Philosophy of Social Science
 
Philosophy of
Social Science
Alexander Rosenberg
Duke University
      
Westview Press was founded in 1975 in Boulder, Colorado, by notable publisher and
intellectual Fred Praeger. Westview Press continues to publish scholarly titles and
high-quality undergraduate- and graduate-level textbooks in core social science disciplines. With books developed, written, and edited with the needs of serious nonfiction
readers, professors, and students in mind, Westview Press honors its long history of
publishing books that matter.
Copyright © 2016 by Westview Press
Published by Westview Press,
A Member of the Perseus Books Group
2465 Central Avenue
Boulder, CO 80301
www.westviewpress.com
All rights reserved. Printed in the United States of America. No part of this book may
be reproduced in any manner whatsoever without written permission except in the
case of brief quotations embodied in critical articles and reviews.
Every effort has been made to secure required permissions for all text, images, maps,
and other art reprinted in this volume.
Westview Press books are available at special discounts for bulk purchases in the
United States by corporations, institutions, and other organizations. For more
information, please contact the Special Markets Department at the Perseus Books
Group, 2300 Chestnut Street, Suite 200, Philadelphia, PA 19103, or call (800) 810-4145,
ext. 5000, or e-mail [email protected].
Library of Congress Cataloging-in-Publication Data
Rosenberg, Alexander, 1946Philosophy of social science / Alexander Rosenberg, Duke University. — Fifth
edition.
pages cm
Includes bibliographical references and index.
ISBN 978-0-8133-4973-2 (paperback) — ISBN 978-0-8133-4990-9 (e-book) 1.
Social sciences–Philosophy. I. Title.
H61.R668 2015
300.1–dc23
2015007880
10 9 8 7 6 5 4 3 2 1
To the memory of Blanca N. Rosenberg,
mentor and teacher to two generations of students
in the School of Social Work, Columbia University, 1958–1979
Contents
Preface to the Fifth Edition
ix
1 WHAT IS THE PHILOSOPHY OF SOCIAL SCIENCE?
1
2 THE METHODOLOGICAL DIVIDE:
NATURALISM VERSUS INTERPRETATION
11
3 THE EXPLANATION OF HUMAN ACTION
35
4 ACTIONS, INTENTIONALITY, AND THE
MIND BODY PROBLEM
55
5 BEHAVIORISM IN THE BEHAVIORAL SCIENCES
73
6 PROBLEMS OF RATIONAL CHOICE THEORY
93
7 SOCIAL PSYCHOLOGY AND THE
CONSTRUCTION OF SOCIETY
121
8 EUROPEAN PHILOSOPHY OF SOCIAL SCIENCE
143
9 HOLISM AND ANTIREDUCTIONISM IN
SOCIOLOGY AND PSYCHOLOGY
173
10 FUNCTIONALISM AS A RESEARCH PROGRAM
195
11 SOCIOBIOLOGY OR THE STANDARD
SOCIAL SCIENCE MODEL?
211
vii
viii
Contents
12 THEORIES OF CULTURAL EVOLUTION
237
13 RESEARCH ETHICS IN SOCIAL INQUIRY
257
14 FACTS AND VALUES IN THE HUMAN SCIENCES
283
15 SOCIAL SCIENCE AND THE ENDURING
QUESTIONS OF PHILOSOPHY
307
Bibliography
Index
323
329
Preface to the Fifth Edition
A generation is a lifetime in the history of a textbook. I wrote the first edition
of this work in the late 1980s, long before the advent of online resources and
well before the current proliferation of handbooks, encyclopedias, companions, and guides to every discipline and subdiscipline. Four revisions later,
however, this old-fashioned introduction to the philosophy of social science
is holding its own in the online informational environment of the twentyfirst century. What is the reason for this? Some hypotheses are disobliging
and disconfirmed: it can’t be that no one cares enough about the subject;
there are, after all, by Google Scholar’s account, a dozen books with titles
that are variations on The Philosophy of Social Science. Nor can it be that no
one plugged into the web is interested in this subject or my take on it: I once
had to threaten legal action to get a plagiarized version of the second edition
taken down from a website.
To explain the persistence of this book across twenty-four years and four
editions, I propose this hypothesis: its particular approach to the philosophy of social science has persisted because of the merit of the book’s central
premises—that social scientists must take sides on philosophical problems,
whether they like it or not, even whether they know it or not, and that the
problems of the philosophy of social science are all versions of one or another fundamental problem of philosophy: problems of metaphysics, of epistemology, of ethics.
Its central premise—that the philosophy of social science is a way of coming to grips with the perennial problems of philosophy—makes this volume
different from the other works that share its title. It does not pretend to be
a tour guide through isms and fashions, nor a précis of major theories and
findings in social science. This is a work of philosophy as well as a textbook.
A great deal has changed in the social sciences since the 1980s, when the
first edition was gestating. Economics, for example, has greatly changed,
owing to events in the economy and changes in the social sciences that it
ix
x
Preface to the Fifth Edition
has refused to take seriously, especially in cognitive social psychology, experimental economics, and evolutionary game theory. Even since the fourth
edition, a new subdiscipline has gone mainstream: neuroeconomics. Economics has also succumbed to attacks on its moral neutrality and indifference to application in human development. These too have made the field
recognizably more like the other social sciences it once feigned to distain.
Meanwhile, anthropology, sociology, and social psychology, along with
politics, have been swept up in a tsunami of Darwinian analysis originating
from tectonic changes in evolutionary biology and biological anthropology.
Owing to the zeal—perhaps trop de zèle—of the second and third generation
of sociologists, evolutionary psychologists, and gene-culture coevolutionary
theorists, there is still, after thirty or forty years, no end in sight to the Darwinization of the social sciences.
Another great change in social science since the first edition of this book
is the increasing willingness of European students of human affairs to be
influenced by naturalistic, empirical, and data-based approaches to social science. The empiricist and quantitative approaches to the sciences of
human affairs had their origins in European thinkers—Durkheim, Weber,
Walras. But that approach was eclipsed and effaced in the middle years of
the twentieth century by Marxism, the Frankfurt School, phenomenology,
structuralism, postmodernism, deconstruction, and critical theory. Now the
European philosophy of social science is moving back toward an appreciation of the older European tradition of social research that English-language
social science maintained through the twentieth century. But the Europeans
bring their intellectual tradition along to this new exchange. This raises
questions of compatibility that few address.
All the social sciences have become much more sensitive to, and much
more influenced by, theories and findings that reflect the experience and perhaps also the special information—if not exclusive knowledge—of women,
ethnic, racial, and other largely marginalized groups. How to incorporate
these new voices and thoughts remains a vexed matter. In this fifth edition,
I devote more discussion to this question.
However, these changes in the social sciences have brought along with
them not so much changes in the philosophical questions they raise, but a
new vocabulary for addressing the persistent philosophical questions that
face the social scientist. This edition reflects the new vocabulary of the
human sciences, while continuing the previous editions’ insistence that the
problems social science faces are old wine in new bottles, but just as intellectually intoxicating as ever.
The fourth edition added much new material on the role of models and
equilibrium explanations in economics; new discussions of how speech acts
Preface to the Fifth Edition
xi
create norms and thus construct social practices and institutions; the problem of spontaneous order in the creation of institutions; and the relationships of Rawls’s moral theory to social research and Sen’s capacity theory to
the broad problem of how facts and values intersect. Besides a fuller treatment of feminist philosophy of science, this fifth edition adds a discussion of
the ethics of care that sometimes accompanies feminist views, a discussion
of how advances in neuroscience bear on the social sciences, not just in
neuroeconomics, along with an exposition of postmodernist approaches to
knowledge.
Feedback on the usefulness of previous editions has suggested that instructors often skip over Chapter 4, on the nature of intentionality, because
the chapter proceeds at a level of abstraction from the immediate methodological concerns of social scientists. But it is intentionality that makes the
difference between the methodologies of the social and the natural sciences.
To encourage faculty and students to plow through this philosophically sophisticated but difficult material, I have rewritten parts of Chapter 4 and
signposted its importance elsewhere in the book. You can skip this chapter
without losing the thread of the argument. But readers need to return to it
to see why the problems of the philosophy of social science are instances of
the perennial problems of philosophy going all the way back to Plato.
As in previous editions, I begin with an explanation of why philosophy
is relevant to the human sciences, and then I explore the problems raised by
alternative explanatory strategies of the human sciences. In the twentieth
century these problems spawned familiar theoretical and methodological
movements: behaviorism, structuralism, and a variety of interpretational
theories including critical theory, to name only a few. Even among social
scientists who accepted no labels for their views, these problems facing
their explanations led to significant shifts in the aims and methods advocated for the social sciences. Despite the changes described briefly above,
the challenges facing social science in the twenty-first century remain the
same as those that confronted sociologists such as Emile Durkheim and
Claude Lévi-Strauss, psychologists such as B. F. Skinner, economists such as
Milton Friedman, and cultural theorists such as Michel Foucault or Pierre
Bourdieu. The thinking of these figures and others is sketched in this book,
where it confronts fundamental problems of method and theory raised by
the philosophy of science.
Previous editions mention my debts to many scholars—social scientists
and philosophers. The ones whose lessons have stuck with me the longest
include David Braybrooke, Donald T. Campbell, Martin Hollis, Jonathan
Bennett, Dan Hausman, Harold Kincaid, Martin Trow, Alasdair McIntyre,
and Amartya Sen.
CHAPTER ONE
What Is the Philosophy
of Social Science?
Most sociologists and anthropologists agree on the definition and the domain of their disciplines; the same holds true for many psychologists, political scientists, and almost all economists. The same cannot be said for
philosophers and philosophy. Philosophy is a difficult subject to define,
which makes it difficult to show social scientists why they should care about
it—the philosophy of social science in particular. This chapter provides a
definition of philosophy that makes the subject inescapable for the social scientist. It shows that, whether as an economist or an anthropologist, one has
to take sides on philosophical questions. One cannot pursue the agenda of
research in any of the social sciences without taking sides on philosophical
issues, without committing oneself to answers to philosophical questions. At
a minimum, social scientists need to recognize this fact about themselves.
It is even better if the choices made are based on evidence and argument.
WHAT IS PHILOSOPHY?
Philosophers do not agree among themselves on the definition of their subject. Its major components are easy to list, and the subjects of some of them
are relatively easy to understand. The trouble is trying to figure out what
they have to do with one another, why combined they constitute one discipline—philosophy—instead of being parts of other subjects, or their own
independent areas of inquiry.
The major subdisciplines of philosophy include logic, the search for welljustified rules of reasoning; ethics (and political philosophy), which concerns


      
itself with right and wrong, good and bad, justice and injustice, in the conduct of individuals and states; epistemology or the theory of knowledge,
the inquiry into the nature, extent, and justification of human knowledge;
and metaphysics, which seeks to determine the most fundamental kinds
of things that exist in reality and what the relations between them are. Despite its abstract definition, many of the questions of metaphysics are wellknown to almost all people. For example, Is there a God? or, Is the mind just
the brain, or something altogether nonphysical? or, Do I have free will? are
metaphysical questions most people have asked themselves.
But these four domains of inquiry don’t seem to have much to do with
one another. Each seems to have at least as much to do with another subject
altogether. Why isn’t logic part of mathematics, or epistemology a compartment of psychology? Shouldn’t political philosophy go along with political
science, and isn’t ethics a matter ultimately for people who deliver sermons?
Whether we have free will or the mind is the brain is surely a matter for
neuroscience. Perhaps God’s existence is something to be decided upon not
by an academic inquiry but by personal faith. The question thus remains:
What makes them all parts of a single discipline, philosophy?
The answer to this question organizes this book, and it is pretty clear.
Philosophy deals with two sets of questions: first, questions that the sciences—physical, biological, social, behavioral—cannot answer now and perhaps may never be able to answer; second, questions about why the natural
and social sciences cannot answer the first set of questions.
There is a powerful argument for this definition of philosophy in terms
of its historical relationship with science. The history of science from the
ancient Greeks to the present is that of one compartment of philosophy
after another breaking away from philosophy and emerging as a separate
discipline. But each of these separated disciplines has left philosophy with
a set of distinctive problems, issues the discipline cannot resolve, but must
leave either permanently or at least temporarily for philosophy to deal with.
Mathematics leaves to philosophy questions like, What is a number? Physics
leaves to philosophy questions like, What is time? There are other questions science appears to be unable to address—the fundamental questions
of value, good and bad, rights and duties, justice and injustice—that ethics
and political philosophy address. Questions about what ought to be the case,
what we should do, and what is right and wrong, just and unjust, are called
normative. By contrast, questions in science are presumably descriptive or,
as is sometimes said, positive, not normative. Many of the normative questions have close cousins in the social and behavioral sciences. Thus, psychology will interest itself in why individuals hold some actions to be right and
others wrong; anthropology will consider the sources of differences among
Philosophical Problems of Social Science

cultures about what is good and bad; political science may study the consequences of various policies established in the name of justice; economics will
consider how to maximize welfare, subject to the normative assumption that
welfare is what we ought to maximize. But the sciences—social or natural—
do not challenge or defend the normative views we may hold.
In addition to normative questions that the sciences cannot answer, there
are questions about the claims of each of the sciences to provide knowledge,
or about the limits of scientific knowledge, that the sciences themselves cannot address. These are among the distinctive questions of the philosophy of
science, including questions about what counts as knowledge, explanation,
evidence, or understanding. The philosophy of science is that subdiscipline
of philosophy devoted to addressing these questions.
PHILOSOPHICAL PROBLEMS OF SOCIAL SCIENCE
If there are questions the sciences cannot answer and questions about why
the sciences cannot answer them, why should a scientist, in particular a
behavioral or social scientist, take any interest in them? The reason is simple. Though the sciences cannot answer philosophical questions, individual
scientists have to take sides on the right answers to them. The positions scientists take on answers to philosophical questions determine the questions
they consider answerable by science and choose to address, as well as the
methods they employ to answer them. Sometimes scientists take sides consciously. More often they take sides on philosophical questions by their very
choice of question, and without realizing it. The philosophy of science may
be able to vindicate those choices. At the least, it can reveal to scientists that
they have made choices, that they have taken sides on philosophical issues. It
is crucial for scientists to recognize this, not just because their philosophical
positions must be consistent with the theoretical and observational findings of their sciences. Being clear about a discipline’s philosophy is essential
because at the research frontiers of the disciplines, it is the philosophy of
science that guides inquiry.
As Chapter 2 argues, the unavoidability and importance of philosophical questions are even more significant for the social scientist than for the
natural scientist. The natural sciences have a much larger body of wellestablished, successful answers to questions and well-established methods
for answering them. As a result, many of the basic philosophical questions
about the limits and the methods of the natural sciences have been set aside
in favor of more immediate questions clearly within the limits of each of the
natural sciences.

      
The social and behavioral sciences have not been so fortunate. Within
these disciplines, there is no consensus on the questions that each of them
is to address, or on the methods to be employed. This is true between disciplines and even within some of them. Varying schools and groups, movements and camps claim to have developed appropriate methods, identified
significant questions, and provided convincing answers to them. But among
social scientists, there is certainly nothing like the agreement on such claims
that we find in any of the natural sciences. In the absence of agreement about
theories and benchmark methods of inquiry among the social sciences, the
only source of guidance for research must come from philosophical theories. Without a well-established theory to guide inquiry, every choice of
research question and of method to tackle it is implicitly or explicitly a gamble with unknown odds. The choice the social scientist makes is a bet that
the question chosen is answerable, that questions not chosen are either less
important or unanswerable, that the means used to attack the question are
appropriate, and that other methods are not.
Chapter 2 outlines the alternative choices, bets, and wagers about the best
way to proceed that face the social scientist. When social scientists choose to
employ methods as close as possible to those of natural science, they commit
themselves to the position that the question before them is one that empirical science can answer. When they spurn such methods, they adopt the
contrary view, that the question is different in some crucial way from those
addressed in the physical or biological sciences. Neither of these choices
has yet been vindicated by success that is conspicuous enough to make the
choice anything less than a gamble.
Whether these gambles really pay off will not be known during the lifetimes of the social scientists who bet their careers on them. Yet the choices
must be justified by a theory, either one that argues for the appropriateness of
the methods of natural science to the question the social scientist addresses,
or one that explains why these methods are not appropriate and supplies an
alternative. Such theories are our only reasonable basis for choosing methods of inquiry in the social sciences.
But these theories are philosophical, regardless of whether the person
who offers them is a philosopher or a social scientist. Indeed, social scientists
are in at least as good a position as philosophers to provide theories that
justify methods and delimit research. In the end, the philosophy of social
science not only is inevitable and unavoidable for social scientists, but it
must also be shaped by them as much as by philosophers.
The traditional questions of the philosophy of social science reflect the
importance of the choice among these philosophical theories. And in this
Philosophical Problems of Social Science

book we will examine almost all of those questions at length. By contrast
with this approach to social science, which very self-consciously takes its
inspiration from the natural sciences, there are disciplines that make the
meaning and intelligibility of human affairs central to their explanations.
These social scientists (and the philosophers who embrace their aims and
methods as the right way to proceed) contrast their commitment to understanding with demands for prediction. They are indifferent or hostile to
the notion that their disciplines should provide predictive knowledge about
individuals or groups. In Chapters 7 and 8 we look at this approach.
In Chapters 8 through 10 we also turn to questions about whether the
primary explanatory factors in social science should be large groups of people such as social classes or communities and their properties—so-called
structural properties, as Marx, Durkheim, and other social scientists have
argued—or whether explanations must begin with the choices of individual,
often “rational” human agents, as contemporary economists and some political scientists argue. The differences between the various social sciences,
especially economics and sociology, on this point are so abstract and general that they have long concerned philosophers. The social scientist who
holds that large-scale social facts explain individual conduct, instead of the
reverse, makes strong metaphysical assumptions about the reality of groups
independent of the individuals who compose them. Such a theory—called
holism—also requires a form of explanation called functionalism, which
raises other profound questions about differences between the explanatory
strategies of social and natural science. As a theory that gives pride of explanatory place to social wholes, holism might seem quite unappealing. But
the alternative to it, individualism, as advanced by economists, political scientists, and biologically inspired social scientists, for instance, also faces
equally profound philosophical questions.
Problems of functionalism, holism, and individualism are exacerbated by
the ever-increasing influence of biological science, and especially Darwin’s
theory of natural selection, on all the social and behavioral sciences. This
is the subject of Chapters 11 and 12, which report on several lively debates
at the intersection of biology and the social sciences and their philosophies.
In Chapters 13 and 14 we turn to the relation between the social sciences
and moral philosophy. We examine whether we can expect the social sciences themselves to answer questions about what is right or fair or just or
good. Many philosophers and social scientists have held that no conclusions
about what ought to be the case can be inferred even from true theories about
what is the case. Others have asserted the opposite. No matter who is right, it
will still turn out that alternative approaches to social science and competing

      
moral theories have natural affinities to, and make strong demands on, one
another as well. We must also examine the question of whether there are
morally imposed limits to legitimate inquiry in the social sciences.
Finally, in Chapter 15 I try to show why the immediate choices that social
scientists make in conducting their inquiry commit them to taking sides on
the most profound and perennial questions of philosophy. If this is right,
then no social scientist can afford to ignore the philosophy of social science
or any other compartment of philosophy.
ONE CENTRAL PROBLEM OF THE
PHILOSOPHY OF SOCIAL SCIENCE
The central philosophical dispute about the scope, aim, and prospects for
each of the social sciences taken separately, and all of them together, is what
sort of knowledge they should or can seek. The debate takes place against
a background argument about the nature of understanding in the natural
sciences. There it is widely held that increases in understanding are certified
by improvements in prediction.
Among social scientists who accept the requirement that their discipline
provide the kind of knowledge natural science provides—demographic sociologists, econometricians, experimental social psychologists, or political
scientists interested in voting behavior, for instance—there is a strong commitment to improving prediction as the mark of increasing understanding.
Among social scientists there are debates about how reliable and precise
their respective disciplines’ predictions can be and whether they can get better. But other social scientists reject the demand that their discipline provide
the same kind of understanding natural science offers. These social scientists offer alternative explanations of why their subjects cannot, and should
not, seek predictive knowledge and improvements in it. They provide quite
different accounts of what the aims and objectives of their disciplines can be.
The question centers on the fact that it is human beings, in groups and
individually, whose behaviors, actions, and their consequences we are trying
to understand that make the difference between natural and social science.
It is what shapes the nature and scope of the knowledge social science can
provide. Should the subject matter of these disciplines make the aims and
methods of the social sciences as a whole radically different from those of
the natural sciences?
The natural sciences are often alleged, especially by natural scientists and
others impatient with social science, to have made far greater progress than
the social sciences. Questions naturally arise as to why that is so and what
One Central Problem of the Philosophy of Social Science

can be done to accelerate the progress of social science toward achievements
comparable to those of natural science. But one should notice that these two
questions have controversial philosophical presuppositions: they presuppose
(1) that we know what progress in natural science is and how to measure it;
(2) that, based on our measurements, the natural sciences have made more
progress; and (3) that the social sciences aim for the same kind of progress
as the natural sciences.
If you agree that progress in the social sciences leaves much to be desired
compared with the natural sciences, then you must be able to substantiate
those three presuppositions. However, if you consider that the social sciences
cannot or should not implement the methods of natural science in the study
of human behavior, you will reject as misconceived the invidious comparison between the natural and the social sciences, along with the presuppositions on which it is based. But if you conclude that the study of human action
proceeds in a different way and is appraised with different standards than
the natural sciences, then you will have equally strong presuppositions about
the aims and achievements of social science to substantiate.
Chapters 2 through 4 of this volume outline the arguments both for and
against the claims that the social sciences have failed to progress and that
this failure needs explanation. Both arguments have one view in common:
a neat compromise is impossible. Such a compromise would suggest not
that social science has made as much progress as have the natural sciences,
but that it has made some. It would suggest that very broadly the methods
of the social sciences are the same as those of natural science, though their
specific concepts are distinctive and the interests the social sciences serve
are sometimes different. The compromise view holds that the lack of progress in social science is a consequence of the complexity of human processes,
which is much greater than that of natural processes. It also identifies limits
on our understanding that stem from the regulations, mores, and inhibitions barring controlled experiments on human beings. If this view is right,
the problems of social science are mainly practical instead of philosophical.
Though this is a possible view, much of the work of philosophers and social
scientists who have dealt with the philosophy of social science suggests that
this nice compromise is a difficult one to maintain. Most of the rest of this
book expounds arguments that one way or another attempt to undercut this
philosophically deflationary compromise. We will reconsider it often.
Some philosophers and social scientists reject as uninteresting or unimportant the question of whether the social sciences have progressed as fast as
natural science. They hold that the question is peripheral to the philosophy
of social science. On their view, the social sciences raise distinctive philosophical problems that have nothing to do with any comparison to other

      
disciplines. On this view, the chief goal of the philosophy of social science is
to understand the disciplines involved, without casting an eye to comparative questions that are at best premature and at worst a distraction.
Those social scientists and others who demand predictive improvement
as the litmus test of advancement in the social and behavioral sciences
condemn this attitude as complacent: it is indifferent to human needs and
aspirations, which social science is called upon to serve, for the extent to
which social science can ameliorate and improve the human condition is a
function of its similarity to natural science as a source of predictively useful
knowledge that can be applied in the way physics is applied to engineering.
There are several controversial counterarguments to this demand that
social science show the sort of predictive improvement that natural science
manifests and provide us with the sort of technological mastery that natural
science confers.
First, this demand seems to assume that the social sciences are all of one
piece, and most stand or fall together in regard to their predictive powers.
It may be that some social sciences are rightly viewed as potential sources
of predictive knowledge if conducted according to the “right methods.” But
in others, the appropriate methodology may not by any means aim at or
produce this sort of technologically applicable information about human
affairs. Not all the social sciences should be assessed along the same limited
set of dimensions.
Second, demanding that the social sciences show persistent increases in
predictive power can’t make them do it. If there are any impediments to
predictive success and technological application in the nature of human affairs, then no matter how hard anyone tries, predictive improvements can’t
happen. Third, it is often argued that the misguided belief that we already
have such knowledge has been used in the past not to ameliorate the human
condition but to worsen it. Even if we ever acquire such knowledge, the prospects of its beneficent use are dim. Finally, it is argued that the understanding of human affairs may ameliorate the human condition even if it does
not confer on us useful tools for manipulating the social environment, however well intentioned. The sort of understanding some of the social sciences
provide is precisely of this type—it enhances our lives without necessarily
enabling us to control our own, or for that matter others’ lives, any better.
SUMMARY
Philosophy deals with two sorts of questions, ones the sciences cannot answer and ones about why the sciences cannot answer all questions. Since the
Introduction to the Literature

social sciences are domains of greater methodological debate, and of greater
immediate relevance to daily life and to matters about which people really
care a great deal, the philosophical questions about what the social sciences
can and cannot tell us about human life are even more pressing than questions raised by the limits of the natural sciences.
Perhaps the leading question the philosophy of social science faces is
whether we should seek the kind of understanding of human affairs that
natural science gives us about nonhuman processes in nature—knowledge
that enables us increasingly to predict and control phenomena.
Introduction to the Literature
The account of the nature of philosophy and its relationship to science defended here is elaborated in Alex Rosenberg, The Philosophy of Science: A
Contemporary Introduction, in any of its three editions. This text and Peter
Godfrey-Smith’s Theory and Reality introduce the philosophy of science in
two ways: Rosenberg’s book is thematic, identifying the problems and alternative answers to them that philosophers of science have provided. GodfreySmith’s book is organized historically.
The two classical introductions to the philosophy of science are Carl
Hempel, Philosophy of Natural Science, and Ernest Nagel, The Structure of
Science. The latter work is somewhat more difficult and far more comprehensive than Hempel’s. It includes an extended defense of naturalistic philosophy of social science. The appendix to Daniel Hausman’s The Inexact
and Separate Science of Economics is a particularly clear introduction to the
philosophy of science, with special relevance for the social sciences, within
the compass of fifty pages.
CHAPTER TWO
The Methodological Divide:
Naturalism Versus Interpretation
The main lines of dispute about how the social sciences should proceed
turn on disputes in epistemology—the theory of knowledge—in particular, whether predictive success should be a necessary condition for knowledge, as in natural science, or whether we should adopt a different theory
of knowledge to assess the progress of the social sciences. Which theory
of knowledge we choose determines how we assess the progress the social
sciences have made in understanding phenomena in their domains.
NATURALISM VERSUS INTERPRETATION
Epistemology, as noted in Chapter 1, is the study of the nature, extent, and
justification of knowledge. Competing epistemologies are supposed to have
implications for methodology, that is, for choosing the methods that will
provide knowledge, as epistemology defines it. If the epistemology of natural
science is the only correct one, then the methods of the natural sciences are
the only ones that will provide knowledge in social science. If there are other
epistemologies, other conceptions of knowledge, ones more appropriate for
understanding human affairs, then the methods and the theories of the social sciences will inevitably differ from those of natural science.
Any comparison of progress in advancing knowledge by social and natural sciences requires an epistemological starting point: a thesis about what
constitutes knowledge and how to acquire it.
First I outline the epistemology behind the argument that understanding is the same in the natural and the social sciences. Then I set out the


  :   
counterargument that the comparison is based on several epistemological
mistakes about both social and natural sciences. Finally we will see why
all social scientists must, willy-nilly, take sides on this dispute and what it
means, not just for the epistemologies of their disciplines, but for metaphysical issues raised by their decisions about epistemology.
PROGRESS AND PREDICTION
Natural science has provided increasingly reliable knowledge about the
physical world since the seventeenth century. From precise predictions of
the positions of the planets, the natural sciences have gone on to predict the
existence and properties of chemical elements and the mechanisms of the
molecular biology of life. These predictions have given weight to the increasingly precise explanations the natural sciences provide as well. In addition to
systematic explanation and precise prediction, natural science has provided
an accelerating application in technologies to control features of the natural
world. This sustained growth of knowledge and application seems absent
from the sciences of human behavior.
In social disciplines, there seem to be moments at which a breakthrough
to cumulating knowledge has been achieved: Adam Smith’s Wealth of Nations, Durkheim’s work in Suicide, perhaps Keynes’s General Theory of
Employment, Interest, and Money, or Skinner’s Behavior of Organisms, for
instance. But subsequent developments have never confirmed such assessments. Though the social sciences have aimed at predicting and explaining
human behavior and its consequences at least since the Greek historian Thucydides in the fifth century , some say we are really no better at it than
the Greeks.
Therefore, the argument concludes, something is the matter with the social sciences; probably they are not “scientific” enough in their methods.
They need to adopt methods that more successfully uncover laws or, at any
rate, models and empirical generalizations, which can be improved in the
direction of laws or brought together in theories that explain their applications and improve on our predictive and explanatory power when it comes
to human affairs.
Why models, generalizations, and laws? It’s pretty clear that technological
control and predictive success come only through the discovery of general
regularities, which enable us to bend the future to our desires by manipulating present conditions and, perhaps more important, enable us to prevent
future misfortunes by rearranging present circumstances. The only way that
is possible is through reliable knowledge of the future, knowledge of the sort
that only laws can provide.
Progress and Prediction

There are two other less practical and more philosophical arguments for
the importance science attaches to laws. First, the kind of explanation science seeks is causal, and causal knowledge requires regularities. Second,
the certification of scientific claims as knowledge comes from observation,
experiments, and the collection of data. Only generalizations that bear on
the future can be tested by new data.
Why does causation require laws or regularities? Consider how we distinguish a causal sequence from an accidental one. Suppose I strike a wooden
match, which breaks in the middle and ignites into flame. Why do we say
that the striking caused the ignition, not the breaking in half? Because
match strikings are generally followed by flames, whereas match breakings
rarely are. Even this regularity is not exceptional. Sometimes a struck match
will not light. But to explain these failures causally and to prevent them from
happening in the future, we search for further regularities, for example, that
wet matches will not ignite. Our search ultimately leads to laws of chemical
reactions expressed in terms that don’t mention matches and their being
struck. These laws have few or no exceptions and ultimately underwrite the
rough generalizations that experience leads us to frame.
The eighteenth-century British philosopher David Hume was the first
to argue that, independent of our past experiences, there is nothing we can
directly observe in any single, observed sequence of events that enables us
to detect that the first event causes the second; there is no detectable glue attaching a cause to its effects that allows us to distinguish between causal and
accidental sequences. Hume’s observation is reflected in the methods all the
sciences have developed to distinguish causation from mere correlation: by
identifying well-confirmed regularities that stand behind individual causes
and that are absent in cases of mere correlation. Because strict, exceptionless
laws are hard to find in everyday life, we make do with rough-and-ready empirical regularities to underwrite particular causal claims. In the sciences—
natural and social—these rough regularities often take the form of statistical
generalizations. Much statistical methodology is devoted to distinguishing
merely coincidental statistical correlations from correlations that reflect real
causal sequences.
According to Hume, when science traces observed causal sequences back
to fundamental physical regularities, such as Newton’s law that bodies exert
gravitational attraction on one another, there is nothing more to them than
universality of connection. When we reach the most fundamental laws of
nature, they will themselves be nothing more than statements of constant
conjunction of distinct events. In science, a causal explanation must in the
end appeal to laws connecting the event to be explained with prior events.
Indeed, there is no stopping place in the search for laws that are more and
more fundamental. The role accorded to laws has been a continuing feature

  :   
of empiricist philosophy and empirical methodology in science ever since
Hume. And the importance of generalizations, models, and other approximations to laws in all the sciences—natural as well as social—has been
grounded on the role that laws play in underwriting causation.
Since, even in individual cases, our knowledge of causation is based on
the preliminary identification of generalizations, which themselves are refined through the repeated observation of similar sequences, it is no surprise
that such observation is what tests our explanatory and predictive hypotheses and certifies them as justified knowledge. Hume’s analysis of the nature
of causation as constant conjunction means that our knowledge of individual causal sequences is justified only if we can successfully predict further
effects when we observe their causes. If Hume is right about causation, prediction is the sine qua non of causal knowledge.
EMPIRICISM AND LOGICAL POSITIVISM IN
THE PHILOSOPHY OF SOCIAL SCIENCE
After 1900, Hume’s two insights blossomed into a philosophy of science that
held sway for half a century or more and set the agenda of problems in the
philosophy of natural and social science. This philosophy of science was
labeled by its exponents “logical positivism” and sometimes “logical empiricism” or just “positivism” for short.
Logical positivists adopted Hume’s epistemology—empiricism. This is the
thesis that our knowledge of the world can be justified only by the testimony
of the senses—that is, by experience, observation, and experiment. Logical
positivists extended this thesis to a more radical one, that theories that one
could not verify or falsify by experience are, strictly speaking, meaningless.
Using this principle, logical positivists stigmatized much nineteenth-century
philosophy, especially the work of Hegel and his followers, who advanced
theses like, “All history exhibits the self-development of reason,” which
seemed not only grandiloquent but also so vague that one couldn’t know
whether to disagree with it. Logical positivists held that such sentences were
meaningless, more like, “Colorless green ideas sleep furiously” or “‘Twas
brillig, and the slivy toves / Did gyre and gimble in the wabe,” than technical
claims of scientific theory we can’t understand until we have learned calculus
and quantum theory. Positivists wanted to limit meaningful discourse to
what could be tested by the methods of science and to the logical analysis of
the discourse that can be provided by the methods science uses.
Thus, positivists devoted much effort to analyzing the nature of theory
testing in the natural sciences. Part of their motivation for this effort was
Empiricism and Logical Positivism in the Philosophy of Social Science

their epistemological commitment to Hume’s empiricism, which accorded
scientific findings the status of best-established knowledge.
Positivists also devoted much time to developing accounts of the nature
of scientific theories and the structure of scientific explanation. Since scientific explanation uncovers causal mechanisms, it must involve laws. The theory of explanation they advanced is called the “deductive-nomological” or
“covering-law” theory: to explain a particular event, one deduces its occurrence from a set of one or more laws of nature together with a description of
the “initial” conditions that the laws require for the occurrence of the event
to be explained. Thus, we can explain why a car’s radiator burst by deducing
from the facts that the temperature fell below freezing, and the radiator was
full of water, and the law that water expands when it freezes into ice. Similarly, the positivists pointed out, given the law about the expansion of water
at its freezing temperature, we can predict that the full radiator will break
if we know that the temperature is falling below freezing. Thus, according
to positivists, explanation and prediction are two sides of the same coin.
The role of successful prediction in certifying that explanations are correct
was only the beginning of its importance for the positivists’ conception of
knowledge.
The covering-law model of explanation can be extended to account for
how science explains laws, and it can be developed into an analysis of scientific theories. Laws are explained by derivation from other, more general
laws. Thus, we can derive a chemical law—for example, that hydrogen and
oxygen will combine under certain conditions to produce water—by deducing it from more general physical laws governing the chemical bonds
produced through the interaction of electrons. A scientific theory is just a
set of very general laws that jointly enable us to derive a large number of empirical phenomena. According to the positivists, a theory has the structure
of an axiomatic system—rather like Euclidean geometry with its postulates,
or axioms, and its theorems derived from them by logical deduction. But
unlike geometry, the axioms of a scientific theory are not taken to be known
for certain. Rather, positivists held that such axioms are hypotheses, which
are tested by the deduction from them of predictions about observations. If
observations corroborate the predictions, the theory is confirmed to some
degree. But no theory is ever conclusively verified once and for all. Theories, like laws, make universal claims. Our evidence for these claims about
everywhere and always is limited to here and now and in the past. Therefore, scientific knowledge is fallible, always subject to revision, correction,
improvement, as guided by its predictions that go wrong.
To emphasize the hypothetical nature of the basic laws of a theory and
the logical relations between these laws and the observations that test them,

  :   
positivists named their account of theories “hypothetico-deductivism.” Despite the fallibility of science, the positivists (along with pretty much everyone else) held that the history of science is a history of progress, a history of
increasingly powerful predictions and increasingly precise explanations of
the way the world works. And positivists parlayed their analysis of the nature of theories into an account of this progress. Galileo’s laws and Kepler’s
laws could be mathematically derived from Newton’s, and Newton’s from
Einstein’s theories of relativity—special and general, and from quantum mechanics, while all three of these could be deduced from superstring theory.
The progressive accumulation of knowledge in science is thus certified by its
increasing predictive success.
The history of science is the history of narrower theories being “reduced”
to broader theories. One theory is reduced to another when the distinctive
fundamental assumptions of the first theory—its axioms—can be derived
as theorems from the fundamental assumptions of the broader theory. This
will ensure that the predictive successes of a theory are preserved by its successors. Thus, Galileo’s theory of terrestrial mechanics and Kepler’s theory
of the planetary orbits can be derived as special cases from one theory of
mechanics—Newton’s. And Newton’s theory turns out to be a special case
derivable from Einstein’s special theory of relativity. Similarly, the balanced
equations of chemistry follow from the physical theory of the atom, and
Mendelian genetics, discovered in the nineteenth century, turns out to be
derivable from molecular genetics. Or so positivists claimed. Most students
of science accepted this picture of the progress of science as accumulating
more and more knowledge by incorporating and preserving the predictive
successes of older theories in newer ones. But the positivists attempted to
make the picture precise by giving a formal account of the reduction of theories as the logical derivation of one axiomatic system from another.
As noted above, the positivists sought to parlay their empiricist theory
of knowledge, which made predicting observations central, into a theory of
meaning—the so-called principle of verification. This theory of meaning
has its origins in the seventeenth- and eighteenth-century British empiricist
notion that the meaning of a word is the image in the mind that it names. So
“red” means the color experience one has looking at stop signs. The positivists developed this notion into the thesis that words have “empirical” meanings, roughly given by their roles or the contributions they make to testable
sentences—whole statements whose truth or falsity can be determined by
making observations, that is, predicting them and seeing whether they are
borne out. Prediction is thus built into the positivist theory of knowledge,
of explanation, and even of language.
One serious problem for positivists was reconciling their empiricism—the
requirement that meaningful statements be testable in observation—with
Empiricism and Logical Positivism in the Philosophy of Social Science

the unobservable entities, processes, and properties of scientific theories. It
is clear that theoretically indispensable concepts like electron, charge, acid,
and gene name unobserved things that we can make no observational predictions about. Are we to stigmatize the statement, “Electrons have negative
charge” as meaningless because it cannot be tested? No. Positivists held that
statements that use such concepts can be indirectly tested by the derivation
from them of statements that are about observations. The trouble with this
approach is that no particular observations follow from any single theoretical statement; experimental observations follow only from large sets of theoretical statements working together. So we cannot give the observational
pedigree of a single term like “electron” or of just one statement, like “Electrons have quantized angular momentum.”
The realization of this fact about the relation between observations and
theory began the unraveling of positivism as a philosophy of science. If it is
through their predictions that scientific claims face the court of experience,
not one by one but in large groups, then we cannot distinguish the theoretical statements or individual concepts in the group that are meaningful from
the ones in the group that may not be. Moreover, when observations disconfirm a set of theoretical statements that work together to imply observations
that don’t occur, they force us to give up one or more members of this set.
But they don’t point to the one we have to give up, and we can always save
our favorite hypothesis from disconfirmation by making any one of a large
number of other possible changes in our theory.
This “underdetermination of theory by evidence” has serious consequences for empiricist philosophy of science. Most important, it suggests
that theory choice may not be governed exclusively or perhaps even largely
by observation, as empiricism requires. That is because scientists’ observational evidence does not confirm the theory scientists actually endorse any
more strongly than it does any number of alternatives we can construct by
making slight changes in the theories scientists actually believe. Accordingly, when scientists embrace specific theories, it cannot be just because
those are more strongly confirmed by observation than others. It must be
that the scientists’ theoretical choices are driven by nonempirical factors.
What might the nonempirical factors be? By the 1970s, many philosophers of science were beginning to search for these factors. Searching for the
nonempirical factors means, in effect, giving up positivism and its empiricist
epistemology as an account of scientific method. In fact, it may involve giving up the positivist and empiricist claim that science provides ever increasing objective knowledge. For if the factors that govern theory choice are, say,
psychological or social or ideological instead of empirical and logical, then
the source of insight into the nature of science will be psychology or sociology or political science or economics or history. Certainly the only way to

  :   
tell how science has actually proceeded is to explore the history of science.
So, by the 1970s, the history of science had become a crucial component of
any attempt to understand the nature of natural science. Soon after that,
sociologists began to seek nonevidential factors that determine scientific
consensus among social forces. Eventually, each of the social sciences could
boast of a subdiscipline devoted to understanding the character of science
and scientific change. Besides the sociology of science, there emerged, for
example, the economics of science, which sought to show how scientific research reflects the rational distribution of scarce research resources in the
face of uncertainty. Students of gender and gender politics sought to show
that scientific practices, and in some cases scientific theories, were the result
of male domination and discrimination based on race, class, and gender.
These enterprises had little influence on the course of the philosophy of science itself, though they had a good deal of impact on the ways in which each
of these social sciences viewed itself as a science. And all of them reduced the
influence of logical positivism within the social and behavioral disciplines.
Meanwhile, for quite different reasons, logical positivism as a viable
movement among philosophers had disappeared by the 1970s. This disappearance was not due to the acceptance among philosophers of views that
cast doubt on science’s objectivity or improvement. The eclipse of logical
positivism was due much more to philosophers’ own studies of the history
of science, and especially sciences such as biology. What they learned was
that scientific developments in these disciplines do not honor the narrow
strictures of positivism. Moreover, no one could solve the problems required
to vindicate its philosophical program. What former positivists and their
students remained wedded to was a vision of science as objective knowledge, which, though fallible, is characterized by persistent improvement in
explanatory depth, as revealed in its predictive power for observations. In its
expansion and deepening of our understanding, models, general laws, and
universal quantitative theories continued to be recognized as playing an indispensable role. For these philosophers and for social scientists influenced
by them, the question remains why the same sort of progress in providing
models, laws, and theories of ever-increasing predictive power with respect
to observations—which is a feature of natural science—does not characterize the social sciences.
THE EMPIRICIST’S DIAGNOSIS: WHY SOCIAL
SCIENCE FAILS TO DISCOVER LAWS
Why have the social sciences not provided increasing amounts of cumulating scientific knowledge with technological payoff for predicting and
The Empiricist’s Diagnosis: Why Social Science Fails to Discover Laws

controlling social processes? The social sciences have failed, despite long
attempts, because they have not uncovered laws or even empirical generalizations that could be improved in the direction of real laws about human
behavior and its consequences. This diagnosis calls for both an explanation
of why no laws have been discovered and a proposal about how we can go
about discovering them.
One compelling explanation is that social science is just much harder
than natural science: the research object is us, human beings, and we are
fiercely complicated systems. It is therefore no surprise that less progress can
be made in these disciplines than in ones that deal with such simple objects
as quarks, chemical bonds, and chromosomes. After all, the human being is
subject to all the forces natural science identifies as well as those of psychology, sociology, economics, and so forth. Teasing out the separate effects of
all the forces determining our behavior is more formidable a task than that
which faces any other discipline.
Add to this the restrictions of time, money, and morality on the sorts of
experiments needed to uncover causal regularities, and the relatively underdeveloped character of social science should be no surprise. Perhaps the
complexity of human behavior and its causes and effects are beyond our
cognitive powers to understand. Perhaps there are laws of human behavior,
but they are so complicated that human beings are not clever enough to
uncover them. Or perhaps we haven’t given social science enough time and
effort; perhaps breakthroughs in, say, computer simulation will enable us to
extract models, generalizations, and eventually laws from data about human
behavior. On this view, the social sciences are just “young sciences.” By and
large, they are or can be scientific enough in their methods, but they just
require more time and resources to produce the social knowledge we seek.
These explanations for the failure of social science to enable us to discover
the laws governing human behavior have not convinced many students of
the social sciences. Is human behavior so much more complicated than
nonhuman processes? Sure, but science has always successfully coped with
complexity in other cases. Are the social sciences really young by comparison with the natural sciences? From when should we date the social disciplines? From the post–World War II infusion of research money, statistical
methods, cheap computation, and improved scientific education of social
scientists? Should we date social science from the self-conscious attempts,
like Durkheim’s in the late nineteenth century, to establish a quantitative
science of society? Or should we go back to the eighteenth-century Marquis
de Condorcet’s or the seventeenth-century Thomas Hobbes’s development
of rational choice theories of human behavior? Some will argue that the
search for laws in social science goes back to Thucydides and the Peloponnesian War in the fifth century . Certainly the desire to understand and

  :   
predict human behavior is at least as old as the desire to understand natural
phenomena, and the search for laws of human behavior goes back past Machiavelli for sure.
For some philosophers and for even more social scientists, the claim that
human behavior is too complicated to understand or that the social sciences
are young rings hollow. Twentieth-century behaviorists in all the social sciences provided good illustrations of these attitudes. These social scientists
offer a different explanation for the failure to discover laws. To begin with,
behaviorists didn’t accept the argument from the complexity of human beings to the difficulty of discovering laws about them. They note that as natural science developed, its subject matter became more complex and more
difficult to work with. Indeed, to advance knowledge in physics nowadays,
vast particle accelerators must be built to learn about objects on which it
is extremely difficult to make even the most indirect observations. But the
increasing complexity of research in the natural sciences has not resulted in
any slowdown in scientific advance. Quite the contrary, if anything the rate
of “progress” has increased over time. Thus, by itself, complexity can hardly
be an excuse for the social sciences.
Moreover, the argument continues, the social sciences have had a great
advantage over the natural sciences. It is an advantage that makes their comparative lack of progress hard to explain as merely the result of complexity
and the difficulties of experimentation. In the natural sciences, the greatest
obstacle to advancement has been conceptual, not factual; that is, advances
have often been the result of realizing that our commonsense descriptive
categories needed to be changed because they were a barrier to discovering
generalizations. Thus, the Newtonian revolution was the result of realizing
that commonsense notions about change, forces, motion, and the nature of
space needed to be replaced if we were to uncover the real laws of motion.
We had to give up our commonsense suppositions that there is a preferred
direction in space, that the earth is at rest, that if something is moving there
must be a force acting on it. Instead we must view motion at constant velocity as the absence of net forces, consider “down” as just the direction
toward the strongest local source of gravity, and accept that Earth is moving
at about seven hundred feet per second. Similarly, the pre-Darwinian conception of unchangeable, immutable species must be surrendered if we are
ever to coherently entertain an evolutionary theory, still less to accept one
that explains diversity by appeal to blind variation and natural selection
changing old species into new ones.
But in the social sciences, the change of fundamental categories has not
been thought necessary. In fact, since the very beginnings of the philosophy of social science in the late nineteenth century, it has been argued
The Empiricist’s Diagnosis: Why Social Science Fails to Discover Laws

by important social scientists and philosophers that these disciplines must
invoke the same framework of explanatory concepts that people use in everyday life to explain their own and other people’s actions—the categories
of beliefs, desires, expectations, preferences, hopes, fears, wants, that make
actions meaningful or intelligible to ourselves and to one another. The reason
often given for insisting on explanations of behavior that show its meaning
for the agents who engage in it is that the perspective of social science is fundamentally different from that of natural science. The perspective of natural
scientists is that of spectators of the phenomena they seek to discover. The
social scientist is not just a spectator of the social domain, but a participant,
an agent, a player in the human domain. Theories in natural science cannot
change the nature of the reality that the physicist or chemist or biologist
studies, but theories in social science can and often do. As participants in social life learn about these theories, their actions may change in light of them.
This goes for social scientists as well as those whose actions and behavior
they study. If laws and theories in social science must be ones that reveal the
meaning of behavior and make it intelligible to the human agents who engage in it, they will have to employ the categories and concepts in which we
humans have always understood our own actions and their consequences for
others. Notice that those who hold this view need to give us an argument for
why the social scientist cannot adopt the perspective of observer and must
adopt the perspective of participant. The fact that many social scientists do
so is not an argument that they inevitably must.
On the other hand, even those who seek a social science that, like natural
science, provides only an observer’s perspective and not that of a participant,
more often than not embrace the conceptual repertoire of common sense.
For what they seek are the causes and consequences of our actions, and they
agree with common sense that these actions are determined by our desires
and our beliefs. Accordingly, almost all social scientists have long searched
for models, generalizations, and ultimately laws connecting actions, beliefs,
and desires.
It is therefore hard to avoid the conclusion that the social sciences never
had to face the greatest obstacle to advancement in the natural sciences:
the need to carve out entirely new ways of looking at the world. Thus, we
might expect progress to have been possible or perhaps even more rapid in
the social than in the natural sciences. The absence of progress makes the
excuse—that these are young disciplines that face subjects of great complexity—unconvincing to many social scientists and some philosophers.
In fact, behaviorists and others argue that the basic categories of social
science are wrong. The reason no laws have been uncovered is that the categories of action, desire, belief, and their cognates have prevented us from

  :   
discovering the laws. And many social scientists seek to supplant those categories with new ones, for example, operant conditioning, sociological functionalism, and sociobiology.
It is easy to see how a category scheme can prevent us from uncovering laws or regularities, even when they would otherwise be easy to find.
Suppose we define fish as “aquatic animal” and then attempt to frame a
generalization about how fish breathe. We do so by catching fish and examining their anatomy. Our observation leads to the hypothesis that fish
breathe through gills. Casting our nets more widely, we begin to trap whales
and dolphins, and then we modify our generalization to “all fish breathe
through gills, save whales and dolphins.” But then we start to drag along the
ocean floor and discover lobsters, starfish, and crabs, not to mention jellyfish
floating at the surface, all breathing in different ways. There’s no point in
adding more and more exceptions to our generalization. There isn’t just one
generalization about how all fish breathe, not as we have defined fish. The
trouble is obvious: it’s our definition of fish as aquatic animal. A narrower
definition, such as “scaly aquatic vertebrate,” will not only, as Aristotle says,
“carve nature closer to the joints”—that is, reflect its real divisions more accurately—but also enable us to frame simple generalizations that stand up to
testing against new data. Indeed, the difference between a “kind-term” like
gold, which reflects real divisions in nature, and one like fake gold, which
does not, is that there are laws about the former and not the latter. Philosophers call the kind-terms that figure in laws “natural kinds.” The search
for generalizations and laws in science is at the same time a search for these
natural kinds.
What if the terms desire, belief, action do not name natural kinds? What
if they just don’t carve nature at the joints? Then, like our example, fish,
every generalization that employs those terms will be so riddled with exceptions that there are no laws we can discover stated in these terms. In
consequence, the explanations that employ them would inevitably have little predictive power. One solution to the social sciences’ problem would be
to find new explanatory variables to replace the “unnatural” kinds. Social
scientists are infamous for introducing such neologisms—terms like reinforcer from behaviorism, repression from psychoanalysis, alienation from
Marxian theory, or anomie from Durkheim’s sociological tradition. Advocates of each of these theories promise that the application of their preferred
descriptive vocabulary will enable the social sciences to begin to progress in
the way the natural sciences have. If these social scientists are correct, their
disciplines will indeed turn out to be young sciences. For in the absence of
their preferred system of kinds and categories, the social sciences are rather
like chemistry before Lavoisier: trying to describe combustion in terms of
Rejecting Empiricism for Intelligibility

“phlogiston” instead of “oxygen,” and failing because there is no such thing
as phlogiston.
It is important to keep in mind that social scientists and philosophers
have challenged every step in this chain of reasoning: the claim that the natural sciences show progress and the social sciences do not; the assumptions
about what progress in the growth of knowledge consists in; the role of laws
in providing knowledge; the purported explanations of why the social sciences have not yet uncovered any laws; and the prescriptions about how they
should proceed if they hope to uncover laws. Let us examine this challenge.
REJECTING EMPIRICISM FOR INTELLIGIBILITY
Those who reject the argument that natural science has progressed and social science has languished take up their counterargument at the very foundations of the philosophy of natural science. To begin with, it is sometimes
held that the natural sciences have not in fact made the kind of progress
ordinarily attributed to them. In making that point, they sometimes exploit
the account of science advanced in Thomas Kuhn’s Structure of Scientific
Revolutions. This book has been the most heavily cited work among social
scientists writing about method and philosophy in the years since 1962 when
it was published. Kuhn’s book did more to undermine the dominance of
logical positivism in the philosophy of natural and social science than any
other single volume. It challenged several of the details of the positivist picture of science sketched in our discussion above. In particular it raised two
challenges against the claim that predictive success is a universal criterion in
all disciplines of progress in attaining knowledge. First, Kuhn gave reason to
suppose that natural sciences do not really progress in the way orthodox history of science portrays. Second, and more radically, he challenged the role
of prediction in the epistemology of natural science altogether, arguing that
the demand for it was a temporary fashion of Enlightenment science. Some
of Kuhn’s readers, especially social scientists, interpreted Kuhn as claiming
that instead of progress, the history of scientific theories from Aristotle to
Einstein has been characterized by change without overall improvement.
Thus the history of science is the succession of very general theories, or
what Kuhn called “paradigms,” that replace one another, without making
net improvements on their predecessors.
According to the opponents of the thesis that science shows cumulative
“progress,” the reason scientific theories do not build on their predecessors is
very roughly that they constitute irreconcilably different conceptual schemes.
Their notion is that scientific theories are more like poems, meaningful in

  :   
one language but never adequately translated into another. Accurate comparisons between theories is impossible, for too much is lost in translation
from one to another. Of course, a theory-free language to describe observations would enable us to compare two theories for predictive success if they
shared the same observation language. But there is no such theory-neutral
stance, and therefore one theory’s confirming data will be another theory’s
experimental error. In retrospect, the absence of a theory-neutral language
of observation can explain most of the failures of the logical positivists’ program for understanding science and vindicating an empiricist epistemology.
The claim that science shows persistent improvements in predictive success
about what can be observed certainly becomes more controversial.
The appearance of progress in science, Kuhn held, is the result of scientists in each generation rewriting the history of their subjects so that the
latest view can be cloaked in the mantle of success borne by the scientific
achievements it replaces. Positivists failed to see this situation because they
accepted the early twentieth-century rewrite of the history of physics as the
objective truth about what actually happened in the history of science. Instead, it was just part of the new paradigm’s attempt to obscure its victory
over the old one. Careful study of the cases of what positivists called the
reduction of narrower theories by broader ones—like the reduction of Newton’s theory to Einstein’s—will show that no such thing took place. Succeeding theories are incompatible with one another, so neither can be derived
from the other. What can be reduced to the later, newer theory is just the
newer theory’s inadequate “rewrite” of the older theory.
In fact, Kuhn seemed to claim that the whole idea that predictive success should constitute a transdisciplinary criterion for scientific knowledge
is part of a conceptual scheme: positivism, or empiricism, associated with
Newtonian science. But this paradigm has now been replaced in physics by
the theories of relativity and quantum mechanics. The conceptual schemes
of those theories deprive Newtonian demands on scientific method of their
authority. Newtonian science made prediction a requirement of scientific
achievement because it was a deterministic theory of causal mechanisms.
But quantum mechanics has revealed that the world is indeterministic; thus,
definitive prediction can no longer be a necessary condition of scientific
success. Nor does it make sense to search for causal mechanisms described
in strict and exceptionless laws. The fundamental laws of quantum physics
are statements of probabilities.
For the same reasons that scientific standards change within each of the
natural sciences, they differ extensively between them and across the divide
between natural and social sciences. Indeed the differences between standards in the natural and social sciences must be wider than the others. The
Rejecting Empiricism for Intelligibility

reason for these differences is of course the vastly different “paradigms” that
ground theory in each discipline. Thus, the charge that the social sciences
have made less progress than the natural sciences is often said to rest on a
myopic absolutism—a view that improperly generalizes from the methodological recipes of the obsolete paradigm of Newtonian physics.
Of course, within some disciplines prediction and practical application
are important ways of “articulating the paradigm.” But to identify what kinds
of predictions, if any, are appropriate to a social science, we must first identify
its “ruling paradigm.” If we find the right paradigm, we will be able to see that
in the light of its standards, the social sciences are progressing perfectly well,
thank you. We will see that whether there is as much progress in the social
as in the natural sciences is a question not worth answering. The differences
between progress in physics or biology and the type of progress that characterizes the human sciences are too great even to allow comparison.
Unlike the natural sciences, which aim at causal theories that enable us to
predict and control, the social sciences seek to explain behavior by rendering
it meaningful or intelligible. They uncover its meaning, or significance, by
interpreting what people do. The interpretation of human behavior, in this
view, is not fundamentally causal. Nor is intelligibility provided by the discovery of laws or empirical generalizations of any interesting sort.
The social sciences are concerned with the part of human behavior ordinarily described as action and not with mere movements of, or at the surface of, the body. Speech, not snoring; jumping, not falling; and suicide, not
mere death, are the subject matter of some of the social sciences. The parts
of social science that do not deal directly with individual action—demography, econometrics, and survey research, for example—deal with actions’
consequences and their aggregation into large-scale events and institutions.
Though understanding the meaning of actions is not directed at uncovering causes, it certainly satisfies some standards of predictive success: the
correct interpretation of human actions enables us to navigate successfully
in a society of other human beings. When we step back and consider the
reliability of the predictions we make regarding the behavior of others, we
cannot fail to be impressed with the implicit theory that growing up in society has provided us. This theory, known as “common sense” or “folk psychology,” tells us obvious things we all know about ourselves and others.
For instance, people do things roughly because they want certain ends and
believe their actions will help attain them. Folk psychology includes such
obvious truths as that being burned hurts, medium-sized objects in broad
daylight are detected by normal observers, and thirst causes drinking.
Folk psychology is a theory in which we repose such great confidence that
nothing in ordinary life would make us give it up. That leads some people to

  :   
hold that folk psychology isn’t a theory, but something more fundamental, a
“form of life,” a way of living. After all, a theory is something we could give
up; it is composed of models, empirical generalizations, or even laws that
are subject to testing by experience. But when we try to express the central
principles of folk psychology, we seem to produce only banal and obvious
principles or ones with gaping exceptions. It’s probably true by definition
that people act in ways that they believe will attain their desires. And it’s
plainly false that thirst always causes drinking. We can dream up lots of
exceptions to that generalization. If folk psychology is a theory, it’s surely
very different from theories in the natural sciences.
Whether it is a scientific theory or not, folk psychology is still the best
theory we have for predicting the behavior of people around us, and it’s
the one we employ when we explain our own and others’ behavior. What
is more, folk psychology had already reached a high degree of predictive
power well before the dawn of recorded history, long before we acquired
a comparably powerful theory in natural science. Folk psychology enables
us to predict by identifying the meaning of behavior—by showing that it is
action undertaken in the light of beliefs and desires.
Social science, it is argued, is and should be the extension and development of this resource. It inherits the predictive strength of folk psychology.
But unlike natural science, the main aim of social science is not to increase
the predictive power of folk psychology. Rather, the aim is to extend folk
psychology, from the understanding of everyday interactions of individuals,
to the understanding of interactions among large numbers of individuals in
social institutions, and among individuals whose cultures and forms of life
are very different from our own.
Opponents of a “scientific” approach to the social sciences claim that
much of the apparent sterility and lack of progress in these disciplines is
the result of slavish attempts to force folk psychology into the mold of a
scientific theory of the causes and effects of action. The social scientists and
philosophers who oppose the scientific approach and those who support it
agree that in certain areas the social sciences have not progressed. But the
diagnosis of the former does not blame the lack of progress on the complexity of social life, the inability to undertake experiments, or the failure to
find the appropriate “natural kinds.” Rather, opponents of naturalism hold
that many social scientists have misunderstood folk psychology, mistakenly
treating it as a causal theory, to be improved by somehow sharpening its
predictive power. The result, as occurred in microeconomic theory, for example, has been to produce general statements that are not laws: they are either vacuous definitions or flatly false statements. In other disciplines, such
as psychology or parts of sociology, that misunderstanding has produced
jargon-ridden pseudoscience.
Rejecting Empiricism for Intelligibility

The explanation of why parts of social science seem to find themselves
at a dead end can be found here. Folk psychology has reached its maximal
level of predictive power. That is because folk psychology is not a causal
theory, to be improved by the kind of means scientists employ to improve
theory in natural science. The predictive power of folk psychology is a sort
of by-product of its real goal, which is to provide understanding through
interpretation. When we accept this as the aim of social science, we will recognize the important advances it has attained. Doubts about progress will
be shown to be not only groundless but also fundamentally misconceived.
Proponents of this view invite us to consider how much more we now
know about other cultures, their mores, morals, institutions, social rules
and conventions, values, religions, myths, art, music, and medicine, than
we knew a century ago. Consider how much more we know about our own
society as a result of what we have learned about other societies. Our understanding of these initially strange peoples is not the product of “scientific investigation.” It is the result of the cultural anthropologist’s “going
native,” attempting to learn about a foreign culture from the inside, coming
to understand the meaning of his subject’s actions in the terms his subject
employs. Our understanding also reflects important discoveries about the
hidden, deeper meanings behind behavior, meanings that social scientists
have revealed.
This hard-won knowledge represents progress in two ways. We can understand people of differing cultures. Indeed, we can acquire as much predictive confidence about them as our own folk psychology provides us about
ourselves. For what we are learning is in effect their folk psychology. Moreover, learning about other cultures teaches much about our own. Specifically,
it leads us to see that what we might identify as universal or true or optimal
in our beliefs, values, and institutions is really parochial, local, and merely
convenient for some of us. Coming to understand another and very different
society by learning the meaning of its features is a cure for moral absolutism,
xenophobia, racism, and other ills. That is how social science progresses. It
is not meant to provide us with the means to control the behavior of others.
Indeed the understanding it provides may enable us to free ourselves from
the (often unnoticed) control of others or society itself. Rather, social science
is meant to provide interpretations of the actions of others and of ourselves
that will enable us to place our own society in perspective.
Another thing that a scientific approach to human behavior misses, by
substituting causal inquiry for understanding the meaning of human action,
is the moral dimension of social science. The natural sciences aim, in part,
at technological progress. That’s what makes predictive power so important
for them. The social sciences aim at improving the human condition. This
aim entails choices the natural sciences are not called upon to make, moral

  :   
choices about what will count as improvements and what will not. Making
these choices requires us to identify the real meaning of social institutions,
as opposed to their apparent meaning. That is why so much social science
involves the critique of social arrangements and institutions as unjust to one
group or another. This critique will eventually emancipate human beings
from their mistaken beliefs about the meanings of social events and institutions of control and exploitation.
Influential social scientists since Max Weber have argued that theories
about human behavior, human action, and human institutions need to
uncover both causal laws and interpretative meanings, and that this dual
requirement is what distinguishes social from natural sciences. But many
social scientists have held that the conceptual apparatus we need to uncover
the meaning of human events, individual or aggregate, is irreconcilable with
the search for causal laws. In their view, the idea that we should replace
our explanatory system with one that “carves nature at the causal joints” is
based on a fundamental misunderstanding of the nature and aims of social
science. One central philosophical problem for this view is very clear: What
sort of conceptual confusion has led so many philosophers and social scientists down the blind alley of attempts to construct and advance a discipline
that apes the wildly inappropriate methods of natural science? How could
so many smart people be so wrong about what they are doing?
TAKING SIDES IN THE PHILOSOPHY OF SOCIAL SCIENCE
The two arguments summarized in this chapter cover a lot of ground, including both very practical questions of social scientific method and the
most fundamental problems of philosophy. The arguments reflect polar extremes on a continuum along which most social scientists should be able
to locate themselves. But though they are extreme views, these positions
have real proponents. More important, all social scientists take sides on the
problems the positions reflect, whether they want to or not. And that is what
makes the philosophy of social science relevant to social science.
The extreme views, that social science is not scientific enough and that
it is not supposed to be scientific at all, disagree on too much ever to be
reconciled. No one is going to convince a proponent of either extreme that
the view on the other end of the continuum is right. The reason is that the
differences between them rest on very fundamental issues of philosophy,
claims about epistemology, metaphysics, and ethics. Since these issues were
first raised by Plato almost 2,400 years ago, philosophers have not been able
to settle them.
Taking Sides in the Philosophy of Social Science

Why should the rest of us bother about these issues? They cannot be
settled, and we don’t occupy the extreme positions in the philosophy of social science. For these reasons, many social scientists aren’t interested in the
subject at all. They seem to have a good reason not to be, if the problems of
philosophy are insoluble. And yet, however insoluble these problems may be,
they certainly are not irrelevant.
Between the polar extremes in the philosophy of social science, there are
many intermediate theories of the nature of social science that seek to reconcile the social sciences’ differences from natural science with the demand
that they be truly scientific in the natural scientist’s sense of the term. But
partisans of the extreme views agree with one another that such compromises are in one way or another incoherent—attempts to have one’s cake
and eat it, too. In trying to give the differences and the similarities between
the sciences—social and natural—their due, the compromises turn out to
be contradictory or inconsistent, or just plain false. In philosophical matters, the policy of finding a happy medium that splits the difference between
rival theories is often impossible, for the positions are logically incompatible. Picking and choosing components of these two philosophies, with a
view to developing a “third way,” may result in an incoherent position. Or
if the position is coherent, the resulting theory may not be strong enough
to withstand the arguments advanced by proponents of one or both of the
extreme positions.
For example, economists and political scientists are committed to explaining action in terms of the quantitative models of “rational choice theory,” according to which individual expectations and preferences cause
human actions. These social scientists need to explain why we have secured
no predictively reliable laws about the causes of individual action. It won’t do
simply to say that rational choice models are idealizations or approximations
like those of, say, physics—approximations that will eventually be improved
in the direction of laws. For ideal models in physics have what economic
models conspicuously lack: great predictive power. In addition, physical,
chemical, and biological models seem capable of refinement in ways no rational choice theory has yet manifested.
Of course, economists might try to show why no regularities governing
human action are necessary to explain economic choice causally. Such an
argument would in effect claim for economic processes a sort of causality
unknown in natural science. That causality would need a significant and
unavoidably philosophical explanation. Alternatively, economists could
adopt the view that the knowledge they provide is not causal but, at best,
information that helps us interpret the actions of consumers in late capitalist
society. Anyone acquainted with modern economic theory will recognize

  :   
that view as unacceptable to economists. But unless they can provide some
philosophical underpinning for their theory, economists are vulnerable to
the charge that their explanatory variables are not natural kinds. That is, the
explanatory variables of rational choice theory need to be surrendered in
any serious causal theory of human behavior. In effect, finding an intermediate position for economics involves facing several classical metaphysical
problems about causation.
The sociologist or cultural anthropologist faces a different sort of philosophical problem. Anyone who brings back an account of the meanings of
actions, rituals, or symbols of other cultures must assure us that the account
is correct, that it constitutes an addition to knowledge about the culture
reported. How can we tell whether it is information or misinformation?
This question becomes especially pressing if we, like some anthropologists, reject the demand that our theories about cultural meanings identify
causes and have predictive consequences we can test. Social scientists who
reject improving predictive success as a mark of knowledge have taken sides
willy-nilly in the most profound disputes of epistemology. For certification
as knowledge by means of observed predictions is the touchstone of empiricism. The only alternative these social scientists can adopt is some version
of rationalism, the epistemology according to which at least some of our
knowledge is justified independently of experience, a priori.
Social scientists who wish to embrace the natural scientific approach to
human behavior often also hope to learn from their research how morally
to better the human condition. They must face several of the thorniest problems of moral philosophy. First, they must show how to derive what ought
to be the case—the moral improvement—from what is the case, as revealed
by social research. This is a derivation widely held to be impossible by philosophers. What is needed is nothing less than an explanation of how we
can acquire moral knowledge “scientifically.” Moreover, if acquiring moral
knowledge is possible, they must show why such knowledge does not justify
paternalistic imposition of its particular claims on a potentially unwilling
society.
NATURALISM VERSUS INTERPRETATION
Most empirical social scientists believe that prediction and interpretation
can be reconciled. They believe that there is a causal theory of human behavior and that we can uncover models, regularities, and perhaps eventually
laws that will enable us to predict human action. Let us call such social scientists “naturalists” to indicate their commitment to methods adapted from
Naturalism Versus Interpretation

the natural sciences. Naturalists believe they can endorse the methods of
natural science while doing justice to the meaningfulness and significance
of human action. Thus, they do not think anything can force them to choose
between these two commitments.
Naturalism’s pursuit of reconciliation of prediction and interpretation
has been subject to repeated objection over the course of the past hundred
years. Many current controversies about social science are but reiterations
of this objection and replies to it. Prominent social scientists, historians,
philosophers, and cultural critics have held that we cannot do justice to actions as meaningful while at the same time seeking a naturalistic or scientific explanation of them. These critics of naturalism hold that the aim of
the social sciences must be interpretation, and this means they cannot be
experimental, empirical, or predictive sciences. They have adopted a succession of labels since the late nineteenth century: idealists, phenomenologists,
structuralists, ethnomethodologists, and students of semiotics, hermeneutics, postmodernism, and deconstruction. These views share a rejection of
naturalism and a commitment to interpretation. Therefore, we may refer to
their views as antinaturalism or interpretative social science.
The history of science presents both naturalists and antinaturalists with
the same problem. Prehistoric civilizations explained all natural events—
especially catastrophes—in terms of the purposes of supernatural agents.
Today, religions continue to do so. In each of the revolutions in Western science, the greatest obstacle to scientific advance has been the conviction that
only purposes or meanings that made things intelligible could really explain
them. The history of natural science is one of ever-increasing explanatory
and predictive power. Science has achieved that by successively eliminating
meaning, purpose, or significance from nature. After Galileo, the stars and
planets were deprived of the goals Aristotelian science attributed to them;
then Newton showed that force, acceleration, and gravitational attraction
were enough to explain all motion. Eventually Darwin showed that the fitness of flora and fauna to their environments was to be explained without
attributing purpose to them or intentions to their creator. Contemporary
molecular biology has revealed the purely chemical underlying mechanism
for all the biological processes that seemed originally to be explained by the
goals they seek. Now the only arena in which explanations appeal to purposes, goals, intentions, and meaning is their “home base,” human action.
The record of the history of science requires every social scientist to face
the question, Why should human behavior be an exception to this alleged
pattern? Why should meaning, purpose, goal, and intention, which have no
role elsewhere in science, have the central place they occupy in social science? The obvious answer is that people, unlike (most) animals, vegetables,

  :   
and minerals, have minds, beliefs, desires, intentions, goals, and purposes.
These things give their lives and actions meaning, significance, make them
intelligible. But what is so different about minds from everything else under
heaven and earth that makes the approach to understanding people so
different or so much more difficult than everything else? Every potential
answer to this question is general enough, metatheoretical enough, and abstract enough to count as an exercise in the philosophy of social science.
Introduction to the Literature
Among introductions to the philosophy of social science, the best of the
earlier generation of texts is Alan Ryan, Philosophy of Social Sciences. Ryan
has also edited an anthology on the subject, The Philosophy of Social Explanation. Michael Martin and Lee McIntyre, eds., Readings in the Philosophy
of Social Science, is by far the most complete anthology of influential papers
written in the past generation on the subject. It supersedes two older anthologies that include many papers discussing topics treated in this book, L. I.
Krimerman, ed., The Nature and Scope of Social Science, and M. Brodbeck,
ed., Readings in the Philosophy of Social Sciences. Introductions to the literature in subsequent chapters will identify relevant papers in Martin and
McIntyre. Papers in Martin and McIntyre of special relevance to this chapter are F. Machlup, “Are the Social Sciences Really Inferior?” and M. Scriven,
“A Possible Distinction Between Traditional Scientific Disciplines and the
Study of Human Behavior.” Steel and Guala’s recent The Philosophy of Social
Science Reader reprints many important articles. Introductions to the literature in future chapters will identify appropriate readings from Krimerman,
MacIntyre and Martin, and Steel and Guala.
The locus classicus of the naturalistic view of method in social science is
David Hume, Treatise of Human Nature and…
Purchase answer to see full
attachment