III. THE DIFFERENT TYPES
OF BEHAVIORISM
It has been argued that behaviorism was basically
a methodological movement
in psychology which laid
down restrictions on the data on which a science
could
be properly based. In the case of its founder, J. B.
Watson,
this central doctrine was supported by an
inductivist view of scientific
method flanked by the
metaphysical doctrine of materialism and by the asso-
ciationism of a peripheralist, or
S-R, type as a psycho-
logical theory. Few
of the later behaviorists shared all
these assumptions. In commenting,
therefore, on the
main features of their theories special attention
will
be paid to their adherence to or departure from these
other
tenets of Watson which have come to be loosely
associated with behaviorism.
1. Early Materialists.
Some of Watson's more im-
mediate
contemporaries had more in common with his
bold, metaphysical brashness,
than his later followers.
Albert P. Weiss, for instance, published a book entitled
A Theoretical Basis of Human Behavior in 1925 in
which he banned consciousness and introspection from
psychology and claimed
that all behavior could be
interpreted in terms of physiochemical
processes. Nev-
ertheless he argued that
what is distinctive of a human
being is that his environment is social.
Psychology is
therefore a bio-social science which is particularly
concerned
with the impact of the social environment
on a biological organism. Weiss
was particularly inter-
ested in child
development and learning, but he never
seriously tackled the conceptual
problems facing his
reductionism, of how features of the social environ-
ment, such as commands, promises, and
moral exhorta-
tion can be analyzed in
purely physical terms.
Another early behaviorist, who showed equal naiveté
about the
environment, was W. S. Hunter. He held
that consciousness or experience for
the psychologist
is merely a name which he applies to what other
people call “the environment.” This suggestion epito-
mizes the epistemological innocence of
most of the
early behaviorists, against which Koffka reacted so
strongly. Hunter, however, distinguished himself in
other ways. He thought
that the new look in psychol-
ogy deserved a
new name and attempted, without any
success, to substitute
“anthroponomy” for “psychol-
ogy.” He also was the first to use the
temporal maze
for the study of motor learning.
A more ingenious and interesting theorist of early
behaviorism was E. B.
Holt. He was one of the first
to try to deal with Freudian phenomena within
a
behavioristic framework and his The Freudian Wish
and its Place in Ethics (1915) is a classic in this tough-
minded tradition, which was later to
include O. H.
Mowrer, J. Dollard, and N. E. Miller. Holt also devel-
oped Watson's idea that thinking is
subvocal talking
and theorized about the connections between language
and conditioning. He thus anticipated later much more
ambitious, if
abortive, attempts to exhibit language as
a system of conditioned
responses.
Karl S. Lashley was a pupil of Watson's who made
distinguished contributions
to the physiology of the
nervous system. This, however, did not prevent
him
from making pronouncements about the subject matter
of psychology
and about its methods, for instance, that
introspection is “an
example of the pathology of scien-
tific
method” (Lashley, 1923). His physiological find-
ings, however, as expressed in his Brain Mechanisms
and Intelligence (1929), did not support other
doctrines
of Watson. His postulates of equipotentiality—that
one
part of the cortex is potentially the same as another
in its
capacity for learning, and of mass action—that
learning is a
function of the total mass of tissue, favored
a centralist theory of
learning rather than Watson's
peripheralist theory. He became very critical
of S-R
theories which postulated a simple connection between
stimulus
and response and which ignored the role of
intervening cortical processes.
The simple switching
function accorded by Watson to the brain, which
he
inherited from Descartes, was denied.
Lashley, however, never departed from Watson's
materialistic standpoint. Like Weiss and Hunter he
believed
that ultimately behavior could be describable
in the concepts of mechanics
and chemistry. It is also
significant that all these early behaviorists
shared an
inductivist view of scientific method. They thought of
the
different sciences as having different subject mat-
ters and as consisting of generalizations about them
derived
from reliable data drawn from these subject
matters. In this respect they
shared not only Watson's
methodological recipes but also the view of
scientific
method from which these recipes arose (Peters, 1951).
2. E. C. Tolman.
One of the most influential and
forceful converts to behaviorism was
E. C. Tolman;
for he was calling himself a “purposive
behaviorist”
as early as 1920, though his definitive work entitled
Purposive Behavior in Animals and Man did not appear
until 1932. He aligned himself with the behaviorists
because he accepted
their central methodological doc-
trine about
the sort of evidence on which a scientific
psychology should be based. He
did not indulge, like
Watson and Weiss, in metaphysical assertions about
the
sorts of entities which there are in the world; he ad-
mitted that, at a common-sense level, men introspect
and manage well enough with mentalistic terms. What
he doubted, however,
was the adequacy of this termi-
nology for
scientific purposes. “Raw feels” are scien-
tifically useless, and mentalistic
terms can be translated
into the language of observable behavior. Tolman,
in
other words, was a conceptual behaviorist rather than
a
materialist, as well as being a behaviorist in his
explicitly stated
methodology.
In the conceptual sphere Tolman made at least three
contributions, two of
which were of permanent impor-
tance. Firstly
he called himself a purposive behaviorist
because he maintained that the
concept of purpose was
irreducible. As has been mentioned (sec. I, 2), he dis-
tinguished between the molecular and the molar level
of behavior, whose unity as segments of
behavior is
provided by the ends towards which movements persist
and
in the attainment of which they are docile. He
accused Watson of not
distinguishing clearly between
the molecular and the molar levels of
analysis and
maintained, against Hull, that behavior at the molar
level is an “emergent” which has descriptive and de-
fining properties of its own. Descriptions of
it cannot
be reduced to or deduced from analyses at the molecu-
lar level.
Secondly Tolman made rather bizarre attempts to
translate mentalistic terms,
which had application at
the molar level, into a behavioristic type of
termi-
nology.
“Consciousness” became “the performance of
a 'sampling' or 'running-back-and-forth' behavior.” He
even
suggested that Freudian personality mechanisms
can be translated into this
type of terminology.
Thirdly, Tolman introduced into psychological the-
ory the notion of intervening variables. Terms like
“instinct” had previously been used, e.g., by
McDougall, not simply to postulate that certain pur-
posive behavior patterns were unlearned; they also had
a
metaphysical dimension to them—a suggestion of
Aristotelian
entelechies, of dynamic mental atoms ac-
tivating behavior. Tolman argued that it was perfectly
legitimate
for a behaviorist to use a term like “drive”
which
did not denote an unobservable entity, but
which was a shorthand symbol for
stating a correlation
between antecedent conditions, e.g.,
food-deprivation,
and variations in behavior, e.g., eating.
This conceptual clarification helped to set psychol-
ogy free to theorize without fear of metaphysics. It
led on to
the use of hypothetical constructs, which did
commit theorists to
postulates about unobservables
usually of a physiological sort. (For this
distinction see
MacCorquodale and Meehl, 1948.) Tolman thus con-
tributed to ridding psychology of the
inductivist myth,
shared by the early behaviorists, that scientists
must
never go beyond what is observed. In fact, however,
the
postulation of unobservables to explain the ob-
served has been one of the most potent sources of
scientific
advance.
In the details of his psychological theory Tolman was
eclectic. He stressed
the importance of both demand
variables and cognitive variables in
behavior, and at-
tempted to state more
precisely assumptions of the sort
which McDougall had incorporated in his
theory of
instincts, i.e., of innate dispositions to pay attention
to
and behave in specific ways towards objects of a
certain class.
In his account of the demand variables Tolman dis-
tinguished first-order drives, which are linked with
specific
antecedent physiological conditions and con-
sequent states of physiological quiescence (e.g., food-
hunger, sex-hunger) from second-order drives (e.g.,
curiosity, constructiveness) which are not so obviously
linked. This
distinction, which was later to become that
between biological and acquired
drives, was important
in the history of behaviorism. On the cognitive
side
Tolman postulated “means-end readinesses”
for
“means-objects” which are innate but docile
relative
to the success of the organism in attaining its goal.
Also in
his account of “behavior supports” he tried
to escape
the sensory atomism of stimulus-response
psychology. He also developed the
concept of the
“sign-Gestalt expectation” to
incorporate the findings
of Gestalt psychology into his assumptions about
the
organism's perceptual field.
Although Tolman emphasized the importance of
innate appetites and aversions
in behavior he was
equally emphatic on the importance of learning, in
which he stressed the role of cognitive variables. He
argued,
also, that the evidence of latent learning was
inconsistent with
Throndike's law of effect. In trial and
error learning a refinement of
sign-Gestalts takes place.
A kind of cognitive map develops of the
different
possibilities as the various alternatives are explored.
Motivational variables are, of course, important in
learning in that they
determine which aspects of a
situation will be emphasized. But learning
depends
primarily on the expectancy of achievement and on
confirmations of the expectancy. In learning animals
and men make
predictions and the maps which they
use to do this are refined more and
more as experience
confirms or falsifies them. As Tolman developed his
theory he became more and more interested in and
convinced of the
importance of cognitive variables. It
is therefore understandable that
behaviorists became
increasingly embarrassed by Tolman's claim that he
was one of them.
3. C. L. Hull.
Behaviorism was basically old philos-
ophy masquerading as a new scientific theory. In the
1930's
philosophers began to be extremely critical of
the old inductivist view of
scientific method, which
most of those in the empirical tradition had
accepted,
though Whewell in the nineteenth century had been
an acute
critic of this view. The role of hypothesis
and deduction in science, which
had been so prominent
in the work of Galileo, was emphasized.
Psychology
began to be influenced by this change of emphasis in
the
philosophical climate. It was suggested, notably by
Kurt Lewin and by Clark
Hull, that psychology was
in a state of disarray, split into warring
factions, be-
cause it had not yet entered its
Galilean phase. Lewin,
a Gestalt psychologist, wrote a detailed
methodological
polemic to this effect in his chapter on
“Aristotelian
and Galilean Modes of Explanation” in
his A Dynamic
Theory of Personality (1935). He
envisaged the use of
the resoluto-compositive method of Galileo to
erect
a field theory in psychology employing postulates taken
from
dynamics.
Clark Hull, unlike other prominent behaviorists, was
not trained in an
animal laboratory. He had established
a reputation for himself as an
ingenious and talented
designer of experiments in concept formation, hypno-
sis, and suggestibility. He next turned to
Pavlov's laws
of conditioning, and Hull's love for mathematics led
him
to set up a hypothetico-deductive model of learn-
ing. He became more and more ambitious and revived
Hobbes's dream of
a mechanical system in which the
laws of human behavior could be deduced
from postu-
lates about “colorless
movements” at the physiological
level. He accepted Tolman's
distinction between mo-
lecular and molar
behaviorism, but differed from
Tolman in thinking that behavior at the
molar level
could ultimately be explained in terms of movements
at the
molecular level. In 1943 he set out his ambitious
program in his
Principles of Behavior, and in 1951 he
published a
revised and more formalized version of his
system in his
Essentials of Behavior.
There was little original in the actual content of
Hull's system save the
appearance of exactitude created
by his technical constructs and
mathematical form of
expression. Hull started from the biological
postulate
of self preservation and maintained that the organism
is in
a state of need when there is a deviation from
optimum conditions for
survival, e.g., lack of food,
water, air. These needs are reduced by
adaptive ac-
tions. The pattern of actions which
lead to a reduction
of a need becomes reinforced—as in
Thorndike's law
of effect. A stimulus which leads to a need-reducing
action may become associated with another stimulus
in accordance with
principles of conditioning, though
Hull believed that there is no
conditioning without
need-reduction.
Hull acknowledged the importance of what Tolman
had called
“intervening variables” in theory con-
struction, and also took over his concept of
“drive.”
He regarded needs as producing primary
animal drives,
which enabled him to correlate observable antecedent
conditions—e.g., of food deprivation with the energy
expended in
behavior, e.g., in eating. He classified
drives on the Darwinian principle
of whether they
tended towards survival of the individual organism or
of the species. Whereas, however, Tolman only postu-
lated such drives in order to explain the activation of
behavior
patterns, Hull postulated them to explain
their acquisition as well, and
their consolidation into
habits. Tolman, as has already been explained,
was
critical of the law of effect. Hull, on the other hand,
tried to
provide a mechanical theory to explain its
operation. He also rejected
Tolman's emphasis on cog-
nitive variables and
claimed that they could be derived
from his fundamental postulate of
stimulus-response
association. Like Watson he was basically a periph-
eralist and an associationist in his
orientation. He
merely attempted to formulate these assumptions more
precisely as part of a mechanical system.
Hull said that his book had been written “on the
assumption that
all behavior, individual and social,
moral and immoral, normal and
psychopathic, is gen-
erated from the same
primary laws; and that the
differences in the objective behavioral
manifestations
are due to the differing conditions under which habits
are set up and function” (Principles of Behavior,
Pref-
ace, p. v). This was programmatic. In fact
his defini-
tions and postulates were not
well rooted in physio-
logical findings,
and precise deductions to the level of
motor behavior were never
made—if indeed they ever
could be made. Unobservables, such as drive-stimuli,
drive-receptors, etc., which were meant to fill in the
mechanical picture
of the workings of needs and drives,
functioned more as hypothetical
constructs relating to
entites whose existence was shadowy and whose inter-
relations were highly obscure. The
main value of his
work was to formulate assumptions about animal
learning at the motor level in a precise enough way
to be refutable. And
most of his assumptions were in
fact refuted, e.g., by Hebb, Young, Harlow,
and others.
His system, however, became popular. Needs and
acquired
drives proliferated which lacked even the
pretence of being anchored to
physiological moorings
(Peters [1958], Chs. 4 and 5). Drive-reduction
became
a classic example of twentieth-century metaphysics.
4. E. R. Guthrie.
Hull had been content to state
empirical laws at the molar level in
terms of actions
such as “biting the floor-bars” and
“leaping the barrier”
on the assumption that laws at
this level of description
could eventually be deduced from physiological
postu-
lates. E. R. Guthrie, on the other
hand, a contemporary
of Hull, eschewed physiological speculation and at-
tempted to reduce behavior at the molar
level to
movements such as muscle contractions and glandular
secretions, between which correlations could be stated.
He claimed that all
such correlations were derivable
from the old associationist law of
contiguity namely
that stimuli acting at the time of a response tend
on
their recurrence to evoke that response. He was an
S-R theorist par
excellence.
Guthrie was one of the few behaviorists to stress the
difference between
acts and movements. An act, he
claimed quite rightly, is a movement, or a
series of
movements, that brings about an end and acts are
classified
in terms of the ends which they bring about.
Learning, he argued, deals
with movements, not with
acts. Thorndike's law of effect concerns acts,
not
movements. It does not therefore deal with the basic
laws of
learning which state correlations between
movements—e.g.,
between the stimulation of a sense
organ and a muscular contraction. In a
famous experi-
ment which he did with Horton
(Guthrie and Horton,
1946) he placed a cat in a box, release from
which
was obtained by touching a pole in the middle of the
floor. It
was demonstrated that the cat tended to repeat
the posture in which it
first touched the pole and
obtained release. This experiment at least
showed that
contiguity is an important principle of learning; it did
not establish that it is the only principle and later
experiments (e.g., by
Seward) showed that improve-
ment in
learning was brought about by providing an
additional reward. Whether
Guthrie's experiment
showed anything about the wider issue of the impor-
tance of movements in learning as
distinct from acts
is quite another question. It is significant that Guthrie
had
to go to extreme lengths in constructing a situation
where no intelligence
was required to escape from the
box, in order to make his reduction look in
the least
plausible. Nevertheless Guthrie was an important figure
in
the history of behaviorism because he at least saw
the importance of the
distinction between movements
and acts, and because he saw it as an
obstacle in the
path of any reductionist program.
5. B. F. Skinner.
Skinner is the last survivor of the
great men of the behaviorist era,
but in many ways
he is the most old-fashioned of all of them in his
methodology; for in Skinner we encounter the pure
strain of the inductivist
doctrine of scientific method.
Skinner believes that a scientist must start
from empir-
ical data and gradually move
towards making inductive
generalizations or laws. Then, at some later
stage, he
may be in a position to formulate a theory which unites
the
laws. He must therefore be very careful to start
from reliable public data.
Skinner admits that men
have “inner lives” which are
of importance to them
as well as to novelists, as Skinner himself
portrayed
in his novel Walden Two. But data drawn
from this
source can never form a reliable basis for a science.
Skinner's polemics against other psychologists, such as
Freudians, who
based generalizations on data drawn
from this inner realm, have been as
forceful as Watson's
polemics against the introspectionists.
Skinner has also accepted the inductivist warning
that a scientist must
never go beyond the observable
in order to explain the observed. He has had
no more
use for physiological speculation about what goes on
inside
the organism than he has had for mentalistic
constructs. He accords a
limited importance to
Tolman's intervening variables such as
“drive” pro-
vided that
it is clearly understood that such terms are
shorthand symbols for
designating the operations by
which the rate of responses can be measured.
To speak
of hunger as a drive, for instance, is to pick out the
effects of operations such as deprivation on the proba-
bility of eating behavior.
Another significant feature of Skinner's approach is
his operationalism,
which has recently been fashionable
as a theory about the language of
science. (See I, 5
above and Peters, 1951.) To Skinner
this meant “the
practice of talking about (1) one's
observations, (2) the
manipulative and calculational procedures involved
in
making them, (3) the logical and mathematical steps
which intervene
between earlier and later statements,
and (4) nothing else”
(Skinner [1945], p. 270). This
doctrine maintained that a term like
“length” or
“hunger” refers not
to a characteristic of an object or
to a state of an organism but to the
experimenter's
operations of observing, manipulating, and measuring
it. It was an offshoot of positivism and of the verifica-
tionist theory of meaning which
came to the fore
during the period between the two World Wars.
This theory of meaning has now been abandoned
by most philosophers. But it
lives on in the meth-
odology of Skinner and
some other behaviorists, where
it has the added appeal of being in line
with the em-
phasis on control and manipulation
of the environment
which was so characteristic of behaviorism in the
Watsonian tradition. Behaviorism was in many respects
an offshoot of
American pragmatism. The experimenter
has not got to trouble his head with
theoretical ques-
tions about why organisms
behave as they do, espe-
cially if reference
might be made to recondite inner
causes in order to answer them. It is
sufficient to see
what forms of behavior develop if one environmental
variable is manipulated rather than another. This will
lead to predictions
which will eventually enable the
experimenter to
“shape” behavior.
Skinner claims that he has no “theory” of behavior
but
only notes correlations. But this, of course, is either
naive or a matter
of stipulation about the use of the
term “theory.” In
fact his work has presupposed a
biological theory of a Darwinian type in
which condi-
tioned reflexes are postulated
as having survival value.
In formulating the laws in accordance with which
these
“reflexes” are built up Skinner in fact revived
many
of the established principles of associationist theory.
In formulating these laws Skinner made an important
distinction between
“respondent” and “operant” be-
havior. This was facilitated by his
introduction of the
Skinner box, which enabled him to study
instrumental
conditioning in a much more controlled way than had
been
possible in Thorndike's puzzle box. In a respond-
ent reaction there is a known stimulus, such as the
ticking of
the metronome, with which a reaction such
as salivation can be correlated
as in classical condition-
ing. In an
operant response, however, such as lever
pressing, there are no known
stimuli with which the
response can be correlated in this way. There
may,
of course, be some form of internal stimulation, but
such
speculations were ruled out by Skinner's opera-
tionalist approach. So operant responses must be re-
garded as functions of experimental conditions such as
food schedules which can be manipulated by the ex-
perimenter. As behavior consists largely of such oper-
ant responses, which are instrumental in obtaining a
variety of goals, Skinner thought that the study of
conditioned operants
and their extinction must provide
the basic laws which would enable
behavior to be
predicted and controlled. One day a theory might be
devised to unify these laws; but the scientist must
proceed to such
“interpretations” in a Baconian man-
ner. He must not “anticipate” Nature
by premature
theorizing—especially if this involves speculations
like
those of Hull about the internal workings of the orga-
nism. Thus Skinner rejected the peripheralist
approach
of Watson but has remained agnostic about the central
processes which mediate between stimuli and re-
sponses. Operant conditioning has been, in fact, an-
other way of reformulating Thorndike's Law of Effect,
in nonmentalistic terminology.
Like Watson, Skinner has not been averse to extend-
ing his conceptual scheme to cover other aspects of
behavior.
For instance in his Science and Human Be-
havior in 1953 Skinner made pronouncements
about
emotions, the names of which serve to classify behavior
with
respect to various circumstances which affect its
probability. In spite,
too, of his hardheaded positivistic
approach in his Verbal Behavior (1957), he outlined
an ambitious scheme for
including language within the
behavioristic framework. This work, however,
was just
about as programmatic as Hull's Principles of
Behavior,
and has been severely criticized by philosophers
and
linguists alike (Chomsky, 1959).
In recent times Skinner has been very much pre-
occupied with providing a technology of teaching in
which skills
and sequences of material are carefully
broken down and the path of
learning systematically
shaped by positive reinforcement. However, his
con-
cept of
“reinforcement,” which has always been criti-
cized for its obscurity and circularity,
has undergone
such changes that his recipes for teaching amount to
little more than injunctions that material should be
logically analyzed and
students should be taken
through it step by step in a way which minimizes
the
repetition of mistakes, and which supplies constant
rewards for
success. This type of procedure, as Skinner
himself has admitted, could be
devised without much
reference to his elaborate laws of operant behavior.