Behavioral and neural predictors of upcoming decisions

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Brain regions like the prefrontal cortex, amygdala, and ventral striatum are crucial for decision-making.
• The precise neural mechanisms linking reinforcement history to future choices are not fully understood.

Purpose of the Study:

• To investigate the relationship between current reinforcements and future decisions using functional magnetic resonance imaging (fMRI) and reinforcement learning (RL) theory.
• To estimate individual value functions that quantify the impact of past rewards on subsequent decision-making.

Main Methods:

• Subjects performed a task involving choices between high-risk (low probability, high reward) and low-risk (high probability, small reward) options.
• Functional magnetic resonance imaging (fMRI) was used to measure brain activity during decision-making.
• Reinforcement learning (RL) principles were applied to model value functions reflecting how past outcomes influenced current choices.

Main Results:

• Individual differences in estimated value functions predicted behavioral strategies, such as repeating high-risk choices after high-risk rewards.
• Value functions also predicted the relationship between neural activity in prefrontal and subcortical regions during one trial and the decision made in the subsequent trial.
• These findings demonstrate that value functions are reflected in both behavioral adjustments and associated neural activity.

Conclusions:

• Value functions derived from reinforcement learning principles provide a framework for understanding how past experiences shape future decisions.
• Neural activity in key brain regions dynamically adjusts based on these value functions, linking neural processing to behavioral strategy.
• This research establishes a novel connection between observable behavior, underlying neural mechanisms, and the computational principles of reinforcement learning in decision-making.