Cooperation in structured populations via coupled reputation and learning: A spatial evolutionary game approach

Abstract

This study presents an agent-based model to investigate cooperation dynamics in spatial evolutionary games by integrating memory-based reputation tracking with heterogeneous adaptive learning. Agents interact on a lattice network and update their strategies based on both neighbors’ historical cooperation rates and payoff differences, governed by a modified Fermi rule with individual sensitivity parameters. Simulation results demonstrate that this dual-layered mechanism sustains cooperation even under strong defection incentives and limited interaction ranges. The model also reveals how memory length and learning heterogeneity jointly influence spatial cooperation patterns and strategy diversity. These findings offer new insights into decentralized mechanisms that promote cooperation in structured populations, with implications for evolutionary biology, distributed systems, and behavioral economics.