Pergamon Person. indiuid. Dijf Vol. 21, No. I, pp. 91-102, 1996 Copyright 0 1996 Elwier Science Ltd Printed in Great Britain. All rights reserved S0191-8869(96)00032-3 0191~8869/96 $15.00+0.00 INTELLIGENCE, ELEMENTARY COGNITIVE COMPONENTS, AND COGNITIVE STYLES AS PREDICTORS OF COMPLEX TASK PERFORMANCE Philip Tucker’ and Peter Warr2 ‘Department of Psychology, University of Wales, Swansea, Wales and ‘Department of Psychology, University of Sheffield, Sheffield, England zyxwvutsrqponmlkjihgfedcbaZYXWVUT (Received 27 June 1995; receivedfor publication 8 February 1996) Summary-A laboratory study is reported in which individual differences underlying variations in the performance of a complex task are examined. Differences in fluid intelligence, elementary cognitive com- ponents (processing speed and working memory) and cognitive styles (tempo, planfulness and complexity) are measured, with a view to comparing between these three types of variable. It is found that fluid intelligence scores are more closely associated with measures of speed rather than accurate task per- formance; processing speed and working memory are both predictive of complex task performance; and differences in self-reported cognitive tempo co-vary with speed of problem-solving but not with speed of routine activity. In general, the inclusion of cognitive style variables does not increase the predictive power of intelligence and the elementary cognitive components. It is concluded that cognitive styles will be most reflected in task performance when the activities involved permit considerable variation in the style under investigation. Finally, an explanation is offered of why fluid intelligence is more predictive of some aspects of complex task performance than others, in terms of their degree of emphasis on speeded cognitive processes. Copyright 0 1996 Elsevier Science Ltd. INTRODUCTION This paper examines cognitive processes in a complex laboratory task in terms of three different approaches to individual variation. Separate literatures have developed in the areas of intelligence, elementary cognitive components, and cognitive styles. We will build on previous research in each of those areas to examine individual differences of the three kinds, and to investigate overlaps between the variables in relation to performance on a complex task. Items to measure intelligence are often classified as covering either ‘crystallised’ or ‘fluid’ processes (e.g. Horn & Cattell, 1966). Psychometric tests of crystallised intelligence include assessments of vocabulary and problem-solving that depend upon specific learned information. Measures of fluid intelligence typically tap aspects of effortful information-processing, for instance in terms of memory span, induction and application of rules, memorising, reasoning, and the solution of abstract problems (Carpenter, Just & Shell, 1990). Standardised tests of fluid intelligence have often been shown to be significant predictors of academic performance and other indices of problem-solving (e.g. Guttman, 1965; 1969; Jensen, 1970; Snow, 1980; 1981) in part because the same types of information-processing are present in both sets of activities. However, the strength of the associ- ations between intelligence and task performance has frequently been found to depend upon the use of relatively difficult tasks and upon the examination of a fairly wide range of ability levels (Crawford, 1991). Differences between experimental participants in their fluid intelligence scores are thus expected to be reflected in their performance of cognitive laboratory tasks when those conditions are met. Until recently, the literature on intelligence testing has emphasised the products of intelligent thinking (in terms of test scores) rather than the processes underlying those products (but see Vernon, 1987; Kyllonen & Christal, 1990). The converse may be suggested about cognitive psy- chologists’ research into elementary cognitive components. Laboratory investigations into encoding, attention and memory processes tend to be motivated by theoretical interest in the nature of cognition, and have rarely addressed the validity of the measures employed in predicting more 91