Selective reinforcement learning deficits in antipsychotic-naïve patients with first-episode schizophrenia-spectrum...

There is a paucity of research examining reinforcement learning (RL) in first-episode schizophrenia-spectrum disorder (FES), particularly among antipsychotic-naïve patients. To date, no study has examined "Go-response bias" among antipsychotic-naïve FES patients. This study investigated 34 antipsychotic-naïve FES patients and 41 demographically-matched healthy-controls (HCs) using two well-validated computerized RL paradigms [Go/NoGo (GNG) and Gain-vs-Loss-Avoidance (GLA)] on win-stay and lose-shift rates (trial-by-trial measures of reward- and punishment-driven RL), procedural RL, value-guided decision-making, and Go-response bias as primary outcomes. Correlations between RL measures, and symptom severity and cognitive functions were assessed as secondary outcomes. Patients showed significantly worse reward-driven procedural RL (in GLA task: t₇₂=-2.4, p= 0.017, d= 0.58) and punishment-driven procedural RL (in both GNG: t_59.0=-2.9, p= 0.005, d= 0.69; and GLA: t₇₂=-2.4, p= 0.019, d= 0.57), relative to HCs. Additionally, patients exhibited significantly lower win-stay rates than HCs in both the first block of trials on the GLA paradigm (t₇₂=-2.4, p= 0.018, d= 0.57), demonstrating reduced ability to utilize positive feedback on a trial-by-trial basis. We did not observe any significant difference between patients and HCs on value-guided decision-making on both tasks, suggesting that expected-value representations were relatively unimpaired. Moreover, no Go-response bias was observed in overall patient sample. Finally, lower lose-shift rates were correlated with more severe negative symptom in both tasks, but failed to survive correction for multiple comparisons. Overall, antipsychotic-naïve FES patients demonstrated mild and circumscribed RL. To differentiate the effects of illness and medication on RL, future longitudinal studies should investigate changes in the severity of RL impairment before and after antipsychotic initiation.