Trust, Lies, and Long Memories: Emergent Social Dynamics and Reputation in Multi-Round Avalon with LLM Agents

2026-04-22

AI Summary

This study investigates how large language model (LLM) agents develop social trust and reputation mechanisms in repeated rounds of Avalon, a hidden-role deception game. By endowing LLM agents with cross-game memory to record and leverage historical interaction data, and combining behavioral log analysis with natural language reasoning, the work reports the first observation of role-dependent reputation formation among LLM agents. Experimental results demonstrate that high-reputation players receive 46% more team invitations. Furthermore, under conditions of high reasoning intensity, 75% of evil-aligned agents proactively build trust during early missions to execute strategic deception—a significant increase compared to 36% under low-intensity conditions—highlighting the critical role of cognitive investment in shaping deceptive strategies.

Technology Category

• Multiagent Systems: Mechanism Design
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Humans and AI: Learning Human Values and Preferences

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

We study emergent social dynamics in LLM agents playing The Resistance: Avalon, a hidden-role deception game. Unlike prior work on single-game performance, our agents play repeated games while retaining memory of previous interactions, including who played which roles and how they behaved, enabling us to study how social dynamics evolve. Across 188 games, two key phenomena emerge. First, reputation dynamics emerge organically when agents retain cross-game memory: agents reference past behavior in statements like "I am wary of repeating last game's mistake of over-trusting early success." These reputations are role-conditional: the same agent is described as "straightforward" when playing good but "subtle" when playing evil, and high-reputation players receive 46% more team inclusions. Second, higher reasoning effort supports more strategic deception: evil players more often pass early missions to build trust before sabotaging later ones, 75% in high-effort games vs 36% in low-effort games. Together, these findings show that repeated interaction with memory gives rise to measurable reputation and deception dynamics among LLM agents.

Problem

• social dynamics
• reputation
• deception
• multi-round games
• LLM agents

Innovation

• reputation dynamics
• strategic deception
• cross-game memory
• LLM agents
• emergent social behavior