Contribution of amygdala to dynamic model arbitration under uncertainty

Area of Science:

• Neuroscience
• Computational Neuroscience
• Behavioral Neuroscience

Background:

• The brain navigates environmental uncertainty using multiple predictive models.
• Simultaneous stimulus- and action-based learning are crucial for predicting outcomes.
• Reliability of these models informs choice behavior.

Purpose of the Study:

• To quantify concurrent learning under varying environmental uncertainty.
• To investigate the roles of the amygdala and ventral striatum in learning and choice.
• To elucidate the dynamic interaction between different learning systems.

Main Methods:

• Utilized a combination of experimental and computational approaches.
• Employed tasks with differing levels of environmental model uncertainty.
• Compared behavioral data from control and lesioned monkeys (amygdala, ventral striatum).

Main Results:

• Identified a dynamic, competitive interaction between stimulus- and action-based learning.
• Demonstrated a distinct role for the amygdala in modulating this interaction.
• Showed that the amygdala adjusts the initial balance between learning systems.

Conclusions:

• The amygdala dynamically regulates the interplay between learning systems, influencing choice behavior.
• This finding offers a novel explanation for previous contradictory observations in learning research.
• Provides a framework for future studies on circuit-level mechanisms of flexible choice under uncertainty.