AMERICAN SOCIOLOGICAL REVIEW
December, 1964 Volume 29, No. S
BRINGING MEN BACK IN *
GEOKGE C. HOHANS
Harvard University
A theory of a phenomenon is an explanation of it, shottmg how it foUows as a conclusion
from general propositions in a deductive system. With all its empirical achievements, the
functional school never produced a theory thai was also an explanation, since jrom its general
propositions about the conditions of social equilibrium no definite conclusions could be drawn.
When a serious effort is made, even by functionalists, to construct an explanatory theory, its
general propositions tum out to be psychological—propositions about the behavior of men,
not the equilibrium of societies.
IAM going to talk about an issue we have
worried over many times. I have worried
over it myself. But I make no excuses
for taking it up again. Although it
is an old issue, it is still not a settled one,
and I think it is the most general inteUectual
issue in sociology. If I have only one chance
to speak ex cathedra, I cannot afford to
say something innocuous. On the contrary,
now if ever is the time to be nocuous.
In the early 'thirties a distinct school of
sociological thought was beginning to form.
Its chief, though certainly not its only, intellectual
parents were Durkheim and Raddiffe-Brown.
I call it a sch(X)l, though not
all its adherents accepted just the same
tenets; and many sociologists went ahead
and made great progress without giving
a thought to it. The school is usually called
that of structural-functionalism, or functionalism
for short. For a whole generation
it has been the dominant, indeed the only
distinct, school of sociological thought. I
think it has run its course, done its work,
and now positively gets in the way of our
understanding social phenomena. And I
propose to ask. Why?
•Presidential Address delivered at the annual
meeting of the American Sociological Association
in Montreal, September 2, 1964.
THE INTERESTS OF FUNCTIONALISM
I begin by reminding you of the chief
interests and assumptions of fimctionalism,
especially as contrasted with what it was
not interested in and took for granted,
for the questions it did not ask have returned
to plague it. If what I say seems
a caricature, remember that a caricature
emphasizes a person's most characteristic
features.
First, the school took its start from the
study of norms, the statements the members
of a group make about how they ought to
behave, and indeed often do behave, in
various circumstances. It was especially
interested in the duster of norms called
a role and in the duster .of roles called an
institution. It never tired of asserting that
its concern was with institutionalized behavior,
and that the unit of social analysis
was not the acting individual but the role.
The school did not ask why there should
be roles at all.
Second, the school was empirically interested
in the interrelations of roles, the
interrelations of institutions: this was the
structural side of its work. It was the sort
of thing the social anthroptologists had been
doing, showing how the institutions of a
primitive society fitted together; and the
810 AMERICAN SOCIOLOGICAL REVIEW
sociologists extended the effort to advanced
societies. They would point out, for instance,
that the nuclear family rather than some
form of extended kinship was characteristic
of industrialized societies. But they were
more interested in establishing what the interrelations
of institutions were than in why
they were so. In the beginning the analyses
tended to be static, as it is more convincing
to speak of a social structure in a society
conceived to be stable than in one undergoing
rapid change. Recently the school
has turned to the study of social change,
but in so doing it has had to take up the
question it disregarded earlier. If an institution
is changing, one can hardly avoid
asking why it is changing in one direction
rather than another.
Third, the school was, to put it crudely,
more interested in the consequences than
io the causes of an institution, particularly
in the consequences for a social system considered
as a whole. These consequences
were the functions of the institution. Thus
the members of the school never tired of
pointing out the functions and dysfunctions
of a status system, without asking why a
status system should exist in the first place,
why it was there to have functions. They
were especially interested in showing how
its institutions helped maintain a society
in equilibrium, as a going concern. The
model for research was Durkheim's effort
to show, in The Elementary Forms of the
Religious Life, how the religion of a primitive
tribe helped hold the tribe together.
Such were the empirical interests of functionalism.
As empirically I have been a
functionalist myself, I shall be the last to
quarrel with them. It is certainly one of
the jobs of a sociologist to discover what
the norms of a society are. Though a role
is not actual behavior, it is for some purposes
a useful simplification. Institutions
are interrelated, and it is certainly one of
the jobs of a sociologist to show what the
interrelations are. Institutions do have consequences,
in the sense that, if one institution
may be taken as given, the other kinds of
institution that may exist in the society are
probably not infinite in nimiber. It is certainly
one of the jobs of a sociologist to
search out these consequences and even,
though this is more difficult, to determine
whether their consequences are good or
bad for the society as a whole. At any rate,
the empirical interests of functionalism
have led to an enormous amount of good
work. Think only of the studies made by
Murdock ' and others on the cross-cultural
interrelations of institutions.
As it began to crystallize, the functional
school developed theoretical interests as
well as empirical ones. There was no necessity
for the two to go together, and the
British social anthro]x>logists remained
relatively untheoretical. Not so the American
sociologists, particularly Talcott Parsons,
who claimed that they were not only
theorists but something called general theorists,
and strongly emphasized the importance
of theory.
Theirs was to be, moreover, a certain
kind of theory. They were students of Durkheim
and took seriously his famous definition
of social facts: "Since their essential
characteristic consists in the power they
possess of exerting, from outside, a pressure
on individual consciousnesses, they do not
derive from individual consciousnesses, and
in consequence sociology is not a corollary
of psychology." - Since Durkheim was a
great man, one can find statements in his
writings that have quite other implications,
but this caricature of himself was the one
that made the difference. If not in what
they said, then surely in what they did,
the functionalists took Durkheim seriously.
Their fundamental unit, the role, was a
social fact in Durkheim's sense. And their
theoretical program assumed, as he did,
that sociology should be an independent
science, in the sense that its propositions
should not be derivable from some other
social science, such as psychology. This
meant, in effect, that the general propositions
of sociology were not to be propositions
about the behavior of "individual
consciousnesses"—or, as I should say, about
men—but propositions about the characteristics
of societies or other social groups as
such.
1 George P. Murdock, Social Structure, New
York: Macmillan, 1949.
*£mile Durkheim, Lfs rigUs de la mitkode
sodologiqut (8th ed.), Paris: Alcan, 1927, pp. 124-
12S.
BRINGING MEN BACK IN 811
Where functionalism failed was not in its
empirical interests but, curiously, in what it
most prided itself on, its general theory.
Let me be very careful here. In a recent
Presidential Address, Kingsley Davis asserted
that we are all functionalists now,*
and there is a sense in which he was quite
right. But note that he was talking about
functional analysis. One carries out functional
analysis when, starting from the existence
of a particular institution, one tries
to find out what difference the institution
makes to the other aspects of social structure.
That is, one carries out the empirical
program of functionalism. Since we have
all learned to carry out functional analyses,
we are in this sense all functionalists now.
But functional analysis, as a method, is
not the same thing as functional theory.
And if we are all functional analysts, we
are certainly not all functional theorists.
Count me out, for one.
The only inescapable office of theory^
is to explain. The theory of evolution is ^
an explanation why and how evolution occurs.
To look for the consequences of institutions,
to show the interrelationships
of institutions is not the same thing as explaining
why the interrelationships are what
they are. The question is a practical and
not a philosophical one—not whether it
is legitimate to take the role as the fundamental
unit, nor whether institutions are
really real, but whether the theoretical
program of functionalism has in fact led
to explanations of social phenomena, including
the findings of functional analysis
itself. Nor is the question whether functionalism
might not do so, but whether it has
done so as of today. I think it has not.
THE NATURE OF THEORY
With all their talk about theory, the
functionalists never—and I speak advisedly—succeeded
in making clear what
a theory was. It must be allowed in their
excuse that, in the early days, the philosophers
of science had not given as clear an
answer to the question as they have now.*
But even then, the functionalists could
have done better than they did, and certainly
the excuse is valid no longer. Today
we should stop talking to our students about
sociological theory until we have taught
them what a theory is.
A theory of a phenomenon consists of \
a series of propositions, each stating a relationship
between properties of nature.
But not every kind of sentence qualifies as
such a proposition. The propositions do
not consist of definitions of the properties:
the construction of a conceptual scheme
is an indispensable part of theoretical work
but is not itself theor>'. Nor may a proposition
simply say that there is some relationship
between the prof>erties. Instead, if
there is some change in one of the properties,
it must at least begin to specify what
the change in the other prop>erty will be.
If one of the properties is absent, the other
will also be absent; or if one of the propjerUes
increases in value, the other will too.
The prop)erties, the variables, may be prob-
abilities.
Accordingly, to take a famous example,
Marx's statement that the economic organization
of a society determines the nature
of its other institutions is an immensely
useful guide to research. For it says: "Look
for the social consequences of economic
change, and if you look, you will surely find
them!" But it is not the sort of proposition
that can enter a theory. For by itself it says
only that, if the economic infrastructure
changes, there will be some change in the
social superstructure, without beginning to
suggest what the latter change will be. Most
of the sentences of sociology, alleged to be
theoretical, resemble this one of Marx's,
yet few of our theorists realize it. And while
we are always asking that theory guide research,
we forget that many statements like
Marx's are good guides to research without
being good theory.
To constitute a theory, the propositions
must take the form of a deductive system.
One of them, usually called the lowest-order
proposition, is the proposition to be ex-
\
» "The M>'th of Functional Analysis as a Special
Method in Sociology and Anthropology," American
Sociological Review, 24 (December, 19S9), pp. 757-
773.
•See especially R. B. Braithwaite, Scientific
Explanation, Cambridge: Cambridge University
Press, 1953.
812 AMERICAN SOCIOLOGICAL REVIEW
\
plained, for example, the proposition that
the more thoroughly a society is industrialized,
the more fully its kinship organization
tends towards the nuclear family. The other
propositions are either general propositions
or statements of particular given conditions.
The general propositions are so called because
they enter into other, perhaps many
other, deductive systems besides the one
in question. Indeed, what we often call a
theory is a duster of deductive systems,
sharing the same general propositions but
having different expUcanda. The crucial
requirement is that each system shall be
deductive. That is, the lowest-order proposition
follows as a logical conclusion from
the general propositions under the specified
given conditions. The reason why statements
like Marx's may not enter theories is that
no definite conclusions may in logic be
drawn from them. When the lowest-order
proposition does follow logically, it is said
to be explained. The explanation of a phenomenon
is the theory of the phenomenon.
A theory is nothing—it is not a theory—
imless it is an explanation.
One may define properties and categories,
and one still has no theorj'. One may state
that there are relations between the properties,
and one still has no theory. One may
state that a change in one property will
produce a definite change in another prop>erty,
and one still has no theory. Not until
one has properties, and propositions stating
the relations between them, and the
propositions form a deductive system—
not imtil one has all three does one have a
theory. Most of our arguments about theory
would fall to the ground, if we first asked
whether we had a theory to argue about.
FUNCTIONAL THEORIES
As a theoretical effort, functionalism never
came near meeting these conditions. Even
if the functionalists had seriously tried to
meet them, which they did not, I think
they would still have failed. The difficulty
lay in the characteristic general propositions
of functionalism. A proposition is not functional
just because it uses the word function.
To say that a certain institution is functional
for individual men in the sense of meeting
their needs is not a characteristic proposition
of functionalism. Instead it belongs
to the class of psychological propositions.
Nor is the statement that one institution
i.s a function of another, in the qua.si-mathematical
sense of function, characteristic.
Though many functional theorists make
such statements, non-functionalists like myself
may also make them without a qualm.
The characteristic general propositions of
functional theory in sociology take the form:
"If it is to survive, or remain in equilibrium,
a social system—any social system—must
possess institutions of T\TeX." For instance,
if it is to survive or remain in equilibrium,
a society must possess conflict-resolving
institutions. By general propositions of this
sort the functionalists sought to meet Durkheim's
demand for a truly independent
sociological theory.
The problem was, and is, to construct
deductive systems headed by such propositions.
Take first the terms equilibrium and
survival. If the theorist chose equilibrium,
he was able to provide no criterion of social
equilibrium, especially "dynamic" or "moving"
equilibrium, definite enough to allow
anything specific to be deduced in logic
from a proposition employing the term. I
shall give an example later. When indeed
was a society not in equilibrium? If the
theorist chose survival, he found this, too,
surprisingly hard to define. Did Scotland,
for instance, survive as a society? Though
it had long been united with England, it
still possessed distinctive institutions, legal
and religious. If the theorist took survival
in the strong sense, and said that a society
had not surN'ived if all its members had
died without issue, he was still in trouble.
As far as the records went, the very few
societies of this sort had possessed institution
of all the types the functionalists said
were necessary for survival. The evidence
put in question, to say the least, the empirical
truth of the functionalist propositions.
Of course the functionalists were at liberty
to say: "If a society is to survive, its members
must not all be shot dead," which was
true as true could be but allowed little to
be deduced about the social characteristics
of surviving societies.
Indeed the same was true of the other
functional propositions. Even if a statement
like: "If it is to survive, a society must
BRINGING MEN BACK IN 813
possess conflict-resolving institutions," were
accepted as testable and true, it possessed
little explanatory power. From the proposition
the fact could be deduced that, given
a certain society did survive, it did possess
conflict-resolving institutions of some kind,
and the fact was thus explained. What
remained unexplained was why the society
bad conflict-resolving institutions of a particular
kind, why, for instance, the jury was
an ancient feature of Anglo-Saxon legal
institutions. I take it that what sociology
has to explain are the actual features of
•actual societies and not just the generalized
(features of a generalized society.
I do not think that members of the functional
school could have set up, starting
with general propositions of their distinctive
type, theories that were also deductive
systems. More important, they did not.
Recognizing, perhaps, that they were blocked
in one direction, some of them elaborated
what they called theory in another. They
used what they asserted were a limited and
exhaustive number of functional problems
faced by any society to generate a complex
set of categories in terms of which
social structure could be analyzed. That is,
I they set up a conceptual scheme. But anal'ysis
is not explanation, and a conceptual
scheme is not a theory. They did not fail
to make statements about the relations between
the categories, but most of the statements
resembled the one of Marx's I cited
earlier: they were not of the tjpe that
enter deductive systems. From their lowerorder
propositions, as from their higherorder
ones, no definite conclusions in logic
could be drawn. Under these conditions,
there was no way of telling whether their
choice of functiorial problems and categories
was not wholly arbitrary. What the functionalists
actually produced was not a theorj'
but a new language for describing social
structure, one among many possible languages;
and much of the work they called
theoretical consisted in showing how the
words in other languages, including that
of everyday life, could be translated into
theirs. They would say, for instance, that
what other people called making a living
was called in their language goal-attainment.
But what makes a theory is deduction, not
translation.
I have said that the question is not whether,
in general, functional theories can be
real theories, for there are sciences that
possess real functional theories. The question
is rather whether this particular effort
was successful. If a theory is an explanation,
the functionalists in sociology were,
on the evidence, not successful. Perhaps
they could not have been successful; at
any rate they were not. The trouble vdth «
their theory was not that it was wrong, I
but that it was not a theory.
AN ALTERNATIVE THEORY
Here endeth the destructive part of the
lesson. I shall now try to show that a more
successful effort to explain social phenomena
entails the construction of theories different
from fimctional ones, in the sense that their
general propositions are of a different kind,
precisely the kind, indeed, that the functionalists
tried to get away from. I shall
try to show this for the verj' phenomena •
the functionalists took for granted and the \
very relations they discovered empirically. ^
I shall even try to show that, when functionalists
took the job of explanation seriously,
which they sometimes did, this other
kind of theory' would appear unacknowledged
in their own work.
The functionalists insisted over and over
again that the minimum unit of social I
analysis was the role, which is a cluster of ^
norms. In a recent article, James Coleman
has written: " . . . sociologists have characteristically
taken as their starting-point a
social system in which norms exist, and
individuals are largely governed by these
norms. Such a strategy views norms as
the governors of social behavior, and thus
neatly bypasses the difficult problem that
Hobbes posed." ^ Hobbes' problem is, of
course, why there is not a war of all against
all.
Why, in short, should there be norms
at aJl? The answer Coleman gives is that,
in the kind of case he considers, norms
arise through the actions of men rationally
calculating to further their own self-interest
in a context of other men acting in the
' James S. Coleman, "Collective Decisions," Sociological
Inquiry, 34 (1964), pp. 166-181.
814 AMERICAN SOCIOLOGICAL REVIEW
same way. He writes: "The central postulate
about behavior is this: each actor will attempt
to extend his power over those actions
in which he has most interest." Starting
from this postulate, Coleman constructs
a deductive system explaining why the
actors adopt a particular sort of norm in
the given circumstances.
I do not want to argue the vexed question
of rationality. I do want to point out what
sort of general proposition Coleman starts
with. As he recognizes, it is much like the
central assumption of economics, though
self-interest is not limited to the material
interests usually considered by economists.
It also resembles a proposition of psychology,
though here it might take the form:
the more valuable the reward of an acti\'ity,
the more likely a man is to perform the
activity. But it certainly is not a characteristic
functional proposition in sociology:
it is not a statement about the conditions
of equilibrium for a society, but a statement
about the behavior of individual men.
Again, if there are norms, why do men
conform to them? Let us lay aside the fact
I that many men do not conform or conform
very indifferently, and assume that they
all do so. Whj' do they do so? So far as
the functionalists gave any answer to the
question, it was that men have "internalized"
the values embodied in the norm.
But "intemalization" is a word and not
an explanation. So far as their own theory
was concerned, the functionalists took conformity
to norms for granted. They made
the mistake Malinowski pointed out long
ago in a book now too little read by sociologists,
the mistake made by early writers
on primitive societies, the mistake of assuming
that conformity to norms is a matter
of ". . . this automatic acquiescence,
this instinctive submission of every member
of the tribe to its laws. . . ." " The alternative
answer Malinowski gave was that
obedience to norms "is usually rewarded
according to the measure of its perfection,
while noncompliance is visited upon the
remiss agent." ^ In short, the answer he
* Bronislaw MaHnowski, Crime and Custom m
Savage Society, Patenon, N.J.: Littlefield, Adams,
19S9, p. 11.
' Ibid., p. 12.
gave is much like that of Coleman and
the psychologists. Later he added the suggestive
remark: "The true problem is not
to study how human life submits to rules—
it simply does not; the real problem is how
the rules become adapted to life.""
The question remains why members of
a particular society find certain of the results
of their actions rewarding and not
others, especially when some of the results
seem far from "naturcilly" rewarding. This
is the real problem of the "intemalization"
of values. The explanation Ls given not by
any distinctively sociological propositions
but by the propositions of learning theory
in psychology.
The functionalists were much interested
in the interrelations of institutions, and it
was one of the glories of the school to have
pointed out many such interrelations. But
the job of a science does not end with
pointing out interrelations; it must try to
explain why they are what they are. Take
the statement that the kinship organization
of industrialized societies tends to be
that of the nuclear family. 1 cannot give
anything like the full explanation, but I
can, and you can too, suggest the beginning
of one. Some men organized factories be^
cause by so doing they thought they could
get greater material rewards than they could
get otherwise. Other men entered factories
for reasons of the same sort. In so doing
they worked away from home and so had
to forgo, if only for lack of time, the cultivation
of the extended kinship ties that
were a source of reward, because a source
of help, in many traditional argricultural
societies, where work lay closer to home.
Accordingly the nuclear family tended to
become associated with factory organization;
and the explanation for the association
is provided by propositions about the
behavior of men as such. Not the needs of
society explain the relationship, but the
needs of men.
Again, functionalists were interested in
the consequences of institutions, especially
their consequences for a social system as
a whole. For instance, they were endlessly
concerned with the functions and dysfunctions
of status systems. Seldom did they
* Ibid., p. 127.
BRINGING MEN BACK IN 815
ask why there should be status systems in
the first place. Some theorists have taken
the emergence of phenomena like status '
systems as evidence for Durkheim's contention
that sociology was not reducible tof
psychology. What is important is not the
fact of emergence but the question how the-^
emergence is to be explained. One of the
accomplishments of small-group research
is to explain how a status system, of course
on a small scale, emerges in the course of
interaction between the members of a
group.* The explanation is provided by^
psychological propositions. Certainly no
functional propositions are needed. Indeed^
the theoretical contribution of small-group
research has consisted "in showing how the
kinds of microscopic variables usually ig-.
nored by sociologists can explain the kinds I
of social situations usually ignored by psy- \
chologists." ^^
What is the lesson of all this? If the very
things functionalists take for granted, like
norms, if the very interrelationships they
empirically discover can be explained by
deductive systems that employ psychological
propositions, then it must be that the
general explanatory principles even of soci-|'
ology are not sociological, as the function-|
alists would have them be, but psychological,
propositions about the behavior of men,
not about the behavior of societies. On the
analogy with other sciences, this argument
by itself would not undermine the validity
of a functional theory. Thermodynamics,
for instance, states propositions about aggregates,
which are themselves true and
genera], even though they can be explained
in turn, in statistical mechanics, by propositions
about members of the aggregates.
The question is whether this kind of situation
actually obtains in sociology. So far
as functional propositions are concerned,
which are propositions about social aggregates,
the situation does not obtain, for
they have not been shown to be true and
general.
EXPLAINING SOCIAL CHANGE
' See GeorRe C. Homans, Social Behavior: Its
Elementary Forms, New York: Harcourt, Brace &
World, 1961, esp. Ch. 8.
>" C. N. Alexander, Jr. and R. L. Simpson, "Balance
Theory and Distributive Justice," Sociological
Inquiry 34 (1964), pp. 182-192.
My next contention is that even confessed
functionalists, when they seriously
try to explain certain kinds of social phenomena,
in fact use non-functional explanaI
tions without recognizing that they do so.
This is particularly clear in their studies
of social change.
Social change provides a .searching test
for theory, since historical records are a
prerequisite for its study. Without history,
the social scientist can establish the contemporaneous
interrelations of institutions,
but may be hard put to it to explain why
the interrelations should be what they are.
With historical records he may have the
information needed to support an explanation.
One of the commonest charges against
the functionalist school was that it could
not deal with social change, that its analysis
was static. In recent years some functionalists
have undertaken to show that the
charge was unjustified. They have chosen
for their demonstration the process of differentiation
in society, the process, for instance,
of the increasing specialization of occupations.
In question as usual is not the fact
of differentiation—there is no doubt that
the over-all trend of social history has been
in this direction—but how the process is
to be explained.
A particularly good example of this new
development in functionalism is Neil Smelser's
book. Social Change in the Industrial
Revolution: An Application of Theory to
the British Cotton Industry 1770-1840.^^
The book is not just good for my purposes:
it is good, very good, in itself. It
provides an enormous amount of well organized
information, and it goes far to explain
the changes that occurred. The amusing
thing about it is that the explanation Smelser
actually uses, good scientist that he is,
to account for the changes is not the functional
theory he starts out with, which
is as usual a non-theory, but a different
kind of theory and a better one.
Smelser begins like any true functionalist.
For him a social system is one kind
of system of action, characterized as follows:
"A social system . . . is composed
"Chicago: University of Chicago Press, 1959.
816 AMERICAN SOCIOLOGICAL REVIEW
of a set of interrelated roles, collectivities,
etc. . . . It is important to remember
that the roles, collectivities, etc., not individuals,
are the units in this last case."
Moreover, "all systems of action are governed
by the principle of equilibrium. According
to the dominant tJTie of equilibrium,
the adjustments proceed in a certain direction:
if the equilibrium is stable, the units
tend to return to their original position;
if the equilibrium is partial, only some of
the units need to adjust; if the equilibrium
is unstable, the tendency is to change,
through mutual adjustment, to a new equilibrium
or to disintegrate altogether." Finally,
"all social systems are subject to
four functional exigencies which must be
met more or less satisfactorily if the system
is to remain in equilibrium." " Note that
by this argument all social s>^stems are in
equflibrium, even systems in process of
disintegration. Though the latter are in
unstable equilibrium, they are still in equilibrium.
Accordingly they are meeting more
or less satisfactorily the four functional
exigencies. You see how useful a deductive
system can be in soda] science? More seriously
you will see that definitions of equilibrium
are so broad that you may draw
any conclusion you like from them.
But for all the explanatory use Smelser
makes of it, this theory and its subsequent
elaboration is so much window-dressing.
When he really gets down to explaining
the innovations in the British cotton textile
industry, especially the introduction of
spinning and weaving machinery, he forgets
his ftmctionalism. The guts of his actual
explanation lie in the seven steps through
which he says the process proceeds:
Industrial difierentiation proceeds, therefore,
by the following steps:
(1) Dissatisfaction with the productive
achievements of the industr>' or its relevant
sub-sectors and a sense of opportunity in
terms of the potential availability of adequate
facilities to reach a higher level of produc-
tivity.
(2) Appropriate symptoms of disturbance
in the form of "unjustified" negative emotional
reactions and "unrealistic" aspirations
on the part of various elements of the popw-
lation.'"
I shall not give the other five steps, as
I should make the same criticism of them
as I now make of the first two. I think
they provide by implication a good explanation
of the innovations of the Industrial
Revolution in cotton manufacturing. But
what kind of an explanation is it? Whatever
it is, it is not a functional one. Where
here do roles appear as the fundamental
units of a social system? Where are the four
functional exigencies? Not a word do we
hear of them. Instead, what do we hear of?
We hear of dissatisfaction, a sense of opportunity,
emotional reactions, and aspirations.
And what feels these things? Is a
role dissatisfied or emotional? No; Smelser
himself says it is "various elements of the
population" that do so. Under relentless
pressure let us finally confess that "variotis
elements of the population" means men.
And what men? For the most part men
engaged in making and selling cotton cloth.
And what were they dissatisfied with? Not
with "the productive achievements of the
industry." Though some statesmen were
certainly concerned about the contribution
made by the industry as a whole to the
wealth of Great Britain, let us, again under
relentless pressure, confess that most of the
men in question were concerned with their
own profits. Let us get men back in, and
let us put some blood in them. Smelser
himself makes the crucial statement: "In
Lancashire in the early 176O's there was
excited speculation about instantaneous
forttmes for the man lucky enough to stumble
on the right invention."" In short,
the men in question were activated by
self-interest. Yet not all self-interests are
selfish interests, and certainly not all the
innovations of the Industrial Revolution can
be attributed to selfishness.
Smelser's actual explanation of technical
innovation in cotton manufacturing might
be sketched in the following deductive system.
I have left out the most obvious steps.
1. Men are more likely to perform an
activity, the more valuable they perceive
the reward of that activity to be.
2. Men are more likely to perform an
activity, the more successful they per"
Ibid., pp. 10-11.
" Ilrid., p. 29. " Ibid., p. 80.
BRINGING MEN BACK IN
ceive the activity is likely to be in getting
that reward.
3. The high demand for cotton textiles
and the low productivity of labor led
men concerned with cotton manufacturing
to perceive the development of
labor-saving machinery as rewarding in
increased profits.
4. The existing state of technology led
them to perceive the effort to develop
labor-saving machinery as likely to be
successful.
5. Therefore, by both (1) and (2) such
men were highly likely to try to develop
labor-saving machinery.
6. Since their perceptions of the technology
were accurate, tiieir efforts were likely
to meet with success, and some of them
did meet with success.
From these first steps, others such as the
organization of factories and an increasing
specialization of jobs followed. But no different
kind of explanation is needed for
these further developments: propo.sitions
like (1) and (2), which I call the value
and the success propositions, would occur
in them too. We should need a further
proposition to describe the effect of frustration,
which certainly attended some of the
efforts at innovation, in creating the "negative
emotional reactions" of Smelser's step
2.
I must insist again on the kind of explanaI
tion this is. It is an explanation using psyI
chological propositions (1 and 2 above),
' psychological in that they are commonly
stated and tested by psychologists and that
they refer to the behavior of men and not
to the conditions of equilibrium of societies
or other social groups as such. They are
general in that they appear in many, and
I think in all, of the deductive systems that
will even begin to explain social behavior.
There is no assumption that the men in
question are all alike in their concrete behavior.
They may well have been conditioned
to find different things rewarding,
but the way conditioning takes place is
itself explained by psychological propositions.
There is no assumption that their
values are all materialistic, but only that
their pursuit of non-material values follows
the same laws as their pursuit of material
817
ones. There is no assumption that they are
isolated or unsocial, but only that the laws
of human behavior do not change just because
another person rather than the physical
environment provides the rewards for
behavior. Nor is there any assimiption
that psychological propositions will explain
everything social. We shall certainly not
be able to explain everything, but our failures
will be attributable to lack of factual
information or the intellectual machinery
for dealing with complexity—though the
computers will help us here—and not to
the propositions themselves. Nor is there
any assumption here of psychological reductionism,
though I used to think there
was. For reduction implies that there are
general sociological profwsitions that can
then be reduced to psychological ones. I
now suspect that there are no general sociological
propositions, propositions that hold
good of all societies or social groups as i
such, and that the only general proposi- \
tions of sociology are in fact psychological.
What I do claim is that, no matter what
we say our theories are, when we seriously
try to explain social phenomena by constructing
even the veriest sketches of deductive
systems, we find ourselves in fcict,
and whether we admit it or not, using what
I have called psychological explanations.
I need hardly add that our actual explanations
are our actual theories.
I am being a little unfair to functionalists
like Smelser and Parsons if I imply
that they did not realize there were people
around. The so-called theory of action made
a very good start indeed by taking as its
paradigm for social behavior two persons,
the actions of each of whom sanctioned,
that is, rewarded or punished, the actions
of the other.** But as soon as the start
was made, its authors disregarded it. As the
theory of action was applied to society,
it appeared to have no actors and mighty
little action. The reason was that it separated
the personality system from the social
system and proposed to deal with the latter
alone. It was the personality system
\
Talcott Parsons and Edward Shils (eds.),
Toward a General Theory of Action, Cambridge,
M Harv'ard University Press, 1951, pp. 14-16.Mass.:
818 AMERICAN SOCIOLOGICAL REVIEW
that had "needs, drives, skills, etc." >• It
was not part of the sodal system, but only
conducted exchanges with it, by providing
it, for instance, with disembodied motivation."
This is the kind of box you get into
when you think of theory as a set of boxes.
For this reason, no one should hold their
style of writing against the functionalists.
The best of writers must write clumsily
when he has set up his intellectual problem
in a clumsy way. If the theorist will only
envisage his problem from the outset as
one of constructing explanatory projKJsitioDs
and not a set of categories, he will
come to see that the personal and the social
are not to be kept separate. The actions
of a man that we take to be evidence of
his personality are not different from his
actions that, together with the actions of
others, make up a sodal system. They are
the same identical actions. The theorist
will realize this when he finds that the same
set of general propositions, induding the
success and the value proposition mentioned
above, are needed for explaining the phenomena
of both personality and sodety.
CONCLUSION
If sodology is a science, it must take
seriously one of the jobs of any science,
which is that of providing explanations for
the empirical relations it discovers. An explanation
is a theory, and it takes the
" Smelser, op. dt., p. 10.
" Ibid., p. 33.
form of a deductive system. With all its
talk about theory, the functionalist school
did not take the job of theory seriously
enough. It did not ask itself what a theory
was, and it never produced a functional
theory that was in fact an explanation.
I am not sure that it could have done so,
starting as it did with propositions about the
conditions of social equilibrium, propositions
from which no definite conclusions could be
drawn in a deductive system. If a serious
effort is made to construct theories that will
even begin to explain social phenomena, it
turns out that their general propositions are
not about the equilibrium of societies but
about the behavior of men. This is true even
of some good functionalists, though they will
not admit it. They keep psychological explanations
under the table and bring them
out furtively like a bottle of whiskey, for
use when they reaUy need help. What I
cisk is that we bring what we say about
theory into line with what we actually do,
and so put an end to our intellectual hypocrisy.
It would unite us with the other
social sciences, whose actual theories are
much like our actual ones, and so strengthen
us all. Let us do so idso for the sake of
our students. I sometimes think that they
begin with more understanding of the real
nature of sodal phenomena than we leave
them with, and that our double-talk kills
their mother-wit. Finally, I must acknowledge
freely that everything I have said
seems to me obvious. But why cannot we
take the obviotis seriously?