Model-based action planning involves cortico-cerebellar and basal ganglia networks

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Human action selection involves learning, planning, and memory retrieval.
• Reinforcement learning (RL) categorizes these into exploratory, model-based, and motor-memory strategies.

Purpose of the Study:

• To investigate the neural substrates underlying distinct RL action selection strategies.
• To differentiate brain activity associated with exploratory, model-based, and motor-memory RL.

Main Methods:

• Functional magnetic resonance imaging (fMRI) experiment.
• Sequential action selection task designed to elicit specific RL strategies.
• Analysis of brain activity patterns under distinct experimental conditions.

Main Results:

• Exploratory RL: Increased activity in ventromedial prefrontal cortex and ventral striatum.
• Model-based RL: Activation in dorsolateral prefrontal cortex, dorsomedial striatum, and lateral cerebellum.
• Motor-memory RL: Engagement of supplementary motor area, putamen, and anterior cerebellum.

Conclusions:

• Distinct neural networks support different reinforcement learning strategies.
• A specific prefrontal-basal ganglia-cerebellar network is implicated in model-based RL action selection.