| Children at age three perform poorly on a variety of flexible representation tasks. One task, the Dimensional Change Card Sort (DCCS), requires the flexible use of two competing sets of rules to sort cards. It has been suggested that children at this age have not yet developed a complex-enough representational system to foster flexible performance on the DCCS. Cognitive Complexity and Control (CCC) Theory posits that successful performance hinges on the ability to: (a) reflect upon the two competing pairs of rules, and then (b) form a higher-order rule hierarchy in which conflict between pairs of rules can be reconciled. The CCC account suggests that performance depends solely on the child's cognitive abilities, given that the level of difficulty (demands) of the task is held constant. An alternate account suggests that task demands and characteristics play a role in task performance. The hypothesis that task features can improve performance on flexible representation tasks was tested in two experiments. In the first experiment, children who were allowed to choose when to switch sorting dimensions performed significantly better than children in the typical DCCS task. In the second experiment, children performed significantly better on three DCCS variants, in comparison to typical DCCS performance. Variants in which more than one task characteristic was altered improved performance to a greater extent than variants in which one characteristic was altered. Within-child analysis indicated that the same child often successfully performed the DCCS variant and failed the typical DCCS; the reverse was not often the case. The results suggest that Cognitive Complexity and Control Theory may not adequately explain three-year-old performance on flexible representation tasks; instead, task characteristics may interact with developing cognitive abilities to affect performance. Results of this study and other studies of flexible representation tasks were interpreted in terms of an insufficient system of differential inhibition and activation weighting. This system is described in terms of a basic, theoretical connectionist model of parallel distributed processing. |
