4

A Review of B. F. Skinner's Verbal Behavior

Noam Chomsky .

1 addressed is that of giving a “functional

A great many linguists and philoso- analysis” of verbal behavior. By3y Function-
phers concerned with language have ex- al analysis, Skinner means identification.
pressed the hope that their studies might of the variables that control this behavior
ultimately be embedded in a framework and ‘specification of how they interact to
provided by behaviorist psychology, and determine a: particular verbal response.
that refractory areas of investigation, par- Furthermore, the controlling variables are _
ticularly those in which meaning is in- to be described completely in terms of
volved, will in this way be opened up to such notions as stimulus, reinforcement,
fruitful .exploration. Since this volume deprivation, which have been given a rea-
[Verbal Behavior (New York: Appleton- sonably clear meaning in animal experi-
Century-Crofts, 1957)—Ed. ] is the first mentation. In other words, the goal of the
large-scale attempt to incorporate the book is to provide a way to predict and
major aspects of linguistic behavior within control verbal behavior by observing and
a behaviorist framework, it merits and manipulating the physical environment of
will undoubtedly receive careful attention. the speaker.
Skinner is noted for his contributions to Skinner feels that recent advances in
the study of animal behavior. The book the laboratory study of animal behavior
under review is the product of study of permit us to approach this problem with a
linguistic behavior extending over more certain optimism, since “the basic pro-
than twenty years. Earlier versions of it cesses and relations which give verbal be-
have been fairly widely circulated, and havior its special characteristics are now
there are quite a few references in the psy- fairly well understood . . . the results [of
chological literature to its major ideas. this experimental work] have been sur--
The problem to which this book is prisingly free of species restrictions. Re-
cent work has shown that the methods
can be extended to human behavior with-
From Language 35, no. 1 (1959): 26-58. Re-
out serious modification (3).?
printed by permission of the Linguistic Society of
America and the author. Sections 5-10 have been It is important to see clearly just what
omitted (the notes are therefore not numbered it is in Skinner’s program and claims that
consecutively). makes them appear so bold and remark-
4. A Review of B. F. Skinner’s Verbal Behavior 49

able. It is not primarily the fact that he and defended this definition of a problem
has set functional analysis as his problem, as if it were a thesis which other investi-
gators reject. The differences that arise
or that he limits himself to study of ob-
servables, i.e., input-output relations. between those who affirm and those who |
What is so surprising is the particular lim- deny the importance of the specific “con-
itations he has imposed on the way in tribution of the organism” to learning and
which the observables of behavior are to performance concern the particular char-
be studied, and, above all, the particular- acter and complexity of this function, and
ly simple nature of the function which, he the kinds of observations and research
‘claims, describes the causation of behav- necessary for arriving at a precise specifi-
ior. One would naturally expect that pre- cation of it. If the contribution of the or-
ganism is complex, the only hope of pre- .
diction of the behavior of a complex or-
ganism (or machine) would require, in dicting behavior even in a gross way will
- addition to information about external be through a very indirect program of re-
stimulation, knowledge of the internal search that begins by studying the detailed
structure of the organism, the ways in character of the behavior itself and the
which it processes input information and particular capacities of the organism in-
organizes its own behavior. These charac- volved.
teristics of the organism are in general a Skinner's thesis is that external fac-
complicated product of inborn structure, tors consisting of present stimulation and
the genetically determined course of mat- the history of reinforcement (in particular,
uration, and past experience. Insofar as the frequency, arrangement, and with-
independent neurophysiological evidence holding of reinforcing stimuli) are of over-
is not available, it is obvious that infer- - whelming importance, and that the gener-
ences concerning the structure of the or- al principles revealed in laboratory studies
ganism are based on observation of be- of these phenomena provide the basis for
havior and outside events. Nevertheless, understanding the complexities of verbal
one’s estimate of the relative importance behavior. He confidently and repeatedly
of external factors and internal structure voices his claim to have demonstrated that
in the determination of behavior will have the contribution of the speaker is quite |
an important effect on the duration of re- trivial and elementary, and that precise
search on linguistic (or any other) behav- prediction of verbal behavior involves ~
ior, and on the kinds of analogies from only specification of the few external fac-
animal behavior studies that will be con- tors that he has isolated experimentally
sidered relevant or suggestive. with lower organisms.
.x» Putting it differently, anyone who Careful study of this book (and of the
sets himself the problem of analyzing the research on which it draws) reveals, how-
causation of behavior will (in the absence ever, that these astonishing claims are far
It indicates, furthermore,
of independent neurophysiological evi- from justified.
dence) concern himself with the only data that the insights that have been achieved
available, namely the record of inputs to in the laboratories of the reinforcement
the organism and the organism's present theorist, though quite genuine, can be
response, and will try to describe the func- applied to complex human behavior only
tion specifying the response in terms of in the most gross and superficial way, and
the history of inputs. This is nothing more that speculative attempts to discuss lin-
than the definition of his problem. There guistic behavior in these terms alone omit
are no possible grounds for argument trom consideration factors of fundamental
here, if one accepts the problem as legiti- importance that are, no doubt, amenable
mate, though Skinner has often advanced to scientific study, although their specific
50 Noam Chomsky

character cannot at present be precisely divides the responses of the animal into
formulated. Since Skinner’s work is the two main categories. Respondents are
most extensive attempt to accommodate purely reflex responses elicited by particu-
human behavior involving higher mental lar stimuli. Operants are emitted’ re-
faculties within a strict behaviorist schema sponses, for which no obvious stimulus
of the type that has attracted many lin- can be discovered. Skinner has been con-
guists and philosophers, as well as psy- cerned primarily with operant behavior.
chologists, a detailed documentation is of The experimental arrangement that he in-
independent interest. The magnitude of troduced consists basically of a box with a
-the failure of this attempt to account for bar attached to one wall in such a way
verbal behavior serves as a kind of mea- that when the bar is pressed, a food pellet
sure of the importance of the factors omit- is dropped into a tray (and the bar press is
ted from consideration, and an indication recorded). A rat placed in the box will
of how little -is really known about this soon press the bar, releasing a pellet into
remarkably complex phenomenon. the tray. This state of affairs, resulting
The force of Skinner’s argument lies from the bar press, increases the strength
in the enormous wealth and range of ex- of the bar-pressing operant. The food pel-
amples for which he proposes a functional let is called a reinforcer; the event, a rein-
analysis. The only way to evaluate the forcing event. The strength of an operant:
success of his program and the correctness is defined by Skinner in terms of the rate
of his basic assumptions about verbal be- of response during extinction (i.e., after
havior is to review these examples in de- the last reinforcement and before return
tail and to determine the precise character to the pre-conditioning rate).
of the concepts in terms of which the func- Suppose that release of the pellet is
tional analysis is presented. Section 2 of conditional on the flashing of a light.
this review describes the experimental Then the rat will come to press the bar
context with respect to which these con- only when the light flashes. Thisis called
cepts are originally defined. Sections 3 stimulus discrimination. The response is
and 4 deal with the basic concepts—stim- called a discriminated operant and the
ulus, response, and reinforcement—Sec- light is called the occasion for its emission:
tions 6 to 10 with the new descriptive ma- this is to be distinguished from elicitation
chinery developed specifically for the de- of a response by a stimulus in the case of.
scription of verbal behavior. In Section 5 the respondent.? Suppose that the appara-
we consider the status of the fundamental tus is so atranged that bar-pressing of only
claim, drawn from the laboratory, which a certain character (e.g., duration) will
serves as the basis for the analogic guesses release the pellet. The rat will then come
about human behavior that have been to press the bar in the required way. This
proposed by many psychologists. The fi- process is called response differentiation.
nal section (Section 11) will consider some By successive slight changes in the condi-
ways in which further linguistic work may tions under which the response will be re-
play a part in clarifying some of these inforced, it is possible to shape the re-
problems. sponse of a rat or a pigeon in very surpris-
ing ways in a very short time, so that
2 rather complex behavior can be produced
Although this book makes no direct by a process of successive approximation.
reference to experimental work, it can be A stimulus can become reinforcing
understood only in terms of the general by repeated association with an already
framework that Skinner has developed reinforcing stimulus. Such a stimulus is
for the description of behavior. Skinner called a secondary reinforcer. Like many
e
51
4. A Review of B. F. Skinner's Verbal Behavior

contemporary behaviorists, Skinner con- clude that behavior has not been demon-
siders money, approval, and the like to be strated to be lawful. In the present state of
secondary reinforcers which have become our knowledge, we must attribute an
reinforcing because of their association overwhelming influence on actual behav-
with food, etc.3 Secondary reinforcers can ior to ill-defined factors of attention, set,
volition, and caprice. If we accept the
be generalized by associating them with a
variety of different primary reinforcers. narrower definitions, then behavior is
Another variable that can affect the lawful by definition (if it consists of re-
rate of the bar-pressing operant is drive, sponses); but this fact is of limited signifi-
‘which Skinner defines operationally in cance, since most of what the animal does
terms of hours of deprivation. His major will simply not be considered behavior.
scientific book, Behavior of Organisms, is Hence, the psychologist either must admit
a study of the effects of food-deprivation that behavior is not lawful (or that he can-
and conditioning on the strength of the not at present show that it is—not at all a
bar-pressing response of healthy mature damaging admission for a developing sci-
rats. Probably Skinner’s most original ence), or must restrict his attention to
contribution to animal behavior studies those highly limited areas in which it is
has been his investigation of the effects of lawful (e.g., with adequate controls, bar-
intermittent reinforcement, arranged in pressing in rats; lawfulness of the observed
various different ways, presented in Be- behavior provides, for Skinner, an im-
havior of Organisms and extended (with plicit definition of a good experiment).
pecking of pigeons as the operant under Skinner does not consistently adopt
investigation) in the recent Schedules of either course. He utilizes the experimental
Reinforcement by Ferster and Skinner results as evidence for the scientific char-
(1957). It is apparently these studies that acter of his system of behavior, and ana-
Skinner has in mind when he refers to the logic guesses (formulated in terms of a
recent advances in the study of animal metaphoric extension of the technical vo-
cabulary of the laboratory) as evidence
behavior.‘
The notions stimulus, response, rein- for its scope. This creates the illusion of a
forcement are relatively well defined with rigorous scientific theory with a very
respect to the bar-pressing experiments broad scope, although in fact the terms
and others similarly restricted. Before we used in the description of real-life and of
can extend them to real-life behavior, laboratory behavior may be mere homo-
however, certain difficulties must be nyms, with at most a vague similarity of
faced. We must decide, first of all, whether meaning. To substantiate this evaluation, .
any physical event to which the organism a critical account of his book must show
is capable of reacting is ‘to be called a stim- that witha literal reading (where the terms
ulus on a given occasion, ‘or ‘only one to of the descriptive system have something
which the organism in-fact’ réacts; ‘and_, like the technical meanings given in Skin-
ner’s definitions) the book covers almost
correspondingly, we must ‘decide whether.
any part of behavior is to be called a_re- of linguistic behavior, and that
no aspect
reading, it is no more
sponse, or only one “Connected with stim- with a metaphoric

uli in lawful ways. Questions. of this sort scientific than the traditional approaches
pose something of a dilemma for the to this subject matter, and rarely as clear
experimental psychologist. If he accepts and careful.§
the broad definitions, characterizing any 3
physical event impinging on the organism
as a stimulus and any part of the organ- Consider first Skinner's use of the no-
ism’s behavior as a response, he must con- tions stimulus and response. In Behavior
52 Noam Chomsky

himself to lost all objectivity in this usage. Stimuli

of Organisms (9) he commits
the narrow definitions for these terms. A are no longer part of the outside physical
part of the environment and a part of be- world; they are driven back into the or-
havior are called stimulus (eliciting, dis- ganism. We identify the stimulus when we
-criminated, or reinforcing) and response, hear the response. It is clear from such
respectively, only if they are lawfully re- examples, which abound, that the talk of
lated; that is, if the dynamic laws relating stimulus control simply disguises a com-
them show smooth and reproducible plete retreat to mentalistic psychology.
curves. Evidently, stimuli and responses, We cannot predict verbal behavior in .
so defined, have not been shown to figure terms of the stimuli in the speaker's envi-
very widely in ordinary human behavior.* ronment, since we do not know what the
- We can, in the face of presently available current stimuli are until he responds. Fur-
evidence, continue to maintain the lawful- thermore, since we cannot control the
ness of the relation between stimulus and property of a physical object to which an
. response only by depriving them of their individual will respond, except in highly
objective character. A typical example of artificial cases, Skinner’s claim that his
stimulus control for Skinner would be the system, as opposed to the traditional one,
permits the practical control of verbal be-
‘ response to a piece of music with the utter-
_ance Mozart or to a painting with the re- havior’ is quite false.
sponse Dutch. These responses are assert- Other examples of stimulus control
ed to be “under the control of extremely merely add to the general mystification.
subtle properties” of the physical object or Thus, a proper noun is held to be a re-
event (108). Suppose instead of saying sponse “under the control of a specific
Dutch we had said Clashes with the wall- person or thing” (as controlling stimulus,
paper, I thought you liked abstract work, 113). I have often used the words Eisen-
Never saw it before, Tilted, Hanging too hower and Moscow, which I presume are
low, Beautiful, Hideous, Remember our proper nouns if anything is, but have
camping trip last summer?, or whatever never been stimulated by the correspond-
else might come into our minds when ing objects. How can this fact be made
looking at a picture (in Skinnerian trans- compatible with this definition? Suppose
lation, whatever other responses exist in that I use the name of a friend who is not
sufficient strength). Skinner could only present. Is this an instance of a proper
the friend as
say that each of these responses is under _ noun under the control of
stimulus? Elsewhere it is asserted that a
the control of some other stimulus prop-
erty of the physical object. If we look at a stimulus controls a response in the sense
red chair and say red, the response is un- that presence of-the stimulus increases the
der the control of the stimulus redness; if probability of the response. But it is ob-
we say chair, it is under the control of the viously untrue that the probability that a
collection of properties (for Skinner, the speaker will produce a full name is in-
object) chairness (110), and similarly for creased when its bearer faces the speaker.
any other response. This device is as sim- Furthermore, how can one’s own name be
ple as it is empty. Since properties are free a proper noun in this sense? A multitude
for the asking (we have as many of them of similar questions arise immediately. It
as we have nonsynonymous descriptive appears that the word control here is
expressions in our language, whatever this merely a misleading paraphrase for the
means exactly), we can account fora wide traditional denote or refer. The assertion
class of responses in terms of Skinnerian (115) that so far as the speaker is con-
functional analysis by identifying the con- cerned, the relation of reference is “simply
emit
trolling stimuli. But the word stimulus has | _ the probability that the speaker will
4. A Review of B. F. Skinner’s Verbal Behavior 53

a response of a given form in the presence identification of a unit of verbal behavior,

of a stimulus having specified properties” but is satisfied with an answer so vague
is surely incorrect if we take the words and subjective that it does not really con-
presence, stimulus, and probability in tribute to its solution. The unit of verbal
their literal sense. That they are not in- behavior—the verbal operant—is defined
tended to be taken literally is indicated by as a class of responses of identifiable form
many examples, as when a response is functionally related to one or more con-
said to be “controlled” by a situation or trolling variables. No method is suggested
state of affairs as “stimulus.” Thus, the for determining in a particular instance —
expression a needle in a haystack “may be what are the « controlling variables, how
controlled as a unit by a particular type of many such units have occurred, or where
situation” (116); the words in a single part their boundaries are in the total response.
of speech, e.g., all adjectives, are under Nor i is any attempt made to specify how
the control ofa single set of subtle prop- much orr What kind of similarity in form.
erties of stimuli (121); “the sentence The or control is required for two physical
boy runs a store is under the control of an events to be considered instances of the
extremely complex stimulus situation” same operant. In short, no answers are
(335); “He is not at all well may function suggested for the most elementary ques-
as a standard response under the control tions that must be asked of anyone pro-
of a state of affairs which might also con- posing a method for description of behav-
trol He is ailing” (325); when an envoy ior. Skinner is content with what he calls.
observes events in a foreign country and an extrapolation of the concept of operant
reports upon his return, his report is un- developed in the laboratory to the verbal
der “remote stimulus control” (416); the field. In the typical Skinnerian experi-
statement This is war may be a response ment, the problem of identifying the unit
to a “confusing international situation” of behavior is not too crucial. It is defined,
(441); the suffix -ed is controlled by that by fiat, as a recorded peck or bar-press,
“subtle property of stimuli which we and systematic variations in the rate of
speak of as action-in-the-past” (121) just this operant and its resistance to extinc-
as the -s in The boy runs is under the con- tion are studied as a function of depriva-
trol of such specific features of the situ- tion and scheduling of reinforcement (pel-
ation as its “currency” (332). No charac- lets). The operant is thus defined with
terization of the notion stimulus control respect to a particular experimental pro-
_ that is remotely related to the bar-pressing cedure. This is perfectly reasonable and
' experiment (or that preserves the faintest has led to many interesting results. It is,
objectivity) can be made to cover a set of however, completely meaningless to
+ examples like these, in which, for exam- speak of extrapolating this concept of
‘ple, the controlling stimulus need not even operant to ordinary verbal behavior.
impinge on the responding organism. Such “extrapolation” leaves us with no
Consider now Skinner's use of the way of justifying one or another decision
notion response. The problem of identify- about the units in the “verbal repertoire.”
ing units in verbal behavior has of course Skinner specifies “response strength”
been a primary concern of linguists, and it as the basic datum, the basic dependent
seems very likely that experimental psy- variable in his functional analysis. In
chologists should be able to provide much- the bar-pressing experiment, response
needed assistance in clearing up the many strength is defined in terms of rate of
remaining difficulties in systematic identi- emission during extinction. Skinner has
fication. Skinner recognizes (20) the fun- argued that this is “the only datum that
damental character of the problem of varies significantly and in the expected
54 Noam Chomsky

direction under conditions which are rele- claim Beautiful!, the speed and energy of
vant to the ‘learning process.’” In the the response will not be lost on the own-
book under review, response strength is er.” It does not appear totally obvious
defined as “probability of emission” (22). that in this case the way to impress the .
This definition provides a comforting im- owner is to shriek Beautiful in a loud,
pression of objectivity, which, however, high-pitched voice, repeatedly, and with
is quickly dispelled when we look into the no delay (high response strength). It may
matter more closely. The term probability be equally effective to look at the picture
has some rather obscure meaning for silently (long delay) and then to murmur
. Skinner in this book.’ We are told, on the Beautiful in a soft, low-pitched voice (by
one hand, that “our evidence for.the con- definition, very low response strength).
tribution of each variable [to response It is not unfair, I believe, to conclude
strength] is based on observation of fre- from Skinner’s discussion of response
quencies alone” (28). At the same time, it strength, the basic datum in functional
appears that frequency is a very mislead- analysis, that his extrapolation of the no-
ing measure of strength, since, for exam- tion of probability can best be interpreted
ple, the frequency of a response may be as, in effect, nothing more than a decision
“primarily attributable to the frequency to use the word probability, with its fa-
of occurrence of controlling variables” vorable connotations of objectivity, as a
(27). It is not clear how the frequency of a cover term to paraphrase such low-status
response can be attributable to anything words as interest, intention, belief, and
ut the frequency of occurrence ofits con- the like. This interpretation is fully justi-
trolling variables if we accept Skinner's fied by the way in which Skinner uses the
view that the behavior occurring in a terms probability and strength. To cite
given situation is “fully determined” by just one example, Skinner defines the pro-
the relevant controlling variables (175, cess of confirming an assertion in science
228). Furthermore, although the evidence as one of “generating additional variables
for the contribution of each variable to to increase its probability” (425), and
response strength is based on observation more generally, its strength (425-29). If
of frequencies alone, it turns out that “we we take this suggestion quite literally, the
base the notion of strength upon several degree of confirmation of a scientific as-
kinds of evidence” (22), in particular (22- sertion can be measured as a simple func-
28): emission of the response (particularly tion of the loudness, pitch, and frequency
in unusual circumstances), energy level with which it is proclaimed, and a general
(stress), pitch level, speed and delay of procedure for increasing its degree of con-
emission, size of letters etc. in writing, firmation would be, for instance, to train
immediate repetition, and—a final factor, machine guns on large crowds of people
relevant but misleading—over-all fre- who have been instructed to shout it. A
quency. . better indication of what Skinner prob-
_Of course, Skinner recognizes that ably has in mind here is given by his de-
these measures do not co-vary, because scription of how the theory of evolution,
(among other reasons) pitch, stress, quan- as an example, is confirmed. This “single
tity, and reduplication may have internal set of verbal responses . . . is made more
linguistic functions.1° However, he does plausible—is strengthened—by several
not hold these conflicts to be very impor- types of construction based upon verbal
tant, since the proposed factors indicative responses in geology, paleontology, ge-
of strength are “fully understood by every- netics, and so on” (427). We are no doubt
one” in the culture (27). For example, “if to interpret the terms strength and prob-
we are shown a prized work of art and ex- ability in this context as paraphrases of
55
4. A Review of B. F. Skinner's Verbal Behavior

more familiar locutions such as “justified uations and conditions under which they
belief’ or “warranted assertability,” or are reinforcing). Consider first of all the
something of the sort. Similar latitude of status of the basic principle that Skinner
interpretation is presumably expected calls the “law of conditioning” (law of ef-
when we read that “frequency of effective fect). It reads: “if the occurrence of an
action accounts in turn for what we may operant is followed by presence of a rein-
call the listener's ‘belief’ ” (88) or that “our forcing stimulus, the strength is increased”
(Behavior of Organisms, 21). As rein-
belief in what someone tells us is similarly
a function of, or identical with, our ten- forcement was defined, this law becomes
dency to act upon the verbal stimuli which a tautology.?? For Skinner, learning is just
he provides” (160). change in response strength. Although
I think it is evident, then, that Skin- the statement that presence of reinforce-
ner’s use of the terms stimulus, control, ment is a sufficient condition for learning
response, and strength justify the general and maintenance of behavior is vacuous,
conclusion stated in the last paragraph of the claim that it is a necessary condition
Section 2. The way in which these terms may have some content, depending on
are brought to bear on the actual data in- how the class of reinforcers (and appro-
dicates that we must interpret them as priate situations) is characterized. Skinner
mere paraphrases for the popular vocabu- does make it very clear that in his view
lary commonly used to describe behavior reinforcement is a necessary condition for
and as having no particular connection language learning and for the continued
~ with the homonymous expressions used in availability of linguistic responses in the
the description of laboratory experiments. adult.> However, the looseness of the
Naturally, this terminological revision term reinforcement as Skinner uses it in
adds no objectivity to the familiar mental- the book under review makes it entirely
istic mode of description. pointless to inquire into the truth or falsity
of this claim. Examining the instances of
4 what Skinner calls reinforcement, we find
The other fundamental notion bor- that not even the requirement that a rein-
rowed from the description of bar-pressing forcer be an identifiable stimulus is taken
experiments is reinforcement. It raises seriously. In fact, the term is used in such
problems which are similar, and even a way that the assertion that reinforce-
more serious. In Behavior of Organisms, ment is necessary for learning and con-
“the operation of reinforcement is defined tinued availability of behavior is likewise
as the presentation of a certain kind of empty.

stimulus in a temporal relation with either To show this, we consider some ex-
a stimulus or response. A reinforcing amples of reinforcement. First of all, we
stimulus is defined as such by its power to find a heavy appeal to automatic self-
produce the resulting change [in strength]. reinforcement. Thus, “a man talks to him-
There is no circularity about this: some self . . because of the reinforcement he
(163); “the child is reinforced
stimuli are found to produce the change, receives”
the
automatically when he duplicates
others not, and they are classified as rein-
forcing and nonreinforcing accordingly” sounds of airplanes, streetcars . . .” (164);
- (62). This is a perfectly appropriate defi- _ “the young child alone in the nursery may
nition” for the study of schedules of rein- automatically reinforce his own explora-
l behavior when he produces
forcement. Itis perfectly useless, however, ~ tory verba
in the discussion of real-life behavior, un- sounds which he has heard in the speech
less we can somehow characterize the of others” (58); “the speaker who is also
stimuli which are reinforcing (and the sit- an accomplished listener ‘knows when he
Noam Chomsky
56
is re- name though in fact it was not mentioned
has correctly echoed a response’ and
av- or to hear nonexistent words in his child's
inforced thereby” (68); thinking is “beh
behav- babbling (259), to clarify or otherwise in-
ing which automatically affects the the effect of a. stimulus which
so” tensify
er and is reinforcing because it does
be serves an important discriminative func-
(438; cutting one’s finger should thus
king); tion (416), and so on.
reinforcing, and an example of thin
or co- From this sample, it can be seen that
“the verbal fantasy, whether overt
the the notion of reinforcement has totally
vert, is automatically reinforcing to
lost whatever objective meaning it may
speaker as listener. Just as the musician
ed ever have had. Running through these ex-
plays or composes what he is reinforc
by hearing, or as the artist paints what amples, we see that a person can be rein-
ker forced though he emits no response at all,
reinforces him visually, so the spea
he is and that the reinforcing stimulus need not
engaged in verbal fantasy says what
is impinge on the reinforced person or need
reinforced by hearing or writes what he
not even exist (it is sufficient that it be
reinforced by reading” (439); similarly,
iza- imagined or hoped for). When we read
care in problem solving, and rational likes
g | that a person plays what music he
tion, ‘are automatically self-reinforcin , thinks
he likes (165)
one (165), says what
(442-43). We can also reinforce some
as such (sinc e what he likes (438-39), reads what books
by emitting verbal behavior it re-
he likes (163), etc., BECAUSE he finds
this rules out a class of aversive stimula-
v- inforcing to do so, or that we write books
tions, 167), by not emitting verbal beha are
or inform others of facts BECAUSE we
ior (keeping silent and paying attention,
reinforced by what we hope will be the
199), or by acting appropriately on some we
[the ultimate behavior of reader or listener,
future occasion (152: “the strength of orce-
can only conclude that the term reinf
speaker's] behavior is determined mainly The
by the behavior which the listener will ex-
ment has a purely ritual function.
phrase “X is reinforced by Y (stimulus,
hibit with respect to a given state of af- used
ral state of affairs, event, etc.)” is being
fairs’; this Skinner considers the gene likes
the as a cover term for “X wants Y,” "X
case of “communication” or “letting etc.
‘cases, of Y,” “X wishes that Y were the case,”
listener know”). In most such
Invoking the term reinforcement has no
course, the speaker is not present at the this
, explanatory force, and any idea that
time when the reinforcement takes place ty or
the paraphrase introduces any new clari
as when “the artist . . . is reinforced by ng,
- others” objectivity into the description of wishi
effects his works have upon... delus ion. The only
liking, etc., is a serious
(224), or when the writer is reinforced by -
effect is to obscure the important differ
the fact that his “verbal behavior may para- —
of ences among the notions being
reach over centuries or to thousands the latit ude
phrased. Once we recognize
listeners or readers at the same time. The is be-
with which the term reinforcement
writer may not be reinforced often or im- ents
may ing used, many rather startling comm
mediately, but his net reinforcement that
great lose their initial effect—for instance,
be great” (206; this accounts for the is “con-
the behavior of the creative artist
“strength” of his behavior). An individual of re-
find it reinforcing to injure trolled entirely by the contingencies
may also hoped
inforcement” (150). What has been
someone by criticism or by bringing bad indic a-
for from the psychologist is some
news, or to publish an experimental result
to tion how the casual and informal descrip-
which upsets the theory of a rival (154), lar
be tion of everyday behavior in the popu
describe circumstances which would
reinforcing if they were to occur (165) , to vocabulary can be explained or clarified
care-
own in terms of the notions developed in
avoid repetition (222), to “hear” his
vior 57
4. A Review of B. F. Skinner’s Verbal Beha

per- ed, the present state of motivation of the

ful experiment and observation, or
me. listener, etc., the behavior may vary from
haps replaced in terms of a better sche
in which rage to pleasure, from fixing the object to
A mere terminological revision,
from the laboratory is throwing it out, from simply not using it
a term borrowed
to trying to use it in the normal way (e.g.,
used with the full vagueness of the ordi-
in- to see if it is really out of order), and so
nary vocabulary, is of no conceivable
terest.
on. To speak of “conditioning” or “bring-
It seems that Skinner's claim that all ing previously available behavior under
control of a new stimulus” in such a case
verbal behavior is acquired and main-
is just a kind of play-acting at science (cf.
tained in “strength” through reinforce-
also 43n).
ment is quite empty, because his notion of
reinforcement has no clear content, func-
tioning only as a cover term for any fac- 11
tor, detectable or not, related to acquisi-
The preceding discussion covers all
tion or maintenance of verbal behavior.”
the major notions that Skinner introduces
Skinner’s use of the term conditioning
in his descriptive system. My purpose in
suffers from a similar difficulty. Pavlo-
“discussing the concepts one by one was to
vian and operant conditioning are pro- case, if we take his
show that in each
cesses about which psychologists have
terms in their literal meaning, the descrip-
developed real understanding. Instruction
tion covers almost no aspect of verbal
of human beings is not. The claim that
behavior, and if we take them meta-
instruction and imparting of information
phorically, the description offers no im-
are simply matters of conditioning (357-
provement over various traditional for-
66) is pointless. The claim is true, if we
the term conditioning to cover mulations. The terms borrowed from ex-
extend
perimental psychology simply lose their
these processes, but we know no more
objective meaning with this extension,
about them after having revised this term
and take over the full vagueness of ordi-
in such a way as to deprive it of its rela-
nary language. Since Skinner limits him-
tively clear and objective character. It is,
quite false, if we use self to such a small set of terms for para-
as far as we know, distinctions are
many important
conditioning in its literal sense. Similarly,
phrase,
© obscured. I think that this analysis
sup-
when we say that “it is the function of
ports the view expressed in Section 1, that
predication to facilitate the transfer of re-
elimination of the independent contribu-
sponse from one term to another or from t
tion of the speaker and learner (a resul
one object to another” (361), we have said r-
which Skinner considers of great impo
nothing of any significance. In what sense eved only at
tance, cf. 311-12) can be achi
is this true of the predication Whales are ce
the cost of eliminating all significan
mammals? Or, to take Skinner's example, m, whic h then
from the descriptive syste
what point is there in saying that the effect
operates at a level so gross and crude that
of The telephone is out of order on the ele-
no answers are suggested to the most
listener is to bring behavior formerly con- ions to
mentary questions.‘ The quest
trolled by the stimulus out of order under
which Skinner has addressed his specula-
control of the stimulus telephone (or the is futile
tions are hopelessly premature. It
telephone itself) by a process of simple
to inquire into the causation of verbal be-
conditioning (362)? What laws of condi- about
havior until much more is known
tioning hold in this case? Furthermore,
the specific character of this behavior;
what behavior is controlled by the stimu- in speculating
and there is little point
lus out of order, in the abstract? Depend- out
about the process of acquisition with
ing on the object of which this is predicat-
58 | Noam Chomsky

these completely. It is reasonable to re- ~

much better understanding of what is ac-
quired. gard the grammar of a language L ideally
Anyone who seriously approaches as a mechanism that provides an enumer-
the study of linguistic behavior, whether ation of the sentences of L in something
linguist, psychologist, or philosopher, like the way in which a deductive theory
must quickly become aware of the enor- gives an enumeration of a set of theorems.
mous difficulty of stating a problem which (Grammar, in this sense of the word, in-
will define the area of his investigations, cludes phonology.) Furthermore, the the-
and which will not be either completely ory of language can be regarded as a study
trivial or hopelessly beyond the range of of the formal properties of such gram-
present-day understanding and technique. mars, and, with a precise enough formu-
In selecting functional analysis as his lation, this general theory can provide a
problem, Skinner has set himself a task of uniform method for determining, from
the latter type. In an extremely interesting the process of generation of a given sen-
and insightful paper,*” K. S. Lashley has tence, a structural description which can
implicitly delimited a class of problems give a good deal of insight into how this
which can be approached in a fruitful way sentence is used and understood. In short,
by the linguist and psychologist, and it should be possible to derive from a
which are clearly preliminary to those properly formulated grammar a statement
with which Skinner is concerned. Lashley of the integrative processes and general-
recognizes, as anyone must who seriously ized patterns imposed on the specific acts
considers the data, that the composition that constitute an utterance. The rules of a
and production of an utterance is not sim- grammar of the appropriate form can be
ply a matter of stringing together a se- subdivided into the two types, optional
quence of responses under the control of and obligatory; only the latter must be
outside stimulation and intraverbal asso- applied in generating an utterance. The
ciation, and that the syntactic organiza- optional rules of the grammar can be
tion of an utterance is not something di- viewed, then, as the selective mechanisms
rectly represented in any simple way in involved in the production of a particular
the physical structure of the utterance it- utterance. The problem of specifying these
self. A variety of observations lead him to integrative processes and selective mecha-
conclude that syntactic structure is “a gen- nisms is nontrivial and not beyond the
eralized pattern imposed on the specific range of possible investigation. The re-
acts as they occur” (512), and that “a con- sults of such a study might, as Lashley
sideration of the structure of the sentence suggests, be of independent interest for
and other motor sequences will show . . psychology and neurology (and converse-
that there are, behind the overtly expressed ly). Although such a study, even if suc-
sequences, a multiplicity of integrative cessful, would by no means answer the
processes which can only be inferred from major problems involved in the investiga-
the final results of their activity” (509). He tion of meaning and the causation of be-
also comments on the great difficulty of havior, it surely will not be unrelated to
these. It is at least possible, furthermore,
determining the “selective mechanisms”
used in the actual construction of a partic- that such a notion as semantic generaliza-
ular utterance (522). tion, to which such heavy appeal is made
Although present-day linguistics can- in all approaches to language in use, con-
not provide a precise account of these ceals complexities and specific structure of
integrative processes, imposed patterns, inference not far different from those that
and selective mechanisms, it can at least can be studied and exhibited in the case of
set itself the problem of characterizing syntax, and that consequently the general
4. A Review of B. F. Skinner's Verbal Behavior 59

character of the results of syntactic inves- to him and what conditions of compati-
tigations may be a corrective to oversim- bility the choices must meet, we can pro-
plified approaches to the theory of mean- ceed meaningfully to investigate the fac-
ing. tors that lead him to make one or another
choice. The listener (or reader) must de-
The behavior of the speaker, listener,
and learner of language constitutes, of termine, from an exhibited utterance,
course, the actual data for any study of what optional rules were chosen in the
language. The construction of a grammar construction of the utterance. It must be
which enumerates sentences in such a way admitted that the ability of a human being
that a meaningful structural description to do this far surpasses our present under-
can be determined for each sentence does standing. The child who learns a language
not in itself provide an account of this ac- has in some sense constructed the gram-
tual behavior. It merely characterizes ab- mar for himself on the basis of his ob-
stractly the ability of one who has mas- servation of sentences and nonsentences
tered the language to distinguish sentences (i.e., corrections by the verbal commu-
from nonsentences, to understand new nity). Study of the actual observed ability
sentences (in part), to note certain ambi- of a speaker to distinguish sentences from
guities, etc. These are very remarkable nonsentences, detect ambiguities, etc.,
abilities. We constantly read and hear apparently forces us to the conclusion that
new sequences of words, recognize them this grammar is of an extremely complex
as sentences, and understand them. It is and abstract character, and that the young
out what
easy to show that the new events that we _ child has succeeded in carrying
accept and understand as sentences are from the formal point of view, at least,
not related to those with which we are fa- seems to be a remarkable type of theory
this task is
miliar by any simple notion of formal (or construction. Furthermore,
accomplished in an astonishingly short
semantic or statistical) similarity or iden-
tity of grammatical frame. Talk of gener- time, to a large extent independently of
alization in this case is entirely pointless intelligence, and in a comparable way by
and empty. It appears that we recognize a all children. Any theory of learning must
new item as a sentence not because it cope with these facts.
matches some familiar item in any simple It is not easy to accept the view that a
way, but because it is generated by the child is capable of constructing an ex-
grammar that each individual has some- tremely complex mechanism for generat-
how and in some form internalized. And ing a set of sentences, some of which he
we understand a new sentence, in part, has heard, or that an adult can instanta-
because we are somehow capable of deter- neously determine whether (and if so,
mining the process by which this sentence how) a particular item is generated by this
is derived in this grammar. mechanism, which has many of the prop-
Suppose that we manage to construct erties of an abstract deductive theory. Yet
grammars having the properties outlined this appears to be a fair description of the
above. We can then attempt to describe performance of the speaker, listener, and
and study the achievement of the speaker, learner. If this is correct, we can predict
listener, and learner. The speaker and the that a direct attempt to account for the
listener, we must assume, have already actual behavior of speaker, listener, and
acquired the capacities characterized ab- learner, not based on a prior understand-
stractly by the grammar. The speaker's ing of the structure of grammars, will
task is to select a particular compatible set achieve very limited success. The gram-
of optional rules. If we know, from gram- mar must be regarded as a component in
matical study, what choices are available the behavior of the speaker and listener
60 Noam Chomsky

which can only be inferred, as Lashley has posium on Motivation, 58 (Lincoln, 1956);
W. S. Verplanck, “Learned and Innate Behav- |
put it, from the resulting physical acts.
ior,” Psych. Rev., 52 (1955), 139. Perhaps the
The fact that all normal children acquire
strongest view is that of H. Harlow, who has
essentially comparable grammars of great asserted (“Mice, Monkeys, Men, and Mo-
complexity with remarkable rapidity sug- tives,” Psych. Rev., 60 [1953], 23-32) that “a
gests that human beings are somehow strong case can be made for the proposition
specially designed to do this, with data- that the importance of the psychological prob-
handling or “hypothesis-formulating” lems studied during the last 15 years has de-
ability of unknown character and com- creased as a negatively accelerated function
plexity.** The study of linguistic structure approaching an asymptote of complete indif-
may ultimately lead to some significant ference.” N. Tinbergen, a leading representa-
insights into this matter. At the moment tive of a different approach to animal-behavior
the question cannot be seriously posed, studies (comparative ethology), concludes a
discussion of functional analysis with the com-
but in principle it may be possible to study
ment that “we may now draw the conclusion
the problem of determining what the built-
that the causation of behavior is immensely
in structure of an information-processing more complex than was assumed in the gen-
(hypothesis-forming) system: must be to eralizations of the past. A number of internal
enable it to arrive at the grammar of a and external factors act upon complex central
language from the available data in the nervous structures. Second, it will be obvious
available time. At any rate, just as the at- that the facts at our disposal are very fragmen-
tempt to eliminate the contribution of the tary indeed”— The Study of Instinct (Toronto:
speaker leads to a “mentalistic” descrip- Oxford Univ. Press, 1951), p. 74.
tive system that succeeds only in blurring 2. In Behavior of Organisms (New York:
important traditional distinctions, a re- Appleton-Century-Crofts, Inc., 1938), Skinner
fusal to study the contribution of the child remarks that “although a conditioned operant
to language learning permits only a super- is the result of the correlation of the response
with a particular reinforcement, a relation be-
ficial account of language acquisition,
tween it and a discriminative stimulus acting
with a vast and unanalyzed contribution prior to the response is the almost universal
attributed to a step called generalization rule” (178-79). Even emitted behavior is held
which in fact includes just about every- to be produced by some sort of “originating
thing of interest in this process. If the force” (51) which, in the case of operant be-
study of language is limited in these ways, havior, is not under experimental control. The
it seems inevitable that major aspects of distinction between eliciting stimuli, discrimi-
verbal behavior will remain a mystery. nated stimuli, and “originating forces” has
never been adequately clarified and becomes
~
even more confusing when private internal
events are considered to be discriminated stim-
Notes
uli (see below).
1. Skinner's confidence in recent achieve- 3. Ina famous experiment, chimpanzees
ments in the study of animal behavior and were taught to perform complex tasks to re-
their applicability to complex human behavior ceive tokens which had become secondary re-
does not appear to be widely shared. In many inforcers because of association with food.
recent publications of confirmed behaviorists The idea that money, approval, prestige, etc.
there is a prevailing note of skepticism with actually acquire their motivating effects on
regard to the scope of these achievements. For human behavior according to this paradigm is
representative comments, see the contributions - unproved, and not particularly plausible.
to Modern Learning Theory (by W. K. Estes Many psychologists within the behaviorist
et al.; New York: Appleton-Century-Crofts, movement are quite skeptical about this (cf.
Inc., 1954); B. R. Bugelski, Psychology of 23n). As in the case of most aspects of human
Learning (New York: Holt, Rinehart & Win- behavior, the evidence about secondary rein-
ston, Inc., 1956); S. Koch, in Nebraska Sym- forcement is so fragmentary, conflicting, and
' 61
4. A Review of B. F. Skinner’s Verbal Behavior

complex that almost any view can find some Skinner’s notion of lawful relation; and, at the
support. same time, noting the importance of the ex-
4. Skinner’s remark quoted above about. perimental data that Skinner has accumulated.
the generality of his basic results must be un- 6. In Behavior of Organisms, Skinner
derstood in the light of the experimental limi- apparently was willing to accept this conse-
quence. He insists (41-42) that the terms of
tations he has imposed. If it were true in any
deep sense that the basic processes in language casual description in the popular vocabulary
are well understood and free of species restric- are not validly descriptive until the defining
tion, it would be extremely odd that language properties of stimulus and response are speci-
is limited to man. With the exception of a few fied, the correlation is demonstrated experi-
scattered observations (cf. his article, “A Case mentally, and the dynamic changes in it are
History in Scientific Method,” The American shown to be lawful. Thus, in describing a child
Psychologist, 11 [1956], 221-33), Skinner is ap- as hiding from a dog, “it will not be enough to
parently basing this claim on the fact that dignify the popular vocabulary by appealing
qualitatively similar results are obtained with to essential properties of dogness or hidingness
bar pressing of rats and pecking of pigeons and to suppose them intuitively known.” But
under special conditions of deprivation and this is exactly what Skinner does in the book
under review, as we will see directly..
various schedules of reinforcement. One im-
mediately questions how much can be based 7. 253f. and elsewhere, repeatedly. As
on these facts, which are in part at least an arti- an example of how well we can control behav-
fact traceable to experimental design and the ior using the notions developed in this book,
definition of stimulus and response in terms of Skinner shows here how he would go about
smooth dynamic curves (see below). The dan- evoking the response pencil. The most effec-
gers inherent in any attempt to extrapolate to tive way, he suggests, is to say to the subject,
“Please say pencil” (our chances would, pre-
complex behavior from the study of such sim-
ple responses as bar pressing should be obvi- sumably, be even further improved by use of
“aversive stimulation,” e.g., holding a gun to
ous and have often been commented on (cf.,
e.g., Harlow, op. cit.). The generality of even his head). We can also “make sure that no
the simplest results is open to serious question. pencil or writing instrument is available, then
Cf. in this connection M. E. Bitterman, J. Wo- hand our subject a pad of paper appropriate to
dinsky, and D. K. Candland, “Some Com- pencil sketching, and offer him a handsome
parative Psychology,” Am. Jour. of Psych., 71 reward for a recognizable picture of a cat.” It
(1958), 94-110, where it is shown that there are
would also be useful to have voices saying
important qualitative differences in solution of pencil or penand... inthe background; signs
comparable elementary problems by rats and reading pencil or pen and... .; or to place a
fish. “large and unusual pencil in an unusual place
5. An analogous argument, in connec- clearly in sight.” “Under such circumstances, it
tion with a different aspect of Skinner's think- is highly probable that our subject will say
ing, is given by M. Scriven in “A Study of pencil.” "The available techniques are all illus-
Radical Behaviorism,” Univ. of Minn. Studies trated in this sample.” These contributions of
in Philosophy of Science, I. Cf. Verplanck’s ‘behavior theory to the practical control of hu-
contribution to Modern Learning Theory, op. man behavior are amply illustrated elsewhere
in the book, as when Skinner shows (113-14)
cit. pp. 283-88, for more general discussion of
the difficulties in formulating an adequate defi- how we can evoke the response red (the device
nition of stimulus and response. He concludes, suggested is to hold a red object before the
quite correctly, that in Skinner's sense of the subject and say, “Tell me what color this is’).
word, stimuli are not objectively identifiable In fairness, it must be mentioned that
independently of the resulting behavior, nor there are certain nontrivial applications of
are they manipulable. Verplanck presents a operant conditioning to the control of human
clear discussion of many other aspects of Skin- behavior. A wide variety of experiments have
ner’s system, commenting on the untestability
shown that the number of plural nouns (for
of many of the so-called “laws of behavior” example) produced by a subject will increase if
and the limited scope of many of the others, the experimenter says “right” or “good” when
and the arbitrary and obscure character of one is produced (similarly, positive attitudes
* 62 Noam Chomsky

on a certain issue, stories with particular con- tendency to act upon these verbal stimuli. We
tent, etc.; cf. L. Krasner, “Studies of the Con- may, of course, turn Skinner's assertion into a
ditioning of Verbal Behavior,” Psych. Bull., 55 very unilluminating truth by defining “tenden-
[1958], for a survey of several dozen experi- cy to act” to include tendencies to answer
ments of this kind, mostly with positive re- questions in certain ways, under motivation to
sults). It is of some interest that the subject is say what one believes is true.
usually unaware of the process. Just what in- 12. One should add, however, that itis in
sight this gives into normal verbal behavior is general not the stimulus as such that is rein-
not obvious. Nevertheless, it is an example of forcing, but the stimulus in a particular situ-
positive and not totally expected results using ational context. Depending on experimental
the Skinnerian paradigm. arrangement, a particular physical event or
8. “Are Theories of Learning Neces-- object may be reinforcing, punishing, or un-
sary?”, Psych. Rev., 57 (1950), 193-216. noticed. Because Skinner limits himself to a
9. And elsewhere. In his paper “Are’ particular, very simple experimental arrange-
Theories of Learning Necessary?” Skinner con- ment, it is not necessary for himto add this
siders the problem how to extend his analysis qualification, which would not be at all easy to
of behavior to experimental situations in formulate precisely. But it is of course neces-
which it is impossible to observe frequencies, sary if he expects to extend his descriptive sys-
rate of response being the only valid datum. tem to behavior in general.
His answer is that “the notion of probability is 13. This has been frequently noted.
usually extrapolated to cases in which a fre- 14. See, for example, “Are Theories of
quency analysis cannot be carried out. In the Learning Necessary?”, op. cit., p. 199. Else-
field of behavior we arrange a situation in where, he suggests that the term learning be
which frequencies are available as data, but we restricted to complex situations, but these are
use the notion of probability in analyzing or not characterized.
formulating instances of even types of behav- 15. “A child acquires verbal behavior
ior which are not susceptible to this analysis” when relatively unpatterned vocalizations,
(199). There are, of course, conceptions of selectively reinforced, gradually assume forms
probability not based directly on frequency, which produce appropriate consequences in a
but I do not see how any of these apply to the given verbal community” (31). “Differential
cases that Skinner has in mind. I see no way of reinforcement shapes up all verbal forms, and
interpreting the quoted passage other than as when a prior stimulus enters into the contin-
signifying an intention to use the word proba- gency, reinforcement is responsible for its re-
bility in describing behavior quite indepen- sulting control. ... The availability of behav-
dently of whether the notion of probability is ior, its probability or strength, depends on
at all relevant. whether reinforcements continue in effect and
10. Fortunately, “In English this presents according to what schedules” (203-4); else-
no great difficulty” since, for example, “rela- where, frequently. ;
tive pitch levels ... are not . . . important” 16. Talk of schedules of reinforcement
(25). No reference is made to the numerous here is entirely pointless. How are we to de-
studies of the function of relative pitch levels cide, for example, according to what schedules
and other intonational features in English. covert reinforcement is arranged, as in think-
11: The vagueness ot the word tendency, ing or verbal fantasy, or what the scheduling is
as opposed to frequency, saves the latter quo- of such factors as silence, speech, and appro-
tation from the obvious incorrectness of the priate future reactions to communicated infor-
former. Nevertheless, a good deal of stretching mation?
is necessary. If tendency has anything like its 46. E.g., what are in fact the actual units
ordinary meaning, the remark is clearly false. of verbal behavior? Under what conditions
One may believe strongly the assertion that will a physical event capture the attention (be
Jupiter has four moons, that many of Sopho- a stimulus) or be a reinforcer? How do we de-
cles’ plays have been irretrievably lost, that the cide what stimuli are in “control” in a specific
earth will burn to a crisp in ten million years, case? When are stimuli “similar”? And so on.
and so on, without experiencing the slightest (It is not interesting to. be told, e.g., that we
4. A Review of B. F. Skinner’s Verbal Behavior 63

say Stop to an automobile or billiard ball be- 48. There is nothing essentially mysteri-
cause they are sufficiently similar to reinforc- ous about this. Complex innate behavior pat-
ing people [46].) terns and innate “tendencies to learn in specific
The use of unanalyzed notions like similar ways” have been carefully studied in lower
and generalization is particularly disturbing, organisms. Many psychologists have been in-
since it indicates an apparent lack of interest in clined to believe that such biological structure
every significant aspect of the learning or the will not have an important effect on acquisi-
use of language in new situations. No one has tion of complex behavior in higher organisms,
ever doubted that in some sense, language is but I have not been able to find any serious
learned by generalization, or that novel utter- justification for this attitude. Some recent
ances and situations are in some way similar to studies have stressed the necessity for carefully
familiar ones. The only matter of serious inter- analyzing the strategies available to the or-
est is the specific “similarity.” Skinner has, ganism, regarded as a complex “information-
apparently, no interest in this. Keller and processing system” (cf. J. S. Bruner, J. J. Good-
Schoenfeld, op. cit., proceed to incorporate now, andG. A. Austin, A Study of Thinking
these notions (which they identify) into their [New York, 1956]; A. Newell, J. C. Shaw, and
Skinnerian “modern objective psychology” by H. A. Simon, “Elements of a Theory of Hu-
defining two stimuli to be similar when “we man Problem Solving,” Psych. Rev., 65
make the same sort of response to them” (124; [1958], 151-66), if anything significant is to be
but when are responses of the “same sort’’?). said about the character of human learning.
They do not seem to notice that this definition These may be largely innate, or developed by
converts their “principle of generalization” early learning processes about which very little
(116), under any reasonable interpretation of is yet known. (But see Harlow, “The Forma-
this, into a tautology. It is obvious that such a tion of Learning Sets,” Psych. Rev., 56 (1949),
definition will not be of much help in the study 51-65, and many later papers, where striking
of language learning or construction of new shifts in the character of learning are shown as
responses in appropriate situations. a result of early training; also D. O. Hebb,
47. “The Problem of Serial Order in Be- Organization of Behavior, 109 ff.). They are
havior,” in L. A. Jeffress, ed., Hixon Sympo- undoubtedly quite complex. Cf. Lenneberg,
sium on Cerebral Mechanisms in Behavior op. cit., and R. B. Lees, review of N. Chom-
(New York: John Wiley & Sons Inc., 1951). sky’s Syntactic Structures in Language, 33
Reprinted in F. A. Beach, D. O. Hebb, C. T. (1957), 406, for discussion of the topics men-
Morgan, H. W. Nissen, eds., The Neuropsy- tioned in this section.
chology of Lashley (New York: McGraw-Hill
Book Company, 1960). Page references are to
the latter.