Optimal policy for value-based decision-making.

Area of Science:

• Cognitive Neuroscience
• Decision Science
• Computational Neuroscience

Background:

• Drift diffusion models (DDMs) are foundational in understanding perceptual decision-making and neural processes.
• While DDMs explain perceptual choices, a robust theoretical framework for their application to value-based decisions is lacking.

Purpose of the Study:

• To establish drift diffusion models as implementing optimal strategies for value-based decisions.
• To elucidate the theoretical underpinnings of decision-making processes across perceptual and value-based domains.

Main Methods:

• Theoretical analysis of drift diffusion models applied to value-based decision scenarios.
• Examination of model optimality under varying task assumptions and utility functions.

Main Results:

• Drift diffusion models optimally implement value-based decisions when decision boundaries collapse over time and depend on prior reward knowledge.
• Optimality is contingent on specific task assumptions; deviations, such as non-linear utility functions, compromise model performance.
• The study provides a theoretical basis for the observed similarities between perceptual and value-based decision processes.

Conclusions:

• Drift diffusion models provide a unifying theoretical framework for both perceptual and value-based decision-making.
• The findings identify critical conditions for optimal decision-making and predict when similarities between decision types may diverge.