Amygdala and Ventral Striatum Make Distinct Contributions to Reinforcement Learning

Area of Science:

• Neuroscience
• Cognitive Science
• Behavioral Economics

Background:

• Reinforcement learning (RL) theories emphasize striatal dopamine in associating choices with outcomes.
• The amygdala's role in Pavlovian processes influencing choice behavior is often overlooked.
• Dopaminergic signals are crucial for learning and decision-making.

Purpose of the Study:

• To investigate the distinct contributions of the ventral striatum (VS) and amygdala to appetitive reinforcement learning.
• To differentiate the roles of the amygdala and VS in deterministic versus stochastic learning tasks.

Main Methods:

• Lesion studies in rhesus macaques targeting the VS or amygdala.
• Utilizing deterministic and stochastic versions of a two-arm bandit reversal learning task.
• Employing a reinforcement learning (RL) model to characterize learning dynamics.

Main Results:

• Amygdala lesions led to decreased learning from positive feedback and reduced choice consistency.
• VS lesions specifically impaired learning in the stochastic task, not the deterministic one.
• VS lesions accelerated choice reaction times, indicating a speed-accuracy trade-off.

Conclusions:

• The amygdala and VS have distinct, non-redundant roles in appetitive reinforcement learning.
• Standard RL accounts need updating to include the amygdala's contribution to Pavlovian influences on choice.
• The findings highlight differential processing of feedback and task demands by the amygdala and VS.