Learning Contextual Reward Expectations for Value Adaptation.

Area of Science:

• Neuroscience
• Decision Science
• Cognitive Psychology

Background:

• Subjective value adapts to expected reward statistics within temporal contexts.
• Mechanisms of learning contextual reward expectations remain poorly understood.

Purpose of the Study:

• Investigate how contextual reward expectations are learned.
• Examine the neural substrates underlying this learning process.

Main Methods:

• Behavioral analysis of a gambling task with varying context alternation rates.
• Reanalysis of functional magnetic resonance imaging (fMRI) data.

Main Results:

• Average reward expectation learned from recent experience without explicit cues.
• Context-dependent learning emerged with fast context alternation; both cue-dependent and independent learning observed with slower alternation.
• Reward prediction error signals identified in ventral tegmental area/substantia nigra and ventral striatum.
• Ventral tegmental area activity correlated inversely with predicted option value, particularly in individuals with high choice adaptation.

Conclusions:

• Behavioral and neural data reveal mechanisms for learning contextual reward expectations.
• The brain utilizes reward prediction errors to adapt choices based on learned contextual values.
• Findings elucidate the neural basis of adaptive decision-making in dynamic environments.