in Geology, Biology and Theology.
Leiden, 1963.
Hull, D. Darwin and his Critics. Cambridge, Mass., 1973.
. “Central Subjects and Historical Narratives.” History and Theory 14
(1975): 253ff.
lltis, C. “‘The Leibnizian-Newtonian Debates: Natural Philosophy and Social ‘
Psychology.” Brit. J. Hist. Sci. 6 (1972-73): 343ff. 3
Jaspers, K. The Great Philosophers. New York, 1962. |
King, M. ‘Reason, Tradition, and the Progressiveness of Science.’’ History i
and Theory 10 (1971): 3ff. :
Knight, D. Atoms and Elements. London, 1970.
Koertge, N. “Theory Change in Science.”’ In Conceptual Change, edited by
Pearce and Maynard, pp. 167ff. Dordrecht, 1973.
Kopnin, P., et. al, eds. Logik der wissenschaflichen Forschung. Berlin,
1969.
Korch, H. Die wissenschaftliche Hypothese. Berlin, 1972.
Kordig, C. The Justification of Scientific Change. Dordrecht, 1971.
Koyré, A. “Review of Crombie’s Robert Grosseteste."" Diogéne no. 16.
October 1956. ;
Kuhn, T. The Structure of Scientific Revolutions. Chicago, 1962. :
. “History of Science.” In International Encyclopedia of the Social
Sciences, pp. 74-83. New York, 1968.
. “Logic of Discovery or Psychology of Research?” In Criticism and i
the Growth of Knowledge, edited by Lakatos and Musgrave, pp. Iff. ‘
Cambridge, 1970.
Lakatos, I. “Proofs and Refutations.”” B.J.P.S. 14 (1963): 1-25, 120~39,
221-43, 296-342.
——.. “Criticism and the Methodology of Scientific Research Programmes.”
Proc. Aristotelian Soc. 69 (1968a): 149ff.
. “Changes in the Problem of Inductive Logic.” In The Problem of ;
Inductive Logic, edited by 1. Lakatos, pp. 315-417. New York, 1968b. {
“Falsification and the Methodology of Scientific Research Pro-
grammes.” In Criticism and the Growth of Knowledge, edited by Lakatos
and Musgrave, pp. 91ff. Cambridge, 1970.
”
. “History of Science and its Rational Reconstructions. In Boston

es
 BIBLIOGRAPHY
251
Studies in the Philosophy of Science, vol. 8, edited by R. Buck
and R.
Cohen, pp. 91ff, 1971.
Lakatos, I., and Zahar, E. “Why did Copernicus’ Research Program
Super-
cede Ptolemy's?” In The Copernican Achievement, edited by
R. Westman,
pp. 354ff. Berkeley, 1975.
Laudan, L. “Griinbaum on the ‘Duhemian Argument’.” Philosoph
y of Science
32 (1965): 295ff. (Reprinted in S. Harding, ed. Can Theories Be
Refuted?
Dordrecht, 1976.)
- “Thomas Reid and the Newtonian Turn of British Methodolo
gical
Thought.” In The Methodological Heritage of Newton, edited by
Butts and
Davis, pp. 103ff. Toronto, 1970.
. “C. S. Peirce and the Trivialization of the Self-Corrective Thesis.””
In
Foundations of Scientific Method in the 19th Century, edited
by R. Giere
and R. Westfall, pp. 275ff. Bloomington, 1973a.
- “G. L. Le Sage: a Case Study in the Interaction of Physics
and
Philosophy.” In Logic, Methodology and Philosophy of Science-IV
, edited
by P. Suppeset. al., pp. 429ff. Amsterdam, 1973b.
- “The Methodological Foundations of Mach’s Opposition to Atomism.”
In Space and Time, Matter and Motion, edited by P. Machamer
and R.
Turnbull, pp. 390ff. Columbus, 1976.
- “Two Dogmasof Methodology.” Philosophy of Science 43 (1976b).
- “The Sources of Modern Methodology.” In Logic, Methodology and
Philosophy of Science-V, edited by R. Butts and J. Hintikka, Dordrecht
,
1977.
Laudan, R. “Ideas and Institutions: the Case of the Geological
Society of
London.” Isis, forthcoming.
Leplin, J. “The Concept of an Ad Hoc Hypothesis.” Stud.
Hist. Phil. Sci. 5
(1975): 309-45.
Lukes, S. “Some Problems about Rationality.” Archives
Européenes de
Sociologie 8 (1967): 247ff.
McEvoy, J. “‘A ‘Revolutionary’ Philosophy of Science.” Philosoph
y of Science
42 (1975): 49ff.
McEvoy, J., and McGuire, J. “God and Nature: Priestley’s Way
of Rational
Dissent.” Hist. Stud. Phys. Sci. 5 (1975).
McGuire, J. “Atoms and the ‘Analogy of Nature’.”” Stud. Hist.
Phil. Sci. 1
(1970): 3¢f.
McGuire, J. E., and Heimann, P. “Newtonian Forces and Lockean
Powers.”
Hist. Stud. in Phys. Sci 3 (1971): 233¢€.
Machamer, P. “Feyerabend and Galileo.” Stud. Hist. Phil. Sci.
4 (1973): 1ff.
McKie, D., and Partington, J. ‘Historical Studies on the Phlogiston Theory,
LIV.” Annals of Science 2 (1937): 361ff; 3 (1938): 1ff and 337ff;
4 (1939):
113ff.
McMullin, E. “The History and Philosophy of Science: a Taxonomy
.” In
Historical and Philosophical Perspectives of Science, edited
by R. Stuewer,
p. 12ff. Minneapolis, 1970.
 ants Lele Sen St Ahantn catoloy A LattatieadNitin o£ta
252 BIBLIOGRAPHY

Mannheim, K. Ideology and Utopia. London, 1936.

. Essays on the Sociology of Knowledge. London, 1952.
Martin, E. Historie des monstres dupuis l'antiquité jusqu‘a nos jours. Paris,
1880.
Masterman, M. ‘“‘The Nature of a Paradigm.” In Criticism and the Growth of
Knowledge, edited by Lakatos and Musgrave, pp. 59ff. Cambridge, 1970.
Maxwell, A. “A Critique of Popper’s Views on Scientific Method.” Phil. Sci.
39 (1972): 31-52.
Merton, R. Social Theory and Social Structure. Chicago, 1949.
. Science, Technology and Society in 17th-century England. New York,
1970.
Mitteistrass, J. “‘Methodological Elements of Keplerian Astronomy.” Stud.
Hist. Phil. Sci. 3 (1972): 203ff.
. Die Méglichkeit von Wissenchaft. Frankfurt am Main, 1974.
Mitroff, I. The Subjective Side of Science. Amsterdam, 1974.
Mutschalow, I. ‘“‘Das Problem als Kategorie der Logik der wissenchaftlichen
Erkenntnis.”’ Voprosy Filosofii 11 (1964): 27-36.
Nelson, L. ““Whatis the History of Philosophy?” Ratio, 1962.
Neurath, O. ‘“‘Pseudorationalismus der Falsifikation.”” Erkenntnis 5S (1935):
353-65.
Nye, M. J. Molecular Reality. London, 1972.
. “Gustave LeBon’s Black Light: a Study in Physics and Philosophy in
France at the Turn of the Century.” Hist. Stud. in the Phys. Sci. 4 (1974):
163ff.
Olson, R. Scottish Philosophy and British Physics, 1750-1880. Princeton,
1975.
Oresme, N. A Treatise on the Uniformity and Difformity of Intensities. Edited
by M. Clagett. Madison, Wisconsin, 1968.
Pepper, S. “On the Cognitive Value of World Hypotheses.” Journal of
Philosophy 33 (1936): 575-77.
Popkin, R. The History of Scepticism from Erasmus to Descartes. Assen, 1960.
Popper, K. The Logic of Scientific Discovery. London, 1959.
. Conjectures and Refutations. London, 1963.
——.. Objective Knowledge. Oxford, 1972.
. “The Rationality of Scientific Revolutions.” In Problems of Scientific
Revolution, edited by R. Harré, pp. 72-101. Oxford, 1975.
Post, H. “Correspondence, Invariance and Heuristics.” Stud. Hist. Phil. Sci.
2 (1971): 213ff.
Quine, W. From a Logical Point of View. Cambridge, Mass., 1953.
Rescher, N. Methodological Pragmatism, forthcoming.
Richter, M. Science as a Cultural Process. New York, 1973.
Roger, J. Les sciences de la vie dans la pensée francaise du X VIIsiécle. Paris.
1963.
Rudwick, M. “Uniformity and Progression.” In Perspectives in the History of Sct-
ence and Technology, edited by D. Roller, pp. 209ff. Norman, Oklahoma, 1971.
Sabra, A. Theories of Light from Descartes to Newton. London, 1967.
 BIBLIOGRAPHY
253
Salmon, W. “‘Bayes's Theorem and the History of Science.
" In Historical and
Philosophical Perspectives of Science, edited by R.
Stuewer, pp. 68ff.
Minneapolis, 1970.
Schaffner, K. “Outlines of a Logic of Comparative Theory
Evaluation.” In
Historical and Philosophical Perspectives of Science, edited
by R. Stuewer,
pp. 311ff.Minneapolis, 1970.
——.. Nineteenth-century Aether Theories. Oxford, 1972.
. “Einstein vs. Lorentz.” Brit. J. Phil. Sci. 25 (1974): 45-78.
Schagrin, M. “Resistance to Ohm’s Law.” Amer. J. of Phys.
31 (1963): 536-47.
Scheffler, 1. Science and Subjectivity. Indianapolis, 1967.
Schofield, R. Mechanism and Materialism. Princeton, 1970,
Shapere, D. “The Structure of Scientific Revolutions.’”’ Phil.
Rev. 73 (1964):
383-94,
——. “Meaning and Scientific Change.” In Mind and
Cosmos, edited by
R. Colodny, pp. 41 ff. Pittsburgh, 1966.
Shapin, S. ‘‘Phrenological Knowledge and the Social Structure
of Early 19th-
century Edinburgh.” Annals of Science 32 (1975): 219ff.
Shapiro, A. “Kinematic Optics: A Study of the Wave Theory
of Light in the
17th-century.” Archive for History of Exact Sciences 11 (1973):
134ff.
Sharikow, W. “Das wissenschaftliche Problem.” In Logik
der wissenschaft-
lichen Forschung, edited by P. Koptin et. af. Berlin, 1972.
Simon, H. “Scientific Discovery and the Psychology of Problem
Solving.” In
Mind and Cosmos, edited by R. Colodny, pp. 22ff. Pittsburg
h, 1966.
Skinner, Q. ‘Meaning and Understanding in the History of
Ideas.” History
and Theory 8 (1969): 3ff.
Sloan, P. ‘John Locke, John Ray and the Probiem of the Natura!
System.”' J.
Hist. Biol. 5 (1972): Uff.
Stallo, J. Concepts and Theories of Modern Physics. Cambridg
e, Mass., 1960.
Stegmiiller, W. “Theoriendynamik . . . ,” Theorie der Wissenschaftgeschi
chte,
edited by W. Diederich, Frankfurt am Main, pp. 167ff.
Suppe, F., ed. The Structure of Scientific Theories. Urbana,
1974.
Thackray, A. “Has the Present Past a Future?” In Historica
l and Philosophical
Perspectives of Science, edited by R. Stuewer. Minneapolis, 1970.
Tornebohm, H. “The Growth of a Theoretical Model.” In Physics,
Logic and
History. London, 1970.
Toulmin, S. ‘Does the Distinction between Normal and Revoluti
onary Science
Hold Water?” In Criticism and the Growth of Knowledg
e, edited by I.
Lakatos and A. Musgrave, pp. 39ff. Cambridge, 1970.
Truesdell, C. Essays in the History of Mechanics. New York,
1968.
Vartanian, A. “Trembley's Polyp, La Mettrie, and
18th-Century French
Materialism.” In Roots of Scientific Thought, edited by P.
Wiener and A.
Noland, pp. 497ff. New York, 1957.
Viner, J. “Adam Smith and taissez faire.” In Adam
Smith, 1776-1926.
Chicago, 1928.
Watkins, J. ‘Influential and Confirmable Metaphysics."”
Mind, N.S. 67
(1958): 344-65.
254 BIBLIOGRAPHY
Watson, R. The Downfall of Cartesianism: 1673-1712. The Hague, 1966.
Whewell, W. The Philosophy of Inductive Sciences, Founded upon their
History. 2v., London, 1840.
. On the Philosophy of Discovery. London, 1860.
Winch, P. ‘‘Understanding a Primitive Society.” Amer. Phil. Quart. 1 (1964):
307ff.
Wittich, D., et al., eds. Problemstruktur und Problemverhalten in der
wissenschaftlichen Forschung. Rostock, 1966.
Zahar, E. “‘Why did Einstein’s Programme Supersede Lorentz’s? I, Il.” Brit.
J. Phil. Sci. 24 (1973): 9Sff., 223ff.
Index of Names

Agassi, 156, 164, 168, 238, 239 Cantor, 230, 231, 233, 244
Aiton, 236 Carnap, 4, 47, 227
Ampeére, 84, 105 Carnot, 23, 90, 91, 92, 94
Aquinas, 97, 131 Chambers, 240
Aristotle, 1, 2, 25, 36, 51, 55, 58, 97, Charleton, 217
112, 131, 134, 160, 214 Clapeyron, 94
Arschloch, 245 Clarke, 63
Clasius, 23, 95
Bacon, 26, 59, 180, 185, 207, 224 Cohen, 239
Barber, 207-208, 243 Collingwood, 47, 121, 147-150, 177, 178,
Barrow, 144 183, 186, 237, 241
Beckman, 239 Comte, 231
Ben-David, 219, 245 Condorcet, 147
Bergson, 241 Conybeare, 228
Berkeley, 46, 135 Copernicus, 21, 46, 47, 55, 110, 112, 141,
Bernard, 58 218, 229, 230, 232
Bernoulli, 25, 99, 104, 135 Costabel, 236
Berzelius, 31, 234, 237 Cotes, 62, 231
Biot, 105, 228 Cramer, 21
Black, 83 Culotta, 231
Boerhaave, 83 Cuvier, 148, 149
Bohr, 71, 169, 182, 232
Boltzmann, 84 Dalton, 113, 117, 149, 234
Borelli, 83 Darwin, 2, 34, 46, 63, 75, 78, 97, 101,
Boring, 102, 103 117, 136, 137, 138, 141, 207, 218, 229,
Boscovich, 104, 135 231
Boyle, 41, 85, 180, 218, 219 Democritus, 182
Brewster, 19 Descartes, 25, 29, 34, 36, 47, 52, 58, 75,
Brongniart, 19 79, 80, 81, 85, 87, 88, 89, 91, 95, 97, 99,
Brooke, 230, 231 101, 103, 104, 105, 117, 127, 139, 144,
Brown, R., 19, 228 166, 176, 180, 218, 233, 240
Brown, T., 214-217, 233, 234, 244, 245 Dilthey, 183
Brush, 228 Duhem, 27, 40-44, 119, 147, 182, 229,
Buchdahl, 62, 230, 231 230, 231
Buffon, 231 Dujardin, 19
Butts, 230 Durkheim, 212, 213, 214, 244

255
256 INDEX

Eddington, 23 Hutchinson, 135

Einstein, 20, 23, 34, 48, 49, 71, 89, 127, Hume, 177
139, 141, 227 Hutton, 62, 105, 145, 148, 233
Elkana, 245 Huygens, 25, 46, 61, 64, 81, 85, 88, 89,
Ellegard, 231 95, 97, 135, 144, 149, 232, 233
Engles, 175
Euclid, 201 litis, 232
Eudoxus, 51
Euler, 135 Jaspers, 241

Faraday, 49, 50, 87, 99, 181, 230 Kant, 62, 135, 179
Feuerbach, 105 Kelvin, see Thomson
Feyerabend, 3, 4, 47, 66, 74, 110, 113, Kepler, 29, 35
141, 143, 148, 156, 168, 182, 227, 231, Knight, 231
234 Koertge, 232, 237
Fischer, 171 Kordig, 237
FitzGerald, 117 Koyré, 58, 59, 230
Forman, 214-217, 234, 244, 245 Kramers, 71
Foucalt, 241 Kuhn, 1, 3, 4, 37, 47, 66, 72-78, 96, 99,
Fourier, 105 100, 109, 110, 113, 133-136, 138, L4t~
Frank, 230 143, 145, 147, 148, 150, #51, 156, 171,
Franklin, 34, 90, 149, 229, 237 174, 175, 206, 207, 215-216, 227, 228-
Frege, 100 234, 236, 237, 238, 240, 243, 244, 245
Freud, 2, 46, 71, 78, 97
Lakatos, 4, 26, 47, 66, 72, 76-77, 78, 91,
Galen, 214, 218 95, 97, 99, 100, 113-115, 118, 129,
Galileo, 24, 25, 29, 33, 55, 58, 94, 112, 147-151, 155, 156, 163, 168-170, 191,
164, 218, 238 227, 228, 229, 232-240, 242, 243
Ghiselin, 231 Lamarck, 141, 218, 240
Giere, 158, 159, 238 Lambert, 60, 135
Gillispie, 238 La Mettrie, 21
Giison, 176, 240 Laudan, L. 229, 230, 231, 233, 235, 236,
Gorsseteste, 131 238, 242
Griinbaum, 26, 77, 114-115, 156, 229, Laudan, R., 231
232, 233, 234, 235, 238 Lavoisier, 24, 94, 234
Griinwald, 201, 243 Leibniz, 46, 61, 63, 64, 88, 89, 104, 27,
_

135, 180, 182, 232

Hales, 92, 93 Lenin, 218
Hanson, 66, 141, 143, 156, 182 Leplin, 235
Hare, 50, 230 Le Sage, 60, 135
Harris, 237 Locke, 46, 59, 175, 183
Hartley, 60, 85, 135 Lorentz, 116, 117, 232
Hegel, 105, 199, 240 Lovejoy, 181, 182
Heimann, 62, 230, 231, 236 Lukes, 236
Heisenberg, 216 Lyell, 80, 81, 135-136, 148, 149, 218,
Hertz, 85, 87 231, 233
Hessen, 219, 245 Lyonnet, 21
Hintikka, 4, 227 Lysenko, 63
Hippocrates, 58
Hobbes, 81, 144, 180, 181 McEvoy, 230, 239
Hodge, 231, 240 McGuire, 62, 230, 231, 236
Holton, 182, 242 Mach, 50, 58, 100
Home, 229, 237 Machamer, 239
Hooke, 81, 144 McKie, 232
Hooykaas, 231 McMullin, 156, 239
Hull, 231, 234 Malebranche, 130
 257
INDEX

Mannheim, 196, 201, 202, 209-213, 219- Rayleigh, see Strutt

221, 242, 243, 244, 245 Régis, 81
Margenau, 237 Reichenbach, 4, 47, 125, 147
Martin, 228
Richter, 207, 214, 220, 243, 245
Marx, 2, 63, 71, 78, 79, 80, 99, 101, 105, Roger, 231
132, 165, 178, 199, 218, 219 Rohauitt, 81
Masterman, 231 Rumford, see Thompson
Maupertuis, 104, 135
Sabra, 230, 232
Maxwell, A., 235
Salmon, 4, 45
Maxwell, J., 71, 85, 87, 94, 117, 230, 243
Schaffner, 118, 233, 235
Mendel, 35, 63
Scheffier, 129, 236
Mersenne, 218
Scheler, 214, 242
Merton, 202, 219, 220, 242, 243, 244, 245
Schofield, 104, 234
Michelson, 34, 87
Shapere, 74, 231, 236
Mill, 26, 177, 179, 209, 210
Shapin, 244, 245
Moliand, 228
Shapiro, 232, 233
Morgan, 196
Simon, 11
Morley, 34, 87
Skinner, B., 46, 183
Musgrave, 232
Skinner, Q., 241, 242
Slater, 71
Nelson, 177, 241 Smith, 105, 230
Neurath, 27, 229 Sorokin, 214
Newton, 2, 23, 24, 25, 29, 34, 46, 47, Spengler, 215
49, 52, 58, 59, 60, 61, 62, 64, 75, 79, Stahl, 97
81, 84, 85, 88, 89, 93, 95, 97, 99, 100, Statlo, 50, 230
101, 103, 104, 105, 117, 127, 135, 137, Stegmiiller, 147
139, 141, 144, 146, 149, 160, 166, 178, Strutt, 38
180, 181, 183, 188, 218, 219, 230, 231, Suppe, 236, 238
232, 233, 238
Nye, 228 Tarski, 77
Teithard, 241
Olson, 230 Thackray, 245
Oresme, 228 Thomas Aquinas, 97, 131
Thompson, 83, 149
Parmides, 125 Thomson, 57, 243
Partington, 232 Tornebohm, 239
Pasteur, 94 Toulmin, 70, 156
Peirce, 125, 147, 179, 223 Trembley, 20
Perrin, 20 Truesdell, 236
Pitcairn, 83
van der Waals, 26
Plank, 94, 182
Vartanian, 20, 228
Plato, 1, 51, 125, 182
Viner, 230
Playfair, 233
Poisson, 10S Watkins, 232
Popkin, 176, 240 Watson, 240
Popper, 4, 9, 26, 36, 47, 77, 114, (45, Weber, 218, 243
124, 129, 136, 147, 148, 149, 227, 228, Wegener, 71
229, 232, 235, 236, 239, 242 Werner, 145
Post, 148, 237 Whewell, 50, 147, 156, 179, 230
Priestley, 62, 231, 234 Winch, 236
Prout, 31, 169 Wolff, 64
Ptolemy, 24, 46, 47, 51, 52, 112, 117,
123, 141 Young, 24, 88, 230

Quine, 27, 141, 182, 229 Zahar, 115, 118, 232, 235, 236

hoy 47987