Reward learning impairments in patients with first-episode schizophrenia-spectrum disorder

Abstract

Reward learning refers to outcome-based learning that involves selecting optimal response choices from feedback which facilitate adaptive behavior. It is believed that reward learning paradigm represents a promising translational target in schizophrenia research. Previous studies generated relatively consistent evidence of rapid learning deficits but mixed findings on gradual learning deficits. Reward learning impairments were also associated with symptoms as well as antipsychotics treatment. The current study aimed to investigate the reward learning impairments and its longitudinal change in patients with first-episode schizophrenia spectrum disorder. A total of 34 patients and 36 healthy control participants were recruited. Patients and controls were matched in terms of age, sex, and education level. All participants were assessed twice: at baseline and after one year. For each assessment time point, data were collected on demographics, clinical and treatment characteristics. Participants were asked to complete a battery of cognitive assessments and two reward learning tasks: the Gain vs. loss-avoidance task and the Go-NoGo task. Patients and controls were compared in terms of cross-sectional reward learning performance at baseline and follow-up. Correlates of reward deficits were examined, and longitudinal analyses were conducted to investigate change of reward learning performance over time. At baseline, it was found that patients had significant rapid learning deficit in win-stay (learning from positive feedback) and gradual learning deficits in learning from both positive and negative feedback. Reward-driven learning impairments were more robust. At one-year follow-up, patients continued to have significant rapid learning deficit in win-stay and gradual learning deficits in learning from negative feedback. Longitudinal analyses demonstrated that patients had significant decrease in win-stay rate in training phase and significantly lower accuracy for punishment-driven stimuli across assessment time points. No deficits in representing expected reward value of stimuli or Go response bias were demonstrated. Correlations were found between different symptom domains (negative symptoms, positive symptoms) and reward learning impairments. Current findings regarding rapid and gradual learning deficits in patients with first-episode schizophrenia spectrum disorder were partially in keeping with that of previous studies. Discrepant findings across studies may be attributable to different sample characteristics in terms of illness chronicity and symptoms severity. The current study provided valuable information regarding the longitudinal change of reward learning deficits in early psychosis patients.