Psychometric Tests as Cognitive Tasks: A New "Structure of Intellect"

Abstract

This largely.theoretical discussion attempts to show how the identified in factor-analytic studies of cognitive abilities can be interpreted. in.termS of current theories anu perimental work in cognitive pSychology. After consideration of the drawbacks of such psychometrically-derived theories of cognitive abilities as those of Guttman, Cattell, and. Guilford,, appeal is made to E. B. Hunt's .distributi/e .1.umory, model and A. Newell's concept of the production system asNossiY hnses for developing an alternative theory. ',Such, a theory of cognitive abilities rests upon the individual differences displayed 'in the parametei-sothe tasks found in typical tests of intelligence. As a first step toward developing a \ew structure of intellect model, a detailed subjective analysis is made of the cognitive processes involved in two tests designed to' measure each of the 24 factors in the.1963 version of the Kit of.ReferenCe Tests for .Cognitive. This analysis\is made by systematic coding of aspects of these tasks according to a scheme for relating -these features tojhe distributive memory model and to the producttp:system concept. It is hypothesized:that factor-analytic common factors arise when two or.mare tasks share features in which there are individual differences with respect to (1) the types and contents of MemorY stores involved, (2) the types and 'sequences of/cognitive operations reqUiredand cognitive strai,egies employed by individual subjects, and (3) the types of ,responses elicited. It is claimed that. Crum this point of view, cognitive tasks are. complex, and cognitive factors resist classification by aay rigid taxonomy such as Guilford's Structure of Intellect model'; there are probably no such things as truly pure factors. It is held that the study of individual differences in abilities canProfit greatly if it Ls closely tied to the experimental analysis of particular cognitive 'tasks; a t the. -same time, work in the experimental tradition should pay close 'attentiot-Cotthe problem of identifying sources of variance that are due to indiVidual differences in task parameters.