Adaptive coding of action values in the human rostral cingulate zone.

Area of Science:

• Neuroscience
• Cognitive Science
• Decision Making

Background:

• Selecting actions in uncertain environments requires learning from experience.
• The human rostral cingulate zone (RCZ) is implicated in integrating action-outcome associations and action valuation.

Purpose of the Study:

• To investigate how the RCZ processes action-outcome associations under varying environmental uncertainty.
• To examine the relationship between RCZ activity, learning rates, and environmental stochasticity.

Main Methods:

• Functional magnetic resonance imaging (fMRI) was used in healthy volunteers.
• A probabilistic reversal learning task with high (HP) and low (LP) reward probabilities was employed.
• Reinforcement learning models were used to derive learning rates.

Main Results:

• Behavioral adaptation to reversals was faster in the HP condition, with higher derived learning rates.
• The RCZ showed stronger responses to negative feedback during reversals in the HP condition.
• RCZ activity related to negative reward prediction errors correlated with the learning rate.

Conclusions:

• RCZ responses are modulated by the information content of the environment.
• The RCZ integrates reinforcement information over a variable window of trials based on environmental stochasticity.
• Action value adjustments in the RCZ covary with environmental uncertainty, optimizing behavioral adaptation.