Decomposing Card-Sorting Performance: Effects of Working Memory Load and Age-Related Changes

Objective: The Wisconsin Card Sorting Test (WCST) is often regarded as a prototypical neuropsychological test of set-shifting ability. However, it has been proposed that WCST performance involves cognitive processes other than set shifting, such as set maintenance and rule inference. Distinguishing between these processes is necessary for the meaningful interpretation of WCST performance deficits in neuropsychological populations. In the present studies, we aimed to concurrently measure processes of set shifting, set maintenance and rule inference in a computerized version of the WCST, and to dissociate these processes based on their dependence on working memory capacity. Method and Results: In Study 1, we manipulated the number of card-sorting rules to vary the demands placed on working memory-dependent processes of rule inference. As predicted, integration errors as a novel measure of rule-inference efficiency were selectively affected by increasing the number of rules from 3 to 4. In Study 2, we examined age-related changes in set shifting, set maintenance, and rule inference. We found a specific association between age and integration errors, indicating that rule inference, but not set shifting or set maintenance, is affected in older individuals. Conclusions: Rule inference on WCST-like card-sorting tasks appeared to be selectively impaired when the amount of information to be integrated in working memory increases or when working memory capacity is reduced (as in older individuals). Our findings indicate that measuring integration errors as an index of a distinct rule-inference process can improve the understanding and interpretability of WCST performance.