Review of Theories of Perception and the Concept of Structure by F.H. Allport
Benbow F. Ritchie
Theories of Perception and the Concept of Structure: A Review and a Critical Analysis With an Introduction to a Dynamic-Structural Theory of Behavior. By FLOYD H. ALLPORT. New York, John Wiley and Sons, 1955. Pp. xxii, 709.
Two purposes guided the author in this book: first, to provide a critical survey of theories about the nature of perception, and, second, to present his views about the problems which such theories should resolve.
In the first three chapters, Allport discusses the proper function of theories in psychology and in doing so covers familiar ground—how define "perception"? what status have sense-data? what are the criteria of objectivity? what is the logical nature of explanation? and so on. The number of philosophical questions raised is of course considerably greater than the number settled or even clarified. The chapter does enable Allport to give us an idea of his epistemological premises, and though these are matters which seem to have little bearing on the behavior of most psychologists, it seems a useful thing to know how a writer on perceptual theories stands with respect to them.
Next, Allport surveys a variety of theories about the nature of perception: the classical core-context theory, the gestalt theory, Lewin's topological theory, Hebb's cell-assembly theory, the Werner-Wapner sensory-tonic theory, Freeman's set theory, Helson's adaptation-level theory, Brunswik's functionalist theory, the Bruner-Postman hypothesis theory, and the cybernetic theories. These chapters constitute about 700/c of the book and they will, I believe, be the most interesting to the general reader, since they provide a lucid and interesting survey of many ideas important to an understanding of perception.
In the concluding three chapters, Allport considers "the unsolved problem of meaning" and develops his concept of "event-structure." The various philosophical issues that enter this discussion are so numerous and so difficult, that a brief and accurate report of his opinions is out of the question. Instead I have chosen to say what I think he means by "the unsolved problem of meaning" and to explain how he thinks the concept of "event-structure" will help to solve it.
The proper analysis of the relation between non-verbal signs and what they signify is the fundamental problem which Allport regards as unsolved, but, before we can see this problem clearly, several semiotical distinctions are essential. We must understand first that the term "meaning" refers to a particular kind of relation among three terms, a sign, an interpreter of the sign, and a significate of the sign. The characteristic which distinguishes this triadic relation from all others is that one of the terms, the sign, has the property of being either true or false, and it is this characteristic of signs which provides the core of the psychological problem of meaning. The task of the psychologist is to provide an analysis of this complex relation in the language of psychology to the end that, given this analysis, we can discover empirical ways of investigating the sign-behavior of animals and men.
Now, in such an analysis, the sign will be identified as a special kind of stimulus, and the interpreter as an animal in a special condition (e.g. disposed to react in certain ways), but how can we identify the significate? When the sign is true, the significate could perhaps be identified by that thing or situation which makes the sign true. But what about when the sign is false? This case makes it clear that the significate cannot be regarded as a particular, but rather must be considered as a class of a certain kind. Consider the sentence "There is a sunflower in bloom at the North Pole." This sign signifies the class of blooming sunflowers in a certain geographic region. If the class has no members (i.e. is empty or null), the sign is false. If, however, there is at least one member of the class, the sign is true, and is said to denote that member. The same analysis can, of course, be made for any non-verbal sign, like a bell or light in a conditioning experiment, or a patterned card in a discriminative problem. This analysis enables us to make one additional important distinction. Classes can be identified and distinguished from each other in two ways. We may consider a class in extension, that is, in terms of its members, or in intension, in terms of the requirements that must be met if something is to be a member of the class. The class of American Nobel-Prize-winners-in-literature consists of Sinclair Lewis, Eugene O'Neil, Pearl Buck, William Faulkner, and Ernest Hemingway. When we have enumerated these individuals we have considered the class in extension. To be a member of this class, a person must (1) have been born in the U.S., (2) have met the requirements set by the Prize committee for work in literature, and (3) have been awarded the prize by a representative of that committee. Any person who meets all of these requirements is a member of the class, and when we consider the class in terms of these requirements, we are considering the class in intension. Considering then the significate of a sign as a class, we can clarify the difference between denoting and designating. A sign denotes the significate in extension, while it designates the significate in intension. The important point here is that two signs may have the same denotation and yet have different designations. The clearest instances are false signs which have the same denotation, namely, the null class, but which have different designations.
Now the problem for the psychologist is to provide an analysis of these complex relations which will permit one to determine the designations as well as the denotations of the signs which control behavior. Determining a sign's denotation is essentially like determining retention by a recognition test. The sign is presented with various potential members of the significate, and the interpreter of the sign then sorts out those that are members from those that are not. But how can we determine the designation of a sign? Here is "the unsolved problem of meaning" and the point at which Allport introduces the concept of "event-structure."
Two alternative ways of conceiving of the designation of signs have been suggested in the past. First there is the attempt in the tradition of British empiricism to deal with such designation by means of images. The defining properties of the class (i.e. the significate) are given in the image aroused by the sign. Thus, the sign designates that class with members similar to the image. This treatment has the advantage of providing the kind of specificity of designation that the problem seems to require, but it has the disadvantage of introducing a mentalistic term ("image") as essential to the analysis. How can we determine the content of another's imagery, and what can we do with Kulpe's evidence of imageless thought? The failure to provide satis-
( 500) -factory answers to these questions led to the decline of the imagery-doctrine. The other way of treating designation employs the CR instead of the image. Signs arouse fractional surrogates of the overt responses aroused by some US. The significate of a sign on this account consists in those stimuli which arouse URs similar to those aroused by the sign. The great advantage of this approach lies, of course, in its obvious objectivity. Its major disadvantage is its failure to provide the kind of designative specificity which the problem requires. How does the response of looking at a red wine differ from the response of looking at a white one? Or, for that matter, how does the drinking of red wine differ from the drinking of white? It seems quite unlikely that the specificity of designation characteristic of human behavior and that of the higher animals can reside completely in differences among responses. The afferent character of what is designated must somehow be incorporated in the analysis.
Before I turn to Allport's conception of "event-structure," I must make a confession to the reader. I proposed at the outset of this review to explain what Allport meant by the "unsolved problem of meaning," but what I have done instead is to say what I think he should have meant by it. My only excuse is that Allport's treatment of the problem is so thoroughly intertwined with his views on a host of philosophical issues that I found it impossible to disentangle the original question. When we come to his concluding chapter ("Outline of a General Theory of Event-Structure') the difficulty is even greater, for there the issues are represented on the canvas of the cosmos where the primeval forces of thermodynamics and gestalt ontology dominate the landscape. As a consequence, all that I can do to represent his views about event-structure is to express what I think might be the relation between what he seems to mean by "event-structure" and what I have described as the problem of meaning.
The concept of "event-structure" appears to be a way of regarding patterns of cortical events as mediators in the designation of signs. In other words, Allport appears to suggest that, through a knowledge of the qualitative structural' properties of cortical processes, we may be able to find a basis for class-designation which has the specificity that images are reputed to have without their concomitant subjectivity. In fact, he seems to say that, when we have begun to specify the structural properties of such processes, we shall be able to derive implications both about physiology and about behavior which will be directly testable. In this sense, perhaps, the concept of "event-structure" solves "the unsolved problem of meaning."
Does all this mean that Allport has coined a new term for whatever was named by the phrase "cognitive structure?" I do not think so. What Allport appears to have done is to specify some of the spatial and temporal characteristics which cortical events must have if they are to play their proper role in sign-behavior. Hebb was the first to return our attention to this problem, while Allport is the first to press the analysis to the biological level, on the one hand, and to the cultural level, on the other.
What will interest the reader, however, is not the key to the problem which Allport offers, but the many locks which he attempts to open with his key. His survey of the theoretical problems in perception is clear and exciting, and I suspect that his book will be valued primarily as an excellent description of the present status of perceptual theory. At the end of Allport's book the reader is left with
( 501) the feeling that he has been on a long trip, in which he has seen so many exciting things that he can only remember one thing clearly—how much more pleasant it would have been with less baggage.
BENBOW F. RITCHIE
University of California