Reward and Punishment Learning Across Common Forms of Psychopathology

Background: The Iowa Gambling Task (IGT) has been widely used to assess decision-making deficits in clinical populations. Previously-used methods do not provide clear evidence whether associations between IGT performance and psychopathology are general across multiple forms of psychopathology or are specific to individual disorders. In this study, we examined IGT performance across anxiety, depression, and substance use disorder.Methods: We tested a large sample of adults (N=287) on the play-or-pass version of the IGT. We characterized behavior using a computational model formalizing parameters underlying task behavior: reward and punishment learning rates, win-frequency sensitivity, and response bias. With the model, we examined associations between IGT performance and lifetime diagnostic history of anxiety, depression, and substance use disorder.Results: The computational model revealed that anxiety, but not depression or substance use, was associated with higher punishment learning rates. The model also showed that depression, but not anxiety or substance use, was associated with higher win-frequency sensitivity.Limitations: Too few cases of current substance use disorder precluded examination of how current substance use affected IGT performance. Only main effects of diagnostic status were examined; thus, future research should examine comorbidity between diagnoses on IGT performance.Conclusions: Findings are consistent with research showing that anxiety is associated with punishment avoidance. In addition, individuals with depression may prioritize how frequently wins occur over the magnitude of those wins. Interventions targeting affective decision-making should consider the role of punishment learning and reward frequency in anxiety and depression, respectively.