[Value representation in the striatum and prefrontal cortex].

Area of Science:

• Neuroscience
• Cognitive Science
• Decision Making

Context:

• The brain must process reward information to guide behavior in dynamic environments.
• Prior research suggests primates utilize at least two distinct reward valuation processes.
• These processes involve direct experience (e.g., temporal-difference learning) and predictive modeling (e.g., inference).

Purpose:

• To elucidate the neural underpinnings of multiple reward prediction mechanisms.
• Investigate the distinct roles of the nigro-striatal and prefrontal networks in value generation.

Summary:

• Four experiments were conducted: single-unit recording in monkeys (caudate, lateral prefrontal cortex, dopamine neurons), human fMRI, and simultaneous recordings during reward inference tasks.
• Findings indicate the nigro-striatal network employs model-free (temporal-difference) learning for reward prediction.
• The prefrontal network utilizes a model-based approach, leveraging categorization and inference for reward prediction.

Impact:

• Differentiates the functional roles of key brain networks in reward processing.
• Provides insight into how the brain combines direct experience and internal models for adaptive decision-making.
• Advances understanding of the neural basis of value generation and prediction in primates.