A Paradigm for the Identification of Independent Cognitive Constructs

A promising approach in recent years has been to develop measures of individual differences based upon componential cognitive theory to supplement or supplant traditional measures. Cognitive tests are developed to measure theoretically based mental operations which can be isolated by the computation of derived measures such as slope and difference scores. Along with others, we believe there are impediments to this approach due to unreliability of derived measures and lack of demonstrated statistical independence of tests of cognitive abilities. This paper describes a methodology for examining measures of individual differences in information processing skills that first follows the tenets of psychometric theory and then addresses cognitive theories. The approach is illustrated by demonstrating its application in tests representing four distinct cognitive paradigms which were administered repeatedly to subjects over three weeks. Recommended direct measures and derived scores for the four paradigms were examined in terms of their stabilities, retest reliabilities, and cross-correlations. Use of these procedures revealed that 1) derived scores had reliabilities near zero, and therefore, their correlations with other variables were equally low, rendering them of little use as individual difference variables, and 2) correlations between basic or nonderived scores were as high as their reliabilities would allow, suggesting that one common factor could account for the majority of the variance. The generality of this repeated measures paradigmatic approach to the identification of individual differences in human ability is illustrated by describing its application to the evaluation of a family of video games, tests of episodic memory, and visual contrast sensitivity at different spatial frequencies.