Neural mechanism for stochastic behaviour during a competitive game

Area of Science:

• Computational neuroscience
• Behavioral economics
• Primate cognition

Background:

• Non-human primates exhibit stochastic choice behavior, particularly in competitive scenarios.
• Understanding the neural basis of dynamic decision-making is crucial.

Purpose of the Study:

• To develop a biologically plausible model of decision-making that explains stochastic and adaptive behavior.
• To investigate the neural mechanisms underlying choice behavior in primates using computational modeling.

Main Methods:

• A biophysical model of decision-making incorporating synaptic plasticity and a reward-dependent stochastic Hebbian learning rule.
• Simulating reinforcement learning dynamics and agent interactions.
• Comparing model outputs with behavioral data from monkeys playing a matching pennies game.

Main Results:

• The model successfully reproduced key features of monkey behavioral data, including quasi-random behavior despite intrinsic biases.
• It demonstrated how interaction and learning dynamics lead to robust stochasticity.
• The model also explained non-random behavior against non-interactive opponents and slow strategy drift via meta-learning and reward maximization.

Conclusions:

• The proposed model provides a viable neural mechanism for generating stochastic choice behavior in competitive environments.
• It integrates reinforcement learning, synaptic plasticity, and meta-learning to explain complex primate decision-making strategies.