Variability in Dopamine Genes Dissociates Model-Based and Model-Free Reinforcement Learning.

Area of Science:

• Neuroscience
• Cognitive Science
• Genetics

Background:

• Behavior is driven by multiple learning systems, including reflexive model-free and flexible model-based learning.
• Dopamine's differential contributions to these systems are not fully understood, with roles suggested in striatal plasticity and prefrontal working memory.
• Previous studies have not clearly distinguished the neural loci of dopaminergic modulation for different learning strategies.

Purpose of the Study:

• To clarify the relationships between dopamine, neural systems, and distinct learning strategies (model-free vs. model-based).
• To investigate whether striatal or prefrontal dopamine function is associated with specific learning strategies.

Main Methods:

• Utilized a genetic association approach in humans combined with two reinforcement learning tasks.
• Model-free and model-based behaviors were isolated using specific task designs.
• Prefrontal dopamine function was indexed by a COMT gene polymorphism; striatal dopamine function was indexed by a DARPP-32 gene polymorphism.

Main Results:

• Variations in prefrontal dopamine function, indexed by the COMT gene, were associated with model-based learning.
• Variations in striatal dopamine function, indexed by the DARPP-32 gene, were associated with model-free learning.
• Striatal dopamine function was most clearly related to the reflexive (model-free) strategy, while prefrontal dopamine function was most clearly related to the flexible (model-based) strategy.

Conclusions:

• Dissociable brain regions, specifically striatum and prefrontal cortex, support distinct choice strategies.
• Striatal dopamine function is primarily linked to reflexive, model-free decision-making.
• Prefrontal dopamine function is primarily linked to flexible, model-based decision-making.