Causal Inference Gates Corticostriatal Learning.

Area of Science:

• Neuroscience
• Cognitive Science
• Decision Making

Background:

• Credit assignment, attributing outcomes to actions or external causes, is vital for learning.
• Bayesian reinforcement learning models suggest causal beliefs modulate reward prediction errors (RPEs).
• Previous research linked reinforcement learning to the striatum and causal inference to prefrontal regions.

Purpose of the Study:

• To investigate the neural basis of reinforcement learning influenced by causal beliefs.
• To examine how the brain integrates causal inference with reinforcement learning mechanisms.

Main Methods:

• Functional magnetic resonance imaging (fMRI) in 31 human participants.
• A behavioral task manipulating beliefs about causal structure.
• Structural equation modeling to assess effective connectivity.

Main Results:

• RPEs modulated by causal beliefs were found in the dorsal striatum.
• Standard RPEs were represented in the ventral striatum.
• Causal beliefs were associated with activity in the anterior insula and inferior frontal gyrus.
• Effective connectivity was observed between the anterior insula and dorsal striatum.

Conclusions:

• Causal beliefs represented in the anterior insula may modulate RPEs in the dorsal striatum for action value updating.
• This suggests a neural architecture integrating causal inference with reinforcement learning.
• Findings clarify the interaction between causal inference and reinforcement learning in adaptive behavior.