Neural Mechanisms of Human Decision-Making

Area of Science:

• Neuroscience
• Computational Neuroscience
• Cognitive Science

Background:

• Human decision-making involves complex neural computations.
• Existing models often lack a unified framework for integrating various brain functions.
• Animal action-selection theories provide a basis for understanding decision processes.

Purpose of the Study:

• To present a neural network model of human decision-making.
• To explain the interaction between brain regions within a proposer-predictor-actor-critic framework.
• To adapt animal action-selection theories for serial operation in humans.

Main Methods:

• Developing a detailed neural network model based on animal data and action-selection theories.
• Modeling task-relevant cortical areas for plan proposal using model-free computations.
• Incorporating medial temporal lobe and cortical areas for outcome prediction.
• Utilizing basal ganglia for plan acceptance/rejection based on reward history.
• Implementing a reward-prediction system for outcome evaluation and dopamine signaling.

Main Results:

• The model explains human decision-making through sequential and hierarchical processing.
• It integrates fast, model-free proposal with slower, model-based prediction.
• The framework accounts for plan rejection and iterative option selection.
• Dopamine acts as a critic signal for training cortical and basal ganglia pathways.

Conclusions:

• The proposer-predictor-actor-critic framework unifies neural mechanisms for complex human decisions.
• The model provides insights into model-based decisions, habitization, and risky behavior.
• Adapting animal action-selection mechanisms explains sophisticated human planning.