SA67. Deficient Predictive Coding During a Visual Oddball Task in Early Illness Schizophrenia

Abstract

Background: The P300 event-related potential component is typically elicited during an “oddball” task by infrequent target stimuli (P3b) and infrequent novel distractor stimuli (P3a) that are randomly interspersed among frequent standard stimuli. Although standards do not generally elicit a P300, they can elicit a P3a when their appearance late in a local series of repeating standards is improbable. In a typical oddball task, local sequences of repeating standards end with the appearance of a deviant (target or novel) stimulus. Toward the end of such local sequences, the probability that a standard will appear decreases, and the probability that a deviant will appear increases. Previously, using an auditory oddball task, we showed that healthy subjects generated a P3a in response to late appearing standards within local sequences of repeating standards, suggesting that they implicitly recognized the conditional improbability of a standard following a local sequence of multiple repeating standards, likely in the service of predictive coding. However, chronic schizophrenia (SZ) patients failed to generate this P3a response to late appearing standards. Here, we examined whether similar P3a effects are present using a visual oddball task and whether similar deficits in P3a generation are evident early in the course of SZ. Methods: ERPs were recorded from 19 early illness SZ patients (ESZ; mean illness duration = 2.5 years) and 43 healthy controls (HC) during a visual oddball task containing frequent (80%) standard stimuli and infrequent deviant stimuli (10% targets requiring a response; 10% novel distractors). Consecutively presented standards (S) following deviant (D) stimuli ranged in conditional probability, (P(S|D) = 1.0, P(S|DS) = .97, P(S|DSS) = .97, P(S|DSSS) = .69, P(S|DSSSS) = .41, P(S|DSSSSS) = .29), with late appearing repeated standards being less probable than early ones. ERPs to standards based on their position within local sequences of repeating standards were generated, and the amplitude of the P3a was measured. Results: There was a significant main effect of Standard Position (P < .001) and a significant Standard Position × Group interaction (P = .027). Follow-up analyses showed that late appearing, relative to early appearing, standards within local sequences of repeating standards, elicited a larger P3a in HC (P < .001), but not in ESZ (P = .133). Conclusion: These results suggest that the previously observed P3a elicited by late appearing standards in local sequences of repeating standards during an auditory oddball task is also evident in healthy subjects performing a visual oddball task. The failure to implicitly process the local sequential stimulus probabilities previously observed in chronic SZ was also evident in ESZ, consistent with the presence of deficient task-based predictive coding early in the illness.