COMPLEXITY AND PREFERENCE IN ANIMALS AND MEN*

Abstract

Complexity and preference are terms that are relevant to a theory of behavior I presented at the Nebraska Symposium on Motivation several years ago.' The research effort I wish to discuss is based on that theoretical statement. The theory differs very little from an earlier theory published by Dember and Earl (see Dember2) and it incorporates concepts developed by Berlyne.' The theory contains a small number of terms or concepts and hypotheses which I should like to review briefly in order to make it possible to use them discriminatively in discussing the research. The terms are: Psychological Event, Stimulus Complexity, Psychological Complexity and Optimal Complexity. From these is developed a theory of Preference or choice among psychological events. A hypothesis concerning expected changes in optimal complexity with experience produces an expected set of Changes in Complexity and Preference with Experience. This much is relevant to the experiments to be discussed here. An additional concept, arousal, is in the theory but not germane to these experiments and will not be discussed further. Psychological Event. This is a term I have used to identify a central unit of activity. A psychological event may be initiated by an external stimulus, but many psychological events have the appearance of spontaneity. A response may result from a psychological event, but many psychological events occur without a visible product. A psychological event may have a conscious representation and, thus, the character of perception, but there are circumstances under which psychological events occur without conscious awareness. Thus the need for such a concept arises from the limitations imposed by the concepts of stimulus, response, perception, and cognition, any or all of which may reflect a psychological event, but none of which is identical. Stimulus Complexity. This term refers to the complexity of the distal stimulus. The essential characteristic of stimulus complexity-as distinguished from psychological complexity-is that it should be possible to assign a number to a stimulus that represents its complexity and that this value should remain fixed. Often this appears to be no problem, but I believe appearances to be deceiving. With certain kinds of stimulus materials, one can perform a counting operation. One can have stimuli that vary in the number of dots, or polygons that vary in the number of sides. On a priori grounds, it seems reasonable that a card with 50 dots is a more complex stimulus than a card with five, or that a polygon with 20 sides is more complex than a polygon with four sides. Yet, if we subject a set of such materials to psychological scaling procedures, it is very unlikely that there will be perfect agreement between our numerical count and the results of the scaling. One is then tempted to undertake to resolve the discrepancies in terms of factors that were not included in the count-configurational or pattern factors. Alternatively, one can choose one of the two values, the numerical count or the results of psychological scaling, and choose to ignore the other. The question then becomes, which does one choose? I think there is no question that the choice must be the psychologically derived