Trial‐History Effects During <scp>WCST</scp> in Treatment‐Resistant Schizophrenia

AIM: Treatment-resistant schizophrenia (TRS) is a clinically important subtype of schizophrenia, but its cognitive characteristics remain incompletely understood. The Wisconsin Card Sorting Test (WCST) is widely used to assess executive dysfunction in schizophrenia, although conventional summary analyses may overlook trial-by-trial behavioral adaptation. This study aimed to examine whether recent trial outcome modulates subsequent responding differently between TRS and non-TRS patients during the WCST. METHODS: We analyzed WCST data from 41 outpatients with schizophrenia, including 13 patients with TRS and 28 with non-TRS. Conventional subject-level indices, including perseveration rate and outcome-dependent reaction time (RT) adjustment, were first compared between groups. We then performed trial-level linear mixed-effects analyses using log-transformed RT as the dependent variable, with previous trial correctness, treatment resistance, their interaction, age, and sex as fixed effects. Additional sensitivity analyses included models with by-subject random slopes for previous trial correctness and analysis in an age-matched sample. RESULTS: Conventional summary analyses showed no significant group differences in perseveration rate or RT measures. In the primary model, previous trial correctness significantly modulated subsequent RT, and the interaction between previous trial correctness and TRS suggested a group difference in outcome-dependent RT modulation. Specifically, the RT difference between trials following error and correct responses appeared smaller in TRS patients than in non-TRS patients. However, sensitivity analyses using by-subject random slopes and an age-matched sample showed similar directions and magnitudes of the interaction but did not reach statistical significance. CONCLUSION: Conventional WCST summary analyses did not distinguish TRS from non-TRS patients, whereas trial-level modeling provided preliminary evidence for possible alteration of outcome-dependent response adjustment in TRS. These findings suggest that dynamic behavioral phenotypes may provide a useful complement to standard neuropsychological indices in characterizing cognitive heterogeneity in schizophrenia, but replication in larger samples is needed.