Lecture 1:The Information Processing Approach to Intelligence

Abstract

The study of intelligence has been heavily influenced by an informal folk psychology of what constitutes cognitive behaviour. In Binet’s original intelligence test (Binet and Simon, 1905), items were based on what knowledgeable observers thought the “average” (or dull, or smart) person could do. In other words, the decision about what to test was based on an informal induction from observations of individual variations in mental competence rather than a formal deduction from theory of how competence is achieved. “Psychometric” theories of intelligence (Thurstone, 1938; Cattell, 1971; Guilford, 1967) replaced informal induction by formal mathematical models. Factor analysis and related techniques are used to determine the statistical adequacy of a particular mathematical model as a representation of the distribution of test scores. Interpreting the factors involves more intuition than formal deduction (McNemar, 1964). The point can be illustrated by considering a hypothetical model of individual performance on algebra word problems; the familiar studies about the time it takes to dig a ditch or the amount of change one gets in a transaction. A psychometrician might be able to predict individual performance by a linear combination of a person’s score on tests of verbal comprehension and numerical computation. Even if the prediction were perfect problem solution would not have been accounted for in any non-statistical sense, because no model of the process of problem solving would have been offered.