The effect of model uncertainty on cooperation in sensorimotor interactions

Area of Science:

• Cognitive Science
• Neuroscience
• Behavioral Economics

Background:

• Decision-makers utilize probabilistic models for perception and action.
• Model uncertainty arises when these probabilistic models are incorrect or incomplete.
• Model uncertainty influences human sensorimotor behavior, leading to risk-sensitive choices.

Purpose of the Study:

• To investigate the impact of model uncertainty on cooperation in sensorimotor interactions.
• To examine cooperation dynamics in a stag-hunt game paradigm.
• To explore how optimistic deviations in opponent modeling affect cooperative outcomes.

Main Methods:

• Simulations were conducted using agent models incorporating opponent modeling.
• A virtual reality environment was used to implement the agent model for human interaction.
• Human subjects played a sensorimotor game against a computer player with manipulated risk sensitivity.

Main Results:

• Simulations showed that optimistic opponent models increased convergence to cooperative outcomes.
• Human subjects adaptively adjusted their cooperation levels based on the computer player's risk sensitivity.
• Initial human play demonstrated characteristic risk-sensitive biases in the sensorimotor interaction.

Conclusions:

• Model uncertainty is a crucial factor influencing cooperation in two-player sensorimotor interactions.
• Optimistic modeling strategies can promote cooperative behavior.
• Human cooperation is dynamically modulated by perceived risk sensitivity and exhibits inherent biases.