Multidimensionality in Executive Function Profiles in Schizophrenia: a Computational Approach Using the Wisconsin Card Sorting Task

Deficits in executive functions (EF) are frequently implicated in the aetiology of schizophrenia. However, no consensus exists regarding the contribution of EF to the dysfunctions observed in these individuals, with high levels of heterogeneity characterising the neurocognitive deficits, and traditional methods of assessment providing limited utility to explore the variation. We examine a proposal referred to as the multidimensional hypothesis, rather than performance on cognitive tasks being primarily dependent on a single EF, that performance is multiply realised by a range of uneven EF profiles. We describe a computational approach, using the WCST, that allows manipulations to ability levels of processes relating to each EF component to be explicitly assessed. Our central objective is to determine whether models exhibiting unique profiles of EF abilities may simulate the performance of three target groups: individuals with schizophrenia, healthy first-degree relatives and controls. For each target group, we show that of the performance measures yielded by the WCST, perseverative errors (PE; commonly assumed to rely heavily on EF) are simulated by multiple models. For a range of secondary measures (cards sorted, categories completed and non-perseverative errors; assumed less specific to EF abilities), our manipulations show little resemblance to the target groups. We view these results to support the multidimensional hypothesis, in that (1) no single EF process accounts for the poorer performance of the schizophrenic group, and (2) performance of PE in the target groups is achieved in models of diverse EF profiles. The implications of these findings for the study of EF in neuropsychological disorders are discussed.