Working memory contributions to reinforcement learning impairments in schizophrenia

Area of Science:

• Cognitive Neuroscience
• Neuropsychology
• Psychiatry

Background:

• Schizophrenia patients exhibit deficits in reinforcement learning (RL) tasks.
• Learning behaviors involve complex interactions between RL systems and prefrontal cortex-dependent working memory (WM) strategies.
• The specific neural systems underlying learning impairments in schizophrenia remain unclear.

Purpose of the Study:

• To differentiate the contributions of RL and WM systems to learning impairments in schizophrenia.
• To utilize a novel task and computational model to assess RL versus WM mechanisms separately.
• To identify the precise source of learning deficits in schizophrenia patients.

Main Methods:

• Developed and employed a computational task to distinguish between incremental RL and capacity-limited WM.
• Administered the task to 49 schizophrenia patients and 36 healthy controls across 15 learning blocks.
• Varied the number of stimuli (2-6) per block to isolate WM influences from RL system effects.

Main Results:

• Both patient and control groups demonstrated working memory effects, confirmed by set size and stimulus repetition delay.
• Schizophrenia patients exhibited significantly poorer overall performance compared to controls.
• Computational modeling and behavioral analysis revealed that deficits were attributable to altered WM capacity and reliability, with RL processes remaining intact.

Conclusions:

• Working memory, not reinforcement learning, is the primary driver of learning impairments observed in schizophrenia.
• These findings underscore the critical role of working memory dysfunction in the cognitive profile of schizophrenia.
• The study provides a refined understanding of the neural basis of learning deficits in schizophrenia.