Bayesian tit-for-tat fosters cooperation in evolutionary stochastic games

Area of Science:

• Decision-making
• Game Theory
• Evolutionary Biology

Background:

• Learning from experience is crucial for complex decision-making.
• Bayesian inferential strategies adapt actions based on new evidence, unlike reactive strategies.
• Social dilemmas explore how altruistic or selfish choices impact collective outcomes.

Purpose of the Study:

• Investigate the resilience of Bayesian strategies against reactive strategies in dynamic environments.
• Analyze how actions influencing environmental states affect future payoffs.
• Determine conditions for the success of Bayesian tit-for-tat strategy.

Main Methods:

• Utilized stochastic games to model environmental state changes conditioned on player actions.
• Examined three distinct rules for state transitions (resource-rich to resource-poor).
• Assessed the invasion resistance of Bayesian tit-for-tat against reactive strategies.

Main Results:

• Bayesian strategies demonstrated resilience against a broad range of reactive strategies.
• Bayesian strategies were more effective in fostering cooperation and sustaining resource-rich states.
• The success of Bayesian strategies was contingent on the strategy pool composition and environmental transition rules.

Conclusions:

• Bayesian strategies offer a robust approach to cooperation in social dilemmas with environmental feedback.
• Environmental dynamics and the prevalence of other strategies significantly influence the effectiveness of Bayesian decision-making.
• Further research into adaptive strategies is warranted for understanding complex social and ecological systems.