Readings in
Philosophy of Psychology
Volume 1
Edited by Ned Block
Harvarp University Press
Cambridge, MassachusettsIntroduction: Behaviorism
Harris Savin
N FIRST READING, the behavior-
ist conclusions of Carl G. Hempel
and of B.F. Skinner, who exemplify the
two main currents of behaviorist thought,
have utterly different bases. Hempel’s
is an epistemological thesis, purporting
to show that, insofar as psychological
sentences are intelligible, they describe
“physical behavior.” For Hempel, that is
to say, there is nothing psychology could
be about except behavior. Skinner, on the
contrary, has not an epistemological the-
sis but advice about the most expedient
way to do psychology. He assumes with-
out argument that the only interesting
objective for psychology is the prediction
of behavior, and he argues that it is fruit-
less, if that is one’s objective, to consider
mental states or mental processes: “The
objection to inner states is not that they
do not exist but that they are not relevant
to a functional analysis.”
Note that neither position entails the
other. On the one hand, Skinner, unlike
Hempel, is open to the possibility that
there are psychological sentences that are
not translatable into physical language.
He insists only that he does not need them
for his science of behavior, Hempel, on
the other hand, has no quarrel with such
psychological notions as purposes and
beliefs, nor does he deny that there may
be psychological laws whose most illumi-
nating formulation will refer to such psy-
chological entities. His only claim is that
it is always possible in principle, even if
far too tedious for the everyday purposes
of scientific psychology, to translate such
laws into physical language. In short,
Hempel's thesis about the ultimate foun-
dations of psychology has no obvious
bearing on the day-to-day activity of psy-
chologists, while Skinner's advice about
how psychologists can best spend their
time has no obvious epistemological con-
tent.
Hempel is most vulnerable to attack
when he claims that “the meaning of a
proposition is established by its condi-
tions of verification.” “Conditions of veri-
fication,” he supposes, must be physical
conditions; any psychological sentence,
therefore, must have the same meaning as
some physical sentence. In reply, one
might try (as Hilary Putnam does in chap-
ter 2) to show that, however closely men-
tal events and behavioral events may, as a
matter of empirical fact, be correlated
with one another, mind statements can-
not, in general, be translated by behavior
statements.
There are two obvious ways to chal-R Harris Savin
lenge Skinner: by questioning his assump-
tion that the only interesting goal for psy-
chology is the prediction of behavior, and
by showing that, even for the prediction
of behavior, it is expedient on occasion to
consider mental states. Skinner notwith-
standing, it surely needs no argument to
establish that some of us look to psychol-
ogy for descriptions of mental states and
for knowledge about how to modify those
states. As for Skinner's lack of interest in
such topics, it is a mere idiosyncrasy,
warranting no discussion here. More in-
teresting is the question whether or not
consideration of mental states may help in
the prediction of behavior. Skinner says
not, and gives the following argument.
Either we know what a person's mental
state is by virtue of what we know of
his history of stimulation and behavior
(knowing he is hungry, for example, be-
cause we know he has not eaten today),
or else we do not have such evidence for
his mental state. In the former case, we
may as well forget about the mental state
and relate subsequent behavior directly to
history (predicting, for example, that the
subject of our experiment will eat on the
basis not of his inferred hunger but of his
observed history of deprivation. If, on the
contrary, we have not inferred our sub-
ject’s mental state from his prior history,
then (ignoring such far-fetched possibili-
ties as telepathy) we do not know what
his mental state is and therefore can make
No predictive use of any putative law re-
lating behavior to mental states,
The latter half of this argument is
sound enough, but Skinner's dismissal of
mental states that are inferred from a sub-
ject's history is unconvincing. Consider,
for amoment, the variety of mental states
that, on different occasions, might figure
in explanations of eating: we may eat be-
cause we are hungry, because we do not
want to offend our hostess, because the
doctor has ordered us to eat, because we
think it imprudent to drink on an empty
stomach, or for any number of other rea-
sons. Such distinctions among states of
mind are obviously crucial for predicting
subsequent behavior—for predicting, for
example, whether we will stop eating
when the hostess leaves the room, when
the doctor pronounces us cured, or when
we learn there is no whiskey in the house.
Skinner would urge that we can nonethe-
less dispense with these motives for eating
and relate the differences in behavior di-
rectly to differences in previous history.
It takes only a moment's reflection to
see that such a program for psychology is
only plausible to the extent that discover-
able principles relate past history to future
behavior. There are, obviously, far too
many significantly different human histo-
ries for anyone to propose that a Practica-
ble psychological theory could consist of
an exhaustive enumeration of histories,
each one paired with a prediction of
subsequent behavior. Skinner, of course,
recognizes this need for principles and,
indeed, claims to have discovered the
necessary principles in his work on oper-
ant conditioning. How widely these prin-
ciples of operant conditioning apply out-
side the laboratory remains a matter of
controversy among psychologists. It suf-
fices for the purposes of the present dis-
cussion to note that Skinner believes, as
does everyone else, that we need princi-
ples if we want to predict behavior from
past history, yet his argument omits to
consider the possibility that mental states
will occur as theoretical constructs in
some such successful system of principles.
Notwithstanding the differences be.
tween Skinner's thesis and Hempel's, their
two behaviorisms evidence a shared con-
viction that people are in no important
respect different from inanimate aggre-
gates of matter. Now, what is important
depends on one's interests and one's point
of view. From certain limited but perfect-
ly valid points of view, a man may use-
fully be regarded as a mass of matter con-
centrated at a single point. For other pur-
Poses, his description may consist of whatPart One Introduction 13
he does in a factory, an army, or an art-
ist’s studio. I do not wish to quarrel with
the decision to study man from any of
these special points of view, nor with any-
‘one who chooses to see to what extent it is
possible to carry out Hempel’s program,
‘or Skinner's. But it is a remarkable and
distressing fact that, for much of the pres-
ent century, the scientific study of man
has been so largely in the hands of those
who wished to study him from one or the
‘other of these two points of view and that
the human sciences have taken so little
serious interest in man as artist, as lover,
as moral or political theorist, or, indeed,
as creator of scientific knowledge. Even
more remarkable and more distressing,
most of those who have developed the
behaviorist point of view in the human
sciences have believed that their particular
point of view was not the result of a large-
ly arbitrary choice. On the contrary, they
have supposed, for reasons such as those
discussed here, that behaviorism was the
only possible basis for the scientific study
of man.
‘One need not impugn the motives of
the founders of behaviorism to note the
compatibility between their view of the
proper study of man and the interests of
those—generals, politicians, advertisers,
and so forth—whose business it is to pre-
dict or control the behavior of masses of
anonymous humanity. Of late, behavior-
ist fervor has greatly diminished, owing in
part perhaps to growing disenchantment
with these nonacademic purposes and in
part also to the examples of Jean Piaget
and Noam Chomsky, who, in their rather
different ways, have made striking ad-
vances in the scientific study of the nature
‘of human knowledge—advances that,
moreover, cannot plausibly be construed
as advances in the study of human be-
havior.