Transdiagnostic inflexible learning dynamics explain deficits in depression and schizophrenia

Area of Science:

• Neuroscience
• Computational Psychiatry
• Cognitive Psychology

Background:

• Deficits in reward learning are common in mental disorders.
• The underlying neuro-computational mechanisms and transdiagnostic nature of these deficits are unclear.

Purpose of the Study:

• To investigate shared and distinct neuro-computational mechanisms of reward learning deficits in major depressive disorder and schizophrenia.
• To explore how reward history influences behavior in these patient groups compared to controls.

Main Methods:

• Combined computational modeling and single-trial electroencephalography (EEG) regression.
• Assessed probabilistic reward learning in patients with major depressive disorder (n=33), schizophrenia (n=24), and healthy controls (n=33).
• Utilized a task requiring integration of reward history to guide choices.

Main Results:

• Both patient groups exhibited reduced learning dynamics with hypersensitive, less decaying learning rates, increasing susceptibility to feedback.
• Schizophrenia patients showed an approach bias, while major depressive disorder patients had heightened sensitivity to disconfirmational feedback.
• Altered neural processing, including a lack of expected value tracking, was observed in both patient groups.

Conclusions:

• Reduced trial-by-trial learning dynamics represent a convergent deficit in reward learning across major depressive disorder and schizophrenia.
• Specific learning impairments, such as approach bias and feedback sensitivity, distinguish between the disorders.