Collaborative Multi-Agent Scripts Generation for Enhancing Imperfect-Information Reasoning in Murder Mystery Games
For researchers working on multimodal reasoning under uncertainty and adversarial conditions, this work provides a scalable method to enhance VLM performance in complex social scenarios.
The paper tackles the challenge of imperfect-information reasoning in multiplayer murder mystery games, where VLMs struggle with multi-hop reasoning under deception. The proposed multi-agent framework generates high-quality scripts and uses a two-stage training strategy (CoT fine-tuning + GRPO reinforcement learning) to significantly boost VLM performance in narrative reasoning, hidden fact extraction, and deception-resilient understanding.
Vision-language models (VLMs) have shown impressive capabilities in perceptual tasks, yet they degrade in complex multi-hop reasoning under multiplayer game settings with imperfect and deceptive information. In this paper, we study a representative multiplayer task, Murder Mystery Games, which require inferring hidden truths based on partial clues provided by roles with different intentions. To address this challenge, we propose a collaborative multi-agent framework for evaluating and synthesizing high-quality, role-driven multiplayer game scripts, enabling fine-grained interaction patterns tailored to character identities (i.e., murderer vs. innocent). Our system generates rich multimodal contexts, including character backstories, visual and textual clues, and multi-hop reasoning chains, through coordinated agent interactions. We design a two-stage agent-monitored training strategy to enhance the reasoning ability of VLMs: (1) chain-of-thought based fine-tuning on curated and synthetic datasets that model uncertainty and deception; (2) GRPO-based reinforcement learning with agent-monitored reward shaping, encouraging the model to develop character-specific reasoning behaviors and effective multimodal multi-hop inference. Extensive experiments demonstrate that our method significantly boosts the performance of VLMs in narrative reasoning, hidden fact extraction, and deception-resilient understanding. Our contributions offer a scalable solution for training and evaluating VLMs under uncertain, adversarial, and socially complex conditions, laying the groundwork for future benchmarks in multimodal multi-hop reasoning under imperfect information.