Behavioral, computational and self-reported measures of reward and punishment sensitivity as predictors of mental health characteristics

Computational psychiatry proposes that behavioral task-derived computational measures can improve our understanding, diagnosis and treatment of neuropsychiatric disorders. However, recent meta-analyses in cognitive psychology suggest that behavioral and computational measures are less stable than self-reported surveys as assessed by test–retest correlations. If extended to mental health measures, this poses a challenge to the computational psychiatry agenda. To evaluate this challenge, we collected cross-sectional data from participants who performed a popular reinforcement-learning task twice (~5 months apart). Leveraging a well-validated neuro-computational framework, we compared the reliability of behavioral measures, computational parameters and psychological and mental health questionnaires. Despite the remarkable replicability of behavioral and computational measures averaged at the population level, their test–retest reliability at the individual level was surprisingly low. Furthermore, behavioral measures were essentially correlated only among themselves and generally unrelated to mental health symptoms. Overall, these findings challenge the translational potential of computational approaches for precision psychiatry.