The Psychologies of Structure,

Function, and Development

A. CHARLES CATANIA New York University

The nineteenth century closed with the promise of

an integrated science of psychology (Titchener,
1898). In the twentieth century, that promise has
yet to be fulfilled. Students of psychology still
are asked to choose theoretical sides. They see
functional accounts of operant behavior pitted
against ethological accounts of behavioral structure, analyses of reinforcement contingencies pitted
against theories of cognitive processing, and descriptions of language as verbal behavior pitted
against psycholinguistic formulations of language
competence. Behaviorism continues to clash with
phenomenology, and empiricism with nativism.
Psychologists are not yet even agreed on whether
theirs is a science of behavior or a science of
mental life.
The development of these controversies has been
described in terms of paradigm clash (e.g., Katahn
& Koplin, 1968; Neisser, 1972; Segal & Lachman,
1972), as if psychology were in the midst of the
kind of scientific revolution described by Kuhn
(1962). The student, whether his mentor be cognitive psychologist or behaviorist, is led to believe
that one or the other paradigm will emerge victorious from the confrontation of incompatible intellectual positions. But this characterization may be
misleading, because it is not clear that the controversies have grown out of incompatible treatments
of common problems. The present account argues
that the important dimensions of psychology are
different from those ordinarily considered when the
history of psychology is interpreted in terms of
paradigm clashes, and that these dimensions have
1

Preparation of this article was supported in part by

National Institutes of Health Grant MH-18506 to New
York University.
Requests for reprints should be sent to A. Charles
Catania, who is now at Department of Psychology, University of MarylandBaltimore County, 5401 Wilkens
Avenue, Baltimore, Maryland 21228.

434 MAY 1973 AMERICAN PSYCHOLOGIST

been emerging through evolution over the past century rather than through revolution in the past
decade.

Titchener's Psychologies
Let us return to Titchener's position at the turn
of the century (Titchener, 1898, 1899a, 1899b):
Psychology was a single science that, like biology,
contained lines of division. Biology included a
science of structure called morphology or anatomy,
a science of function called physiology, and a science of growth or development called embryology
or morphogenesis. By analogy, Titchener saw psychology divided into structural, functional, and developmental components. (He also noted a similar
division at the level of the analysis of species, which
included the sciences of taxonomy, bionomics or
ecology, and paleontology or evolutionism, and
even suggested the possibility that this type of
classification could be extended to the study of
cultures.) Titchener (1899a) described the divisions of psychology in the following way:
we see at once that the psychology of our definition is
(1) a structural psychology, an anatomy or morphology
of mind. Mind is a mass of tangled processes. Our problem is to dissect this complex, and to discover, if we can,
its plan of arrangement. But we may also regard mind . . .
as a system of functions. The mind "does" things for us,
or enables us to "do" things. We shall then have (2) a
functional psychology. And we may, further, discuss the
makeup and working of the child's mind, and the way in
which it passes over into the adult mind. We shall then
have a mental embryology. Our psychology has become
(3) the study of psychogenesis [pp. 21-22].

The present argument, simply, is that Titchener's

classification of psychological problems, appropriate
in his time, remains appropriate today. The critical point is that Titchener recognized the three
psychologies as complementary, not incompatible.
Titchener (1899a) went on to say: "No one of

these three psychologies is 'better' psychology

psychology in a more real sense of the wordthan
any other [p. 22]." Psychologists, however, perhaps including Titchener, seem not to have taken
sufficient note of the statement. The present account will take it as a point of departure. In
elaborating on it, we will first concentrate on the
distinction between structural and functional psychologies, because this distinction seems to lie at
the root of the conflict between cognitive and behavioral psychologies. We will later have occasion
to consider the place of psychogenesis.

Structure and Function

In biology, the distinction between structure and
function was so well established that it supported a
division of the field into such separate departments
as anatomy and physiology. The line between
anatomical research and physiological research was
sometimes difficult to draw, and it remains so today. But it is at least clearly recognized that
studies of biological structure and studies of biological function are concerned with different empirical questions. To say what an organ does, it
may help to know how it is constructed; yet its
function is not studied in the same way as its
structure.
While the analogous distinction is made in psychology, it is sometimes difficult, as in biology, to
draw the line. Nevertheless, there exist some
bodies of research predominantly concerned with
analyzing stimulus structure or response structure,
and other bodies of research predominantly concerned with analyzing behavioral function. An
analysis of the distinctive features of a stimulus,
for example, is concerned with a different problem
than the analysis of the conditions that motivate
the organism's continued attention to those distinctive features.
The relation between these different kinds of
psychological problems has typically been obscured
by different languages of psychology. These differences have developed as consequences of historical accident, the influence of everyday discourse on
technical vocabulary, and the various research
strategies that are appropriate to specific experimental issues. Structural research tends to be described in the cognitive or mentalist vocabulary,
and functional research in the behaviorist vocabulary. But the correlation between these research

concerns and these vocabularies is not a necessary

one. A cognitive psychologist can be concerned
with functional problems (e.g., in distinguishing
between parallel and serial processes, cf. Sternberg,
1970) just as a behavioral psychologist can be
concerned with structural ones (e.g., in analyzing
the stimulus dimensions to which an organism responds in a color-matching task, cf. Wright & Gumming, 1971). In fact, a major argument of the
present account is that psychological controversy
has often originated because the dichotomy between
structure and function has been confused with that
between mentalism and behaviorism. (An instructive comparison is with the history of anatomy and
physiology in biology, in which the respective concerns with structure and with function were not
so highly correlated with vitalistic and mechanistic
positions.)
The point demands a concrete example, and one
has been chosen that may at first seem frivolous:
consider what a psychologist might do if he were
interested in analyzing baseball pitching. He might
begin by concerning himself with the coordination
of the pitch. Through slow-motion photography
or electromyographic recording, he could examine
the sequential patterning of muscle movements, and
he could analyze the relation between the early
and the late parts of the performance. He might
be able to specify the interaction between the ball's
speed and trajectory as it leaves the pitcher's hand
and the magnitude and form of the pitcher's follow-through. He might even venture a mathematical formulation in which fast balls and curves
were distinguished by parameters of his equations,
and his analysis could conceivably provide clues
about what to emphasize in giving instruction to
a novice pitcher. If he were reasonably successful,
to the point, for example, that he could predict the
properties of a given pitch from the early components of muscle activity, he might be tempted to
claim that an exhaustive account of the critical features of pitching was realizable at least in principle
if not attainable in practice. Yet no matter how
exhaustive his analysis of the structure of the pitch,
this psychologist would not be in a position to
deal with the circumstances that determine when
and at what the pitcher throws the ball. Such an
account would require, instead, an analysis of the
functional properties of the pitch: its relation to
antecedent stimulus conditions and subsequent
stimulus consequences.
AMERICAN PSYCHOLOGIST MAY 1973 43S

The example may seem trivial. Yet it is precisely

paralleled by contemporary developments in the
psychology of language and illustrates the different
strategies that have led to conflict between cognitive and behavioral formulations. Studies of both
grammar and phonology (e.g., Chomsky & Miller,
1963; Liberman, 1970) have dealt specifically with
the structure of language and speech. Transformational analyses have been concerned with the
complex coordinations necessary to generate grammatical sentences or comprehensible phonetic utterances. Recognizing that the way a sentence or
utterance ends interacts with characteristics of the
earlier parts of the sentence or utterance is not so
very different from recognizing that the followthrough interacts with the windup and the delivery
even though it occurs after the ball has left the
pitcher's hand. Such analyses can be elaborated at
various levels of complexity; for example, the hierarchical organization of units such as phonemes,
words, and grammatical forms is explicitly featured in accounts of grammatical structure.
Structural analyses of grammar and speech, however, cannot tell us when a person will decide to
speak, or what he will talk about. It is precisely
these latter questions that are the concern of a
functional analysis of language. This is illustrated
even by some of those materials that are taken as
critical examples of the primacy of a structural account. The written sentence "Dropping bombs can
be dangerous" has one of two structures, depending
on whether the speaker is concerned with the
people in the air or those on the ground. There is
no debate about whether these five words in this
particular order can constitute two different sentences. But the structural account cannot help us
to choose between the two structures. The choice
must be based on functional considerations: the
conditions under which the sentence is generated.
But say instead that the words are spoken. Now
the two sentences can be distinguished by different
patterns of stress. Once both phonological and
grammatical analyses are available, an account can
be given of the relation between the two types of
structure. The problem is not surmounted, however ; even at this point, the structural account cannot tell us about the circumstances under which a
sentence with one or the other structure will be
uttered. That issue is again functional.
It is surprising that such relations are so often
overlooked. Yet failures to note them can easily
436 MAY 1973 AMERICAN PSYCHOLOGIST

be documented. The controversy over the psycholinguistic account of grammatical structure versus
the functional analysis of verbal behavior has been
both persistent and prominent in psychology. The
chronology includes, among others, Skinner (1957),
Chomsky (1959), Lenneberg (1967), Dixon and
Horton (1968), MacCorquodale (1970), and
Premack (1970). The two sides of the controversy
were at most times simply concerned with two different kinds of problems: problems of structure and
problems of function (or, equivalently, problems of
competence and problems of performance, cf. McNeill, 1970, p. 146). But the accounts were
couched in languages and contexts that were sufficiently different that the different problems each
was addressing typically went unrecognized (cf.
Catania, 1972).

A Behavior Paradigm
Although we are arguing that the development of
contemporary psychology is not properly interpreted in terms of paradigm clashes, paradigms can
be useful. We may recall that a paradigm is a
model that exhibits essential relations among the
phenomena that it represents. We should not be
surprised if neither a cognitive nor a behavioral
psychologist could come up with a paradigm on
which his respective colleagues could universally
agree. Nevertheless, we shall introduce a paradigm
here to illustrate some of the properties of structural and functional accounts.
The paradigm takes the form S'^RiS 0 ), where
S represents a stimulus and R represents a response. The superscripts, in SD and S, distinguish
between two kinds of stimuli: a discriminative stimulus, SD, which is a stimulus denned in terms of
the events that can occur in its presence, and a
contingent stimulus, Sr, which is a stimulus defined
in terms of its consequential relation to responses.
The expression (R:S) represents the relation of
responses to consequences. This relation is called
a contingency and can be translated, "the effect of
response R on the probability of stimulus Sc."
Thus, the paradigm as a whole represents a contingency that operates in the presence of a discriminative stimulus. The paradigm is not exhaustive; it does not include, for example, responses
that may be elicited by the contingent stimulus. A
more detailed account of the paradigm has been
presented elsewhere (Catania, 1971); the point of

using it here is not to argue the precedence of a

behavioral terminology, but rather to illustrate
some structural and functional dimensions of problems in psychology.
Let us consider the scope of the paradigm. The
contingency represents the way in which behavior
can affect environmental events. If the contingent
stimulus is a food pellet delivered to a hungry rat,
for example, the rat's various responses may make
the delivery of a pellet more likely or less likely,
or might have no effect on pellet delivery. Let us
say that the lever press makes the pellet delivery
more likely, or, in other words, that the lever press
produces a pellet; we call this an instance of positive reinforcement. But if the lever press makes
the pellet delivery less likely, or, in other words,
prevents the pellet delivery, we speak of the procedure as omission training. Finally, if the lever
press has no effect on the likelihood of a pellet delivery, we speak of pellet deliveries as response
independent.
But each of these contingencies might operate
only in the presence of some discriminative stimulus, such as a light. Such cases provide various
examples of discrimination procedures. If, in the
presence of a light, lever presses produce food pellets, we speak of an operant discrimination. If, in
the presence of a light, lever presses prevent the
delivery of food pellets, we speak of discriminated
omission training. And if, in the presence of a
light, food pellets are delivered independently of
lever presses, we speak of a respondent or Pavlovian
conditioning procedure: in the presence of a light,
food pellets are delivered, just as food was delivered
in the presence of various stimuli to Pavlov's dogs.
The contingent stimulus, however, can be aversive instead of appetitive. If the contingent stimulus were electric shock, for example, the three cases
would include punishment, when lever presses produce shock; avoidance or escape, when lever presses
prevent or remove shock; and response-independent
shock, when lever presses have no effect on the
likelihood of shock delivery. These contingencies,
too, can operate in the presence of a discriminative
stimulus. In such cases, we classify the procedures
as discriminated punishment, discriminated avoidance or escape, and respondent defensive conditioning, respectively.
Finally, the contingent stimulus can be effectively neutral. If a rat's lever press produces or
removes a stimulus such as a click or a light, we

can ask about the extent to which the rat learned

about the contingency relation between its lever
press and this stimulus by later pairing the stimulus
with some appetitive or aversive event. We speak
of learning so demonstrated as latent learning. But
relations between responses and such simple events
typically occur in certain settings, and so it is
appropriate to speak of discriminative stimuli for
these contingencies also. In the presence of certain
visual, tactile, and other stimuli, the rat moves
about and encounters various parts of its environment: its movements are responses that have certain consequences, and these relations are also examples of contingencies. Thus, procedures in
sensorimotor learning can be encompassed by
such a paradigm. There may be, in addition, some
neutral events that reliably occur independently of
the rat's responses. Suppose, for example, that in
the presence of a tone, a light consistently flashes;
if we find, through later procedures in which we
pair the tone or the light with other events, that
the rat had learned something about the relation
between these two stimuli, we refer to the initial
procedure as an instance of sensory preconditioning.
Thus, all of the basic learning procedures can
be incorporated into this paradigm. The essential
feature of the analysis of contingencies is in fact
the description of the functional relations among
stimuli and responses. Some stimuli, which we call
discriminative, are stimuli denned in terms of the
events that can occur in their presence. Other
stimuli, which we call contingent, are stimuli defined in terms of their consequential relation to
responses. Our interest in behavior encompasses
both the circumstances under which responses can
occur and the ways in which they can modify the
likelihood of environmental events.

Structural and Functional

Research Strategies
Consider how we proceed if we are interested in
functional questions. We can study various contingencies, (R:S C ), by varying a response's effect
on the probability of some contingent stimulus or
by varying the nature of the contingent stimulus.
For example, when we study the transition from
reinforcement to extinction, we change from a contingency in which a response produces a reinforcing
stimulus to one in which the response no longer
produces that reinforcing stimulus; when we study
AMERICAN PSYCHOLOGIST MAY 1973 437

the effects of satiation, we change from a procedure

in which the contingent stimulus is reinforcing to
one in which the contingent stimulus is no longer
reinforcing. Or we can study the stimulus control
acquired by a discriminative stimulus, SD, by varying the contingencies on which the stimulus is
superimposed or by changing the relation of the
stimulus to a contingency. For example, when we
compare the roles of stimuli in operant and respondent conditioning, we change the contingency
that operates in the presence of a discriminative
stimulus from one in which a response produces a
contingent stimulus to one in which a response has
no effect on presentations of a contingent stimulus;
when we examine stimulus control in complex reinforcement schedules or in types of delay conditioning, we vary the temporal separation between the
discriminative stimulus and the operation of a contingency. (We could now transfer this account to
the baseball pitcher of our earlier example. We
could argue that a given batter at a particular point
in a game represents a discriminative stimulus, SD,
in the presence of which a particular pitch, R, will
be likely to have a particular consequence, S. The
translation is simple enough. To illustrate how
the paradigm bears on structural and functional
questions, however, we might better apply it to
more traditional psychological materials.)
Each of the functional questions in the preceding
examples concerns relations among various terms
in the paradigm. Let us now consider structural
analyses, which deal with the distinguishing properties of individual terms when the relation among
terms is held constant. For example, if a particular
contingency operates in the presence of a particular
discriminative stimulus, we can vary the stimulus
and ask questions about its critical features. This
is our strategy in psychophysical experiments with
both human and animal subjects, but it is also
applied when the stimulus features are more complex. Thus, in a study of how children learn to
read, we can try to identify properties of the letters
of the alphabet that are critical to learning, and
we might find, among the features that we examine,
that the child can be taught to distinguish between
up-down reversals of letters more easily than between left-right reversals (cf. Gibson, 1965). The
questions are about stimulus structure, or, as
Pylyshyn (1972) put it, about "the structure of
the percept":
438 MAY 1973 AMEKICAN PSYCHOLOGIST

The question of the nature of a mental representation

is really a question about which aspects of the pattern of
incoming data are perceived, learned, and retained for
potential future use. However, . . . these aspects that
are distinguished and retained and that play a functional
role in cognition need not be simple classes of physical
properties of the stimulus pattern. They often are rather
abstract properties whose relation to the physical features
of the stimulus may be quite obscure . . . [p. 548].

The account is reminiscent of the distinction between sensation and perception, and we may recall
the extent to which the field of perception has involved structural concerns.
On the other hand, we might be interested in
studying differences among the classes of responses
that can be learned. Again, the relations among
terms of the paradigm are held constant, but this
time we vary the response rather than the discriminative stimulus. We could examine simple response
properties, such as the force or topography of a
rat's lever press, or we could study the differentiation of motor skills in human subjects. Similar
questions, however, can be addressed to more complex modes of responding. In the four decades between Guilford's (1927) "The Role of Form in
Learning" and Johnson's (1968) "The Influence of
Grammatical Units on Learning," our experimental
sophistication has changed, but the basic problem
has remained the same: at issue is the structure of
complex responses. A number of areas in the contemporary psychology of human verbal learning
(e.g., subjective organization in free recall; Tulving,
1962) can be regarded as concerned with the structural properties of complex responses. (To extend
the account to contingent stimuli, we might even
argue that concern in motivation with the factors
that influence the effectiveness of contingent stimuli, as in the analysis of incentives, is a structural
problem in the present sense.)

Hierarchical Organization of
Stimuli and Responses
Some might argue that this kind of an account
misses the point, because an analysis of stimulus
structure alone or of response structure alone will
necessarily omit the complex interaction between
organism and environment that must take place
during cognitive processing (cf. Neisser, 1967).
But each stage of such an interaction must involve
the organism's responses to particular features of
the environment. The resolution, therefore, may

lie with an account of the hierarchical organization

of stimuli and responses. Estes (1971) stated the
issue as follows:
If one who is attempting to describe and predict the
behavior of an adult human learner fails to take account
of these behavioral organizations, and attempts to construct an account in terms only of individual stimulusresponse units, the principles of operation of rewards and
punishments may appear to be quite different from those
revealed in simpler experiments with animals or immature
human learners. Actually, it may be that the principles
of operation of these factors are the same in all cases and
that the difference lies in the nature of the behavioral
units whose probabilities are being modified as a result of
the experiments with various types of outcomes [p. 23].
The tendency to select one response strategy rather than
another in a given situation must itself be modified by
past experience with rewarding or punishing outcomes. A
strategic question which must be fundamental to the further development of theory in this area is that of whether
the laws and mechanisms of reinforcement are the same
for these higher-order behavioral units as for the more
elementary responses studied in most laboratory experiments [p. 31].

We may too often think of responses as single

and discrete events, failing to regard their structural properties as also included among their defining features. (We have similar problems when
we talk about stimuli when we mean to talk about
stimulus properties.) Not simply the closure of a
switch by a lever press, but perhaps also the grammaticality of a sentence that is uttered or the appropriateness of a strategy that is applied can be
regarded as denning properties of those response
classes that can enter into functional relations.
Take Harlow's (1949) learning set experiment
as an example. At one level of analysis, the choice
of one or another stimulus is the response of
interest. But at a hierarchically more complex
level, the acquisition of learning set is denned in
terms of the substitution, for many separate choices
in different problems, of a single, more general response class that can be described in the following
terms: if the stimulus choice produced food this
time, continue to choose it; if it did not, choose
the other stimulus; more economically, we may
speak of "win-stay, lose-shift." It is fruitless to
debate whether this performance should be spoken
of in terms of a complex response class or in terms
of a strategy; in either case, the point is that the
analysis concerns behavioral structure. Questions
about the nature of this structure are orthogonal to
functional questions, such as that of whether rein-

forcement was a learning or a performance variable

in these procedures. (This does not imply that an
analysis of structure will necessarily be irrelevant
to an analysis of function, or vice versa; if structure and function interact in learning, for example,
it is all the more important to be clear about the
difference between structural and functional
questions.)
The following account (Fischer, 1972) of maze
learning provides another illustration:
After . . . familiarizing yourself with one of the more
complete structural systemsperhaps Chomsky's grammar
you might even be able to begin to write that structural
analysis of children's maze performance. While other psychologists were explaining the performance by reducing it
to some lowest common factor like reinforcement or attention, you would be searching for key patterns and for
rules that would relate those patterns to each other. In
a way, it would be like writing a grammar for maze performance. You would need a set of grammatical categories
to describe the "phrase structure" of the children's behavior
in the maze and a set of transformation rules for relating
different types of phrase structures. With this kind of
structural description, you could account for the particulars of maze performance to an extent that is probably
impossible for explanations that merely reduce behavior to
factors like reinforcement and attention [pp. 330-3311.

According to this account, Krechevsky (1932)

was writing a preliminary grammar of maze learning when he carried out his experiments on hypotheses in rats. And this formulation is reasonable,
because Krechevsky's concern, like Fischer's, was
with the structure of the maze performance. Yet
neither the child nor the rat will necessarily move
through the maze, even if the maze has been
thoroughly mastered, on the basis of a maze grammar alone. We may recall Tolman's (1948) rat,
said to be lost in thought at the choice point. The
cognitive map provided a way of describing the
effective stimulus structure of the rat's world. But
a rat with a cognitive map must have occasion to
use it, and procedures that get the rat to perform
and thereby demonstrate its map are functional. In
psycholinguistics, we can similarly imagine Chomsky's speaker lost in thought, capable of grammatical speech but with nothing to say:
In the case of linguistic concepts, such as grammaticality,
such procedures are called generative grammars. Such
grammars do not describe how people go about understanding or generating sentences, but they do describe the
abstract relation that holds between strings of words and
such concepts as "grammatical sentence" [Pylyshyn, 1972,
p. 550].

AMERICAN PSYCHOLOGIST MAY 1973 439

If we wish to understand how the thinker comes to

speak, we must be able to provide a functional as
well as a structural account of language.

Explanation and Description

Let us suppose, however, that we had provided
both a structural and a functional account of language or of maze learning. To what extent would
we have explained the performance? In his quotation above, Fischer contrasted functional factors
like reinforcement and attention with structural
factors like transformation rules as ways of explaining or accounting for the maze performance. But
we may question whether either a structural or a
functional account, or even both in combination,
can ever have the power of explanation. The present view is that both structural and functional
analyses are descriptive rather than explanatory.
Consider the following commentary by Black
(1970) on the psycholinguistic account of language in terms of deep structures and generative
grammars:
Whatever their value, Kepler's laws do not explain the
planetary motions, in any useful sense of "explain": they
replace a crude and unsystematic description ("those orbits
out there", or "the orbits conforming to these readings")
by another description concisely presenting some mathematical properties of the orbits. The same applies, mutatis
mutandis, to the rules constituting a specific generative
grammar. Our initial crude "intuitions" as to what should
count as grammatical or the reverse are replaced by a set
of precise and explicit rules that (approximately and with
idealization) generate a corresponding classification. This
provides valuable insight into structural connections: it
may be said to provide intelligible reasons for what we
previously seemed to be doing by a kind of instinct. But
"explanation" hardly seems the right tag [pp. 454455].

Functional terms like reinforcement, too, have

typically been regarded as explanatory. Yet reinforcement is simply a name for a certain set of
functional relations. If a response in a given class
produces a stimulus, and responses in that class are
strengthened by virtue of its production of that
stimulus, we say that we have observed an instance
of the process of reinforcement. We may wish to
relate this process to other behavioral processes,
or to characterize the properties of stimuli that can
have this reinforcing effect, but that does not
change the nature of the term. It remains a name,
to be applied when it is appropriate like any other
name (Catania, 1973). Thus, we can talk about
440 MAY 1973 AMERICAN PSYCHOLOGIST

stimulus properties and about response properties

at various hierarchical levels of complexity, and we
can outline the kinds of functional relations that
can exist among them. At each level, our concern
is descriptive, and the question of what kinds of
functional relations different stimuli or responses
can enter into is an empirical one.
Much emphasis has recently been given to the
limits of learning (e.g., Breland & Breland, 1961;
Revusky & Garcia, 1970; Rozin & Kalat, 1971;
Seligman, 1970): limits on the kinds of responses
that can be learned, on the kinds of stimuli that
can act effectively on an organism, and on the kinds
of relations that can be established between particular stimuli and particular responses. But
analyses of the structure of effective stimuli, or of
the different structures of those responses that can
be more or less easily learned, are different from
analyses of how stimuli and responses function in
behavior. And if some functional relations turn
out to be less general than was once believed, it
does not follow that their names should be changed
or that they no longer have the status of functional
relations.

Distinguishing the Structure-Function

Dimension from the BehaviorCognition Dimension
According to the present view, questions of stimulus structure and response structure are orthogonally related to questions of behavioral function.
This is not to say that information about structure
will never bear on the analysis of function, or vice
versa. But to the extent that these relations can
be clarified to show that various research areas in
psychology complement rather than conflict with
each other, controversy may give way to more productive interaction. For example, applications of
psychology to education are often divided between
those concerned with the cognitive organization of
the subject matter and those concerned with behavioral methodology in classroom management.
Yet to teach effectively, it is essential to know
both how a subject matter is structured, as stimulus
in the teacher's presentation and as response in
the student's mastery, and how the teacher and
the student can function in a classroom. We must
give attention to cognition in both structural and
functional senses: structurally in the extent to
which cognitive strategies can be regarded as ways

of describing the structural properties of complex

stimuli, and functionally in the analysis of both
the circumstances under which strategies are applied and the consequences of these strategies. At
issue is the question of whether cognitive and
behavioral approaches are inappropriately interpreted, respectively, as structural and functional
psychologies.
The evolution of psychology over the past century has included the successive elaboration of
various structural and functional psychologies.
The properties of these different psychologies have
changed in detail over the years. Gestalt psychology was clearly structural, just as the analysis of
contingencies in schedules of reinforcement was
clearly functional. But structural and functional
psychologies complement each other; they need not
stand in opposition.
Some of the differences have been differences in
language. The clash between the vocabulary of
mentalism and the vocabulary of behaviorism is
readily evident. It is not unfashionable these
days to be a mentalist; only dualism is reprehen'sible. But the debate between mentalism and behaviorism has been along different dimensions than
those of structure and function. The major behaviorist argument has been against causal, not
descriptive, mentalism, and to argue that mental
events are not causes of behavior is not to argue
that private events do not exist. In fact, the possibility of an internally consistent mentalism is
implicit in the notion that a behavioral translation
of mental or cognitive vocabularies is feasible.
The choice of vocabularies will rest with the
effectiveness with which their proponents apply
them. On the side of the mental or cognitive
vocabulary, it might be argued that it lies closer
to everyday talk than some behaviorist languages,
but on examination, the language of information
storage and retrieval, for example, is no less esoteric than the language of reinforcement contingencies. On the side of the behaviorist vocabulary,
it might be argued that it provides greater precision
and less risk of contamination by everyday preconceptions, but behaviorists have yet to discount
the possibility of an internally consistent and effective mentalist vocabulary (cf. Bolles, 1967).
As the arguments are marshaled on both sides, the
tide shifts: epistemological difficulties in behaviorism are balanced by accounts of its relevance to
phenomenology (Day, 1969), while the expanding

front of cognitive research is tempered by its

philosophical limitations (Malcolm, 1971). These
clashes, in any case, may be mere epiphenomena;
the distinction between structure and function is
critical to either viewpoint, and progress will depend more on results than on discussion.

The Place of Psychogenesis

Let us now turn back to Titchener's formulation.
We have seen how various problems of psychology
can be interpreted as problems of structure or of
function. It remains to relate these two subdivisions to problems of development: what Titchener
called psychogenesis. Superficially, it is tempting
to equate this area with developmental psychology,
and in one sense such an equation would be correct. Yet to place the area in its proper perspective, we must be clear about the kind of development that is to be studied. Certainly, we may be
interested in the development, through maturation,
of the capacity to respond to various complexes of
stimuli or to engage in new responses and apply
different strategies. But whether we say we are
concerned with the development of behavior or
with that of the mind, our psychogenesis must include the psychology of learning. If embryology
deals with the development of the organism, and
evolution with the development of a species, then
psychogenesis must deal with the development of
the organism's behavioral or mental capacities; this
is what the psychology of learning is supposed to
be about.
The view resolves some paradoxes. In the analysis of behavior, the functional analysis of contingencies, as in the study of reinforcement schedules (Ferster & Skinner, 1957), created conflict.
The operant analysis of steady-state performances
was not seen as relevant to some problems in the
psychology of learning. This was as it should have
been: on the one hand, the analysis was more concerned with functional relations in the steady-state
maintenance of behavior than with developmental
relations in its acquisition; on the other, it might
have been anticipated that the study of development, too, would divide into functional and structural components (e.g., respectively, Bijou & Baer,
1966; Piaget & Inhelder, 1969).
According to the present view, it was appropriate
for these various parts of psychology to proceed
in different ways. It is only unfortunate that they
AMERICAN PSYCHOLOGIST MAY 1973 441

have often been seen as incompatible. Progress in

these areasstructure, function, and development
has constituted the evolution in psychology over
the past century. It has typically been uncoordinated, and there is much left to be settled. The
detailed character of each area changes from time
to time. The areas may variously merge and
separate again, but they are at least as independently viable as are divisions of the biological
sciences. Hopefully, they will not become totally
isolated from each other.

Paradigms and Clashes

We have argued here that psychology is not in the
midst of a paradigm clash. True paradigm clashes
require points of contact. In the clash between
the astronomies of Kepler and of Ptolemy, the subject matter stayed the same while the point of view
changed; literally, as well as figuratively, the center
did not hold. Different schools of psychology, however, have been concerned with different problems.
This is not to say that the evolution of psychological concepts has never involved conflict. Rather,
it is to say that the clashes have not been along
any simple dimension of psychological problems.
The multiplicity of issues may generally distinguish
clashes in the life sciences from those in the physical sciences. A strong case for this view has been
made in Mayr's (1972) analysis of the Darwinian
revolution:
It is now evident that the Darwinian revolution does
not conform to the simple model of a scientific revolution
as described, for instance, by T. S. Kuhn. . . . It is actually
a complex movement that started nearly 250 years ago;
its many major components were proposed at different
times, and became victorious independently of each other.
Even though a revolutionary climax occurred unquestionably in 1859, the gradual acceptance of evolutionism, with
all of its ramifications, covered a period of nearly 250
years . . . [p. 988].

Neither is the revolution in psychology a revolution

of a decade.
We have argued that Titchener's insight into the
organization of psychology is relevant today. Yet
we may wonder why Titchener, in his structural
account, contributed to an opposition between structuralism and functionalism that has persisted for
most of this century. Part of the answer probably
lies in what it was possible to do in each of these
separate areas in Titchener's time. As Titchener
442 MAY 1973 AMERICAN PSYCHOLOGIST

(1899a, p. 22) noted, structural psychology had

made more progress in his day than either of the
other two psychologies, But it is the progress in
structural, functional, and developmental accounts
that constitutes psychology's evolution and revolution. It is reassuring to note that Boring (1969)
detected signs that Titchener, near the end of his
career, was gradually moving toward behaviorism;
Boring's grounds were simply that Titchener's
theory of meaning was changing into a functional
account. And from that development, too, both
structuralists and functionalists can learn.
REFERENCES
BIJOU, S. W., & BAER, D. M. Operant methods in child
behavior and development. In W. K. Honig (Ed.),
Operant behavior: Areas of research and application.
New York: Appleton-Century-Crofts, 1966.
BLACK, M. Comment. In R. Borger & F. Cioffi (Eds.),
Explanation in the behavioural sciences. Cambridge:
Cambridge University Press, 1970.
BOLLES, R. C. The syntax of the mind. Psychological
Reports, 1967, 21, 493^99.
BORING, E. G. Titchener, meaning and behaviorism. In
D. L. Krantz (Ed.), Schools of psychology: A symposium. New York: Appleton-Century-Crofts, 1969.
BRELAND, K., & BRELAND, M. The misbehavior of organisms. American Psychologist, 1961, 16, 681-684.
CATANIA, A. C. Elicitation, reinforcement, and stimulus
control. In R. Glaser (Ed.), The nature of learning.
New York: Academic Press, 1971.
CATANIA, A. C. Chomsky's formal analysis of natural
languages: A behavioral translation. Behaviorism, 1972,
1, 1-15.
CATANIA, A. C. The nature of learning. In J. A. Nevin
& G. S. Reynolds (Eds.), The study of behavior: Learning, motivation, emotion, and instinct. Glenview, 111.:
Scott, Foresman, 1973.
CHOMSKY, N. Review of B. F. Skinner's Verbal behavior.
Language, 1959, 35, 26-58.
CHOMSKY, N., & MILLER, G. A. Introduction to the formal
analysis of natural languages. In R. D. Luce, R. R.
Bush, & E. Galanter (Eds.), Handbook of mathematical
psychology. Vol. 2. New York: Wiley, 1963.
DAY, W. F. Radical behaviorism in reconciliation with
phenomenology. Journal of the Experimental Analysis
of Behavior, 1969, 12, 315-329.
DIXOM, T. R., & MORTON, D. L. (Eds.) Verbal behavior
and general behavior theory. Englewood Cliffs, N.J.:
Prentice-Hall, 1968.
ESTES, W. K. Reward in human learning: Theoretical issues and strategic choice points. In R. Glaser (Ed.),
The nature of reinforcement.
New York: Academic
Press, 1971.
FERSTER, C. B., & SKINNER, B. F. Schedules of reinforcement. New York: Appleton-Century-Crofts, 1957.
FISCHER, K. W. Structuralism for psychologists. Review
of M. Lane (Ed.), Introduction to structuralism. Contemporary Psychology, 1972, 17, 329-331.
GIBSON, E. J. Learning to read. Science, 1965, 148, 10661072.

GUILFORD, J. P. The role of form in learning. Journal of

Experimental Psychology, 1927, 10, 415-423.
HARLOW, H. F. The formation of learning sets. Psychological Review, 1949, 56, S1-6S.
JOHNSON, N. F. The influence of grammatical units on
learning. Journal of Verbal Learning and Verbal Behavior, 1968, 7, 236-240.
KATAHN, M., & KOPLIN, J. H. Paradigm clash: Comment
on "Some recent criticisms of behaviorism and learning
theory with special reference to Breger and McGaugh
and to Chomsky." Psychological Bulletin, 1968, 69,
147-148.
KRECHEVSKY, I. "Hypotheses" in rats. Psychological Review, 1932, 39, 516-532.
KUHN, T. S. The structure of scientific revolutions. Chicago: University of Chicago Press, 1962.
LENNEBERG, E. H. Biological foundations of language.
New York: Wiley, 1967.
LIBERMAN, A. M. The grammars of speech and language.
Cognitive Psychology, 1970, 1, 301-323.
MACCORQUODALE, K. On Chomsky's review of Skinner's
Verbal behavior. Journal of the Experimental Analysis
of Behavior, 1970, 13, 83-99.
MALCOLM, N. The myth of cognitive processes and structures. In T. Mischel (Ed.), Cognitive development and
epistemology. New York: Academic Press, 1971.
MAYR, E. The nature of the Darwinian revolution. Science, 1972, 176, 981-989.
McNEiLL, D. The acquisition of language. New York:
Harper & Row, 1970.
NEISSER, U. Cognitive psychology. New York: AppletonCentury-Crofts, 1967.
NEISSER, U. A paradigm shift in psychology. Review of
A. Richardson, Mental imagery; S. J. Segal (Ed.),
Imagery: Current cognitive approaches; and A. Paivio,
Imagery and verbal processes. Science, 1972, 176, 628630.

PIAGET, J., & INHELDER, B. The psychology of the child.

(Trans, by H. Weaver) New York: Basic Books, 1969.
PREMACK, D. A functional analysis of language. Journal
of the Experimental Analysis of Behavior, 1970, 14, 107125.
PYLYSHYN, Z. W. Competence and psychological reality.
American Psychologist, 1972, 27, 546-552.
REVUSKY, S., & GARCIA, J. Learned associations over long
delays. In G. H. Bower (Ed.), The psychology of
learning and motivation. Vol. 4. New York: Academic
Press, 1970.
ROZIN, P., & KALAT, J. W. Specific hungers and poison
avoidance as adaptive specializations in learning. Psychological Review, 1971, 78, 459-486.
SEGAL, E. M., & LACHMAN, R. Complex behavior or higher
mental process: Is there a paradigm shift? American
Psychologist, 1972, 27, 46-55.
SELIGMAN, M. E. P. On the generality of the laws of
learning. Psychological Review, 1970, 77, 406-418.
SKINNER, B. F, Verbal behavior. New York: AppletonCentury-Crofts, 1957.
STERNBERG, S. Memory-scanning: Mental processes revealed by reaction-time experiments. In J. S. Antrobus
(Ed.), Cognition and affect.
Boston: Little, Brown,
1970.
TITCHENER, E. B. The postulates of a structural psychology. Philosophical Review, 1898, 7, 449-465.
TITCHENER, E. B. An outline of psychology. New York:
Macmillan, 1899. (a)
TITCHENER, E. B. Structural and functional psychology.
Philosophical Review, 1899, 8, 290-299. (b)
TOLMAN, E. C. Cognitive maps in rats and men. Psychological Review, 1948, 55, 189-208.
TULVING, E. Subjective organization in free recall of "unrelated" words. Psychological Review, 1962, 69, 344-354.
WRIGHT, A. A., & CUMMING, W. W. Color-naming functions for the pigeon. Journal of the Experimental Analysis of Behavior, 1971, 15, 7-17.

AMERICAN PSYCHOLOGIST MAY 1973 443