Deficient Monetary Reward Processing in Schizophrenia: A Meta-Analysis and Computational Modeling of a Monetary Incentive Delay Task

Evidence regarding monetary incentive-processing deficits in schizophrenia (SZ) remains inconsistent. Clarifying how Monetary Incentive Delay (MID) task performance relates to model-derived learning indices may help characterize motivational and incentive-processing abnormalities in SZ. We hypothesized that patients with SZ would show altered behavioral performance and model-derived expected-value and prediction-error (PE) indices associated with clinical symptom severity.

We integrated a meta-analysis of MID studies (30 studies; SZ/healthy controls (HCs), n = 785/755) with an independent subject-level MID task (SZ/HCs, n = 60/74). A Rescorla-Wagner model with a fixed learning rate was used to derive group-comparable mean absolute PE values and expected-value (Q) indices.

The meta-analysis showed small-to-moderate behavioral differences in SZ including prolonged reaction times (RT) and reduced reward-trial correct rate (CR). In an independent subject-level computational study, patients with SZ showed prolonged RT, lower reward- and loss-trial CR, and higher neutral-trial CR than HCs, consistent with reduced cue-specific modulation. Model-derived analyses showed higher absolute reward PE and absolute loss PE (ABSLPE), lower Reward-Q, and more negative Loss-Q in SZ under the current Rescorla-Wagner framework. Loss-related indices, particularly ABSLPE and Loss-Q, were associated with Positive and Negative Syndrome Scale-Negative scores. Trial-level logistic regression analyses further indicated weaker cue-related modulation of hit probability in SZ.

These findings suggest that SZ is associated with altered MID task performance and model-derived PE- and Q-related indices. These indices may help characterize incentive-processing abnormalities in SZ, but their clinical and neurobiological significance requires validation in future longitudinal and multimodal studies.