Effects of contextual processing on visual conditional associative learning in schizophrenia

Background: We adapted visual conditional associative learning paradigms to assess the contextual processing deficit model of schizophrenic cognitive impairment proposed by J.D. Cohen and D. Servan-Schreiber in 1992. In this task subjects learn the associations between four sets of stimuli through the use of feedback. We administered two experimental conditional associative learning conditions: in one, the eight stimuli used to make four pairs were all different; in the other, the pairs were made from different combinations of four identical stimuli, requiring the use of contextual information to mediate correct performance. Two additional associative learning tasks were administered where subjects generated the stimulus pairings or observed the experimenter form the pairs, eliminating the need to learn from feedback. Methods: We tested 37 patients with schizophrenia and 20 healthy control subjects in each conditional associative learning task condition. Results: Patients demonstrated significant impairments on all four conditional associative learning tasks. The demand to process contextual information did not differentially impact patient performance. Patients were better able to learn associations if they generated or observed the pairings rather than utilized feedback to guide learning. Conclusions: Patients with schizophrenia demonstrate pronounced deficits in the ability to utilize feedback to guide learning. We found no evidence of an additional deficit in processing of contextual information.