Mutual benefits: Combining reinforcement learning with sequential sampling models.

Area of Science:

• Cognitive Neuroscience
• Computational Psychiatry
• Decision Science

Background:

• Reinforcement learning (RL) models explain error-driven learning.
• Sequential-sampling models address decision-making processes.
• These modeling traditions have historically developed independently.

Purpose of the Study:

• To review the theoretical background for integrating RL and sequential-sampling models.
• To survey recent empirical efforts toward model integration.
• To highlight the mutual benefits and future promises of this unified framework.

Main Methods:

• Theoretical review of model integration.
• Synthesis of empirical studies on combined modeling approaches.
• Discussion of the implications for cognitive modeling.

Main Results:

• Integration provides a unified framework explaining trial-by-trial choice behavior.
• Combined models account for response time distributions.
• Recent empirical work demonstrates the feasibility and utility of integration.

Conclusions:

• Integrating RL and sequential-sampling models offers a powerful, unified approach.
• This integration benefits both cognitive modeling and model-based cognitive neuroscience.
• The approach promises significant advancements in understanding the neural basis of cognition.