Uncertainty and Structure as Psychological Concepts

Abstract

It was a misfortune of psychology that it lacked a tradition of dealing with rigorous mathematical theories when psychologists were first attracted by information theory. Applications were made with simple-minded identification of psychological concepts with communication terms, without really paying attention to the meaning of the terms in the respective areas. Quite a few experiments were reported measuring human channel capacity under various experimental conditions without asking the basic question: Does the human being comply with the definition of channel in communication engineering? It is true that, in spite of this carelessness, the bulk of experiments re~ ported demonstrated some systematic results as summarized by G. A. Miller in his concept of The Magical Number Seven. However, these experiments also led to various riddles and confusions as illustrated by Garner in Chapter 2 of this book. And this is undoubtedly the reason that many frustrated psychologists finally gave up information theory as useless to psychology. Still, after the waxing and waning of information theory in psychology, an important recognition remained: Information processing is one of the most significant functions of man. The recognition must eventually revive the application of information theory to psychology as a sheer necessity. Probably " application" is not a proper word. A kind of information theory must be developed which is suitable to describe as complicated aa information processing mechanism as man. A first step toward such a theory was taken by McGill in his paper published in Psychometrika in 1954. What I call a misfortune of psychology is this: Instead of taking McGill's mathematical system (called symmetric uncerlainty analysis by Garner and abbreviated here as SUA) as a conceptual tool iu analyzing psychological problems, the tradition of psychology almost forced us to see it as another statistical testing technique analogous to the analysis of variance. As such, SUA was not so handy as the analysis of variance because of the lack of known distributions, and thus SUA failed to acquire popularity. What we needed then, and need now, is a conceptual means which logically bridges information theory to psychology. So the author could not do better in entirely leaving out of the book the significance testing aspect of SUA. I t must be pointed out that SUA is not a model of human behavior. I t is a system of mathematics (or, I would rather say, of logics) so that it is infallible as far as it goes. This aspect of SUA must be clearly remembered. Information theory, developed in communication engineering, is a normative theory. It is