A Symmetric Bayesian Framework for Swarm Information Interaction and Collective Behavior Prediction

This paper studies the information interaction process in Bayesian theorem-based swarm systems. Through theoretical analysis, model construction, and simulation experiments, it explores how Bayesian decision-making utilizes information cascades to update its state step by step in group information interaction. The system operates within a theoretical framework where an underlying symmetry governs the dynamic combination of prior knowledge, neighbor information, and target guidance, leading to spontaneous aggregation behavior similar to biological swarms. A key embodiment of this symmetry is the action–reaction force parity between agents, which ensures local stability.The simulation results show that groups with different prior information exhibit a multi-stage convergence characteristic, which reveals that within each iteration step, the agent adheres to the rules for information-symmetric communication and interaction. This dynamic behavior is a true reflection of natural biological populations and provides theoretical support for practical applications such as traffic management and robot collaboration.