Choice modulates the neural dynamics of prediction error processing during rewarded learning

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Selective engagement with the environment guides learning and its benefits.
• Choice is a critical factor in learning when faced with multiple actions.
• Reward prediction error (RPE) is a key concept in understanding learning from rewarding feedback, but the influence of choice on RPE processing is underexplored.

Purpose of the Study:

• To investigate how choice influences prediction error processing.
• To examine the distinct neural dynamics associated with choice during rewarded learning.

Main Methods:

• Utilized a novel rewarded learning task with intermingled choice and control trials.
• Recorded scalp electroencephalogram (EEG) from healthy adults.
• Employed a temporal difference learning model to infer image valuations and prediction errors.

Main Results:

• Choices were associated with lower overall prediction error magnitudes, particularly during learning.
• Choices induced a higher-amplitude frontocentral event-related potential (ERP) approximately 200 ms post-reward onset.
• This ERP component was negatively correlated with the choice's differential effect on prediction error.

Conclusions:

• Choice exerts a distinct influence on the neural processing of reward signals during learning.
• Existing behavioral, computational, and neurobiological models of rewarded learning should incorporate the specific role of choice.