Separate encoding of model-based and model-free valuations in the human brain

Area of Science:

• Neuroscience
• Cognitive Psychology
• Neuroeconomics

Background:

• Behavioral studies demonstrate dual process theory, distinguishing intuitive (System 1, model-free) and reflective (System 2, model-based) problem-solving.
• The neurobiological underpinnings of these concurrent cognitive systems remain largely unexplored due to difficulties in isolating their distinct neural activations.

Purpose of the Study:

• To introduce a novel neuroeconomic task designed to differentiate subjective valuation and updating signals associated with System 1 and System 2.
• To investigate the neural correlates of dual process decision-making in humans.

Main Methods:

• Development of a novel neuroeconomic task to elicit distinct neural signals.
• Utilizing neuroimaging techniques to measure brain activity during problem-solving.
• Analysis of concurrent value and updating signals in relation to System 1 (model-free) and System 2 (model-based) processes.

Main Results:

• Identified concurrent value signals in the prefrontal cortex: a System 1 reinforcement signal and a System 2 Bayesian signal.
• Localized System 1 updating signals to striatal areas and System 2 updating signals to the lateral prefrontal cortex.
• Observed prefrontal cortex signals preceding optimal choices, while anterior cingulate cortex and globus pallidus signals preceded deviations from optimal reinforcement learning, particularly under high uncertainty.

Conclusions:

• Neurobiological evidence supports the existence of distinct neural substrates for model-free and model-based decision-making processes.
• Neural signals associated with System 1 and System 2 updating precede choices reflecting their respective principles.
• Disagreements between dual systems appear mediated by uncertainty rather than direct conflict, aligning with recent theoretical frameworks.