This study proposes a method to improve the performance of Werewolf agents by switching between predefined strategies based on the attitudes of other players and the context of conversations. While prior works of Werewolf agents using prompt engineering have employed methods where effective strategies are implicitly defined, they cannot adapt to changing situations. In this research, we propose a method that explicitly selects an appropriate strategy based on the game context and the estimated roles of other players. We compare the strategy adaptation Werewolf agents with baseline agents using implicit or fixed strategies and verify the effectiveness of our proposed method.

With the rapid development of LLM-based agents, increasing attention has been given to their social interaction and strategic reasoning capabilities. However, existing Werewolf-based benchmarking platforms suffer from overly simplified game settings, incomplete evaluation metrics, and poor scalability. To address these limitations, we propose WereWolf-Plus, a multi-model, multi-dimensional, and multi-method benchmarking platform for evaluating multi-agent strategic reasoning in the Werewolf game. The platform offers strong extensibility, supporting customizable configurations for roles such as Seer, Witch, Hunter, Guard, and Sheriff, along with flexible model assignment and reasoning enhancement strategies for different roles. In addition, we introduce a comprehensive set of quantitative evaluation metrics for all special roles, werewolves, and the sheriff, and enrich the assessment dimensions for agent reasoning ability, cooperation capacity, and social influence. WereWolf-Plus provides a more flexible and reliable environment for advancing research on inference and strategic interaction within multi-agent communities. Our code is open sourced at https://github.com/MinstrelsyXia/WereWolfPlus.

Large Language Models (LLMs) have shown remarkable capabilities as autonomous agents, yet existing benchmarks either focus on single-agent tasks or are confined to narrow domains, failing to capture the dynamics of multi-agent coordination and competition. In this paper, we introduce MultiAgentBench, a comprehensive benchmark designed to evaluate LLM-based multi-agent systems across diverse, interactive scenarios. Our framework measures not only task completion but also the quality of collaboration and competition using novel, milestone-based key performance indicators. Moreover, we evaluate various coordination protocols (including star, chain, tree, and graph topologies) and innovative strategies such as group discussion and cognitive planning. Notably, gpt-4o-mini reaches the average highest task score, graph structure performs the best among coordination protocols in the research scenario, and cognitive planning improves milestone achievement rates by 3%. Code and datasets are public available at https://github.com/MultiagentBench/MARBLE.

Large language model (LLM)-based agents have recently shown impressive progress in a variety of domains, including open-ended conversation and multi-step decision-making. However, applying these agents to social deduction games such as Werewolf, which requires both strategic decision-making and free-form language interaction, remains non-trivial. Traditional methods based on Counterfactual Regret Minimization (CFR) or reinforcement learning (RL) typically depend on a predefined action space, making them unsuitable for language games with unconstrained text action space. Meanwhile, pure LLM-based agents often suffer from intrinsic biases and require prohibitively large datasets for fine-tuning. We propose Latent Space Policy Optimization (LSPO), an iterative framework that addresses these challenges by first mapping free-form text to a discrete latent space, where methods like CFR and RL can learn strategic policy more effectively. We then translate the learned policy back into natural language dialogues, which are used to fine-tune an LLM via Direct Preference Optimization (DPO). By iteratively alternating between these stages, our LSPO agent progressively enhances both strategic reasoning and language communication. Experiment results on the Werewolf game show that our method improves the agent's performance in each iteration and outperforms existing Werewolf agents, underscoring its promise for free-form language decision-making.

Achieving Artificial General Intelligence (AGI) requires AI agents that can not only make stratigic decisions but also engage in flexible and meaningful communication. Inspired by Wittgenstein's language game theory in Philosophical Investigations, we propose that language agents can learn through in-context interaction rather than traditional multi-stage frameworks that separate decision-making from language expression. Using Werewolf, a social deduction game that tests language understanding, strategic interaction, and adaptability, we develop the Multi-agent Kahneman & Tversky's Optimization (MaKTO). MaKTO engages diverse models in extensive gameplay to generate unpaired desirable and unacceptable responses, then employs KTO to refine the model's decision-making process. In 9-player Werewolf games, MaKTO achieves a 61% average win rate across various models, outperforming GPT-4o and two-stage RL agents by relative improvements of 23.0% and 10.9%, respectively. Notably, MaKTO also demonstrates human-like performance, winning 60% against expert players and showing only 49% detectability in Turing-style blind tests. These results showcase MaKTO's superior decision-making, strategic adaptation, and natural language generation in complex social deduction games.

This paper introduces Werewolf Arena, a novel framework for evaluating large language models (LLMs) through the lens of the classic social deduction game, Werewolf. In Werewolf Arena, LLMs compete against each other, navigating the game's complex dynamics of deception, deduction, and persuasion. The framework introduces a dynamic turn-taking system based on bidding, mirroring real-world discussions where individuals strategically choose when to speak. We demonstrate the framework's utility through an arena-style tournament featuring Gemini and GPT models. Our results reveal distinct strengths and weaknesses in the models' strategic reasoning and communication. These findings highlight Werewolf Arena's potential as a challenging and scalable LLM benchmark.

Communication is a fundamental aspect of human society, facilitating the exchange of information and beliefs among people. Despite the advancements in large language models (LLMs), recent agents built with these often neglect the control over discussion tactics, which are essential in communication scenarios and games. As a variant of the famous communication game Werewolf, One Night Ultimate Werewolf (ONUW) requires players to develop strategic discussion policies due to the potential role changes that increase the uncertainty and complexity of the game. In this work, we first present the existence of the Perfect Bayesian Equilibria (PBEs) in two scenarios of the ONUW game: one with discussion and one without. The results showcase that the discussion greatly changes players' utilities by affecting their beliefs, emphasizing the significance of discussion tactics. Based on the insights obtained from the analyses, we propose an RL-instructed language agent framework, where a discussion policy trained by reinforcement learning (RL) is employed to determine appropriate discussion tactics to adopt. Our experimental results on several ONUW game settings demonstrate the effectiveness and generalizability of our proposed framework.

In this context, the proliferation of fake news and its impact on society has become a matter of serious concern, and it is critical to understand the mechanisms involved. In this study, we specifically explore how the proliferation of fake news is affected by the strategic behavior and interaction dynamics of individuals. In a scenario where a single werewolf is present, we show that certain agents can have a significant impact on group dynamics by manipulating the flow of information. This result suggests a role for "opinion leaders" or "influencers" in the spread of fake news, and the detection of these agents and the mitigation of their influence may be key to understanding and controlling the dynamics of information dissemination. We have developed models of interactions between individuals and the propagation of information using the framework of incomplete information games and unfolding Figure 1: Network Graph with a Single Werewolf games. In particular, we used the concepts of cocktail party effect and repetition dilemma to analyze how the complexity many other agents an agent interacts with, and the repetition of the decisions agents face and their position in the dilemma represents the balance between an agent's incentives social network affect the spread of fake news and the gains to act cooperatively and non-cooperatively. of individual agents.

Large language models (LLMs) have exhibited memorable strategic behaviors in social deductive games. However, the significance of opinion leadership exhibited by LLM-based agents has been overlooked, which is crucial for practical applications in multi-agent and human-AI interaction settings. Opinion leaders are individuals who have a noticeable impact on the beliefs and behaviors of others within a social group. In this work, we employ the Werewolf game as a simulation platform to assess the opinion leadership of LLMs. The game features the role of the Sheriff, tasked with summarizing arguments and recommending decision options, and therefore serves as a credible proxy for an opinion leader. We develop a framework integrating the Sheriff role and devise two novel metrics for evaluation based on the critical characteristics of opinion leaders. The first metric measures the reliability of the opinion leader, and the second assesses the influence of the opinion leader on other players' decisions. We conduct extensive experiments to evaluate LLMs of different scales. In addition, we collect a Werewolf question-answering dataset (WWQA) to assess and enhance LLM's grasp of the game rules, and we also incorporate human participants for further analysis. The results suggest that the Werewolf game is a suitable test bed to evaluate the opinion leadership of LLMs and few LLMs possess the capacity for opinion leadership.

Agents built with large language models (LLMs) have recently achieved great advancements. However, most of the efforts focus on single-agent or cooperative settings, leaving more general multi-agent environments underexplored. We propose a new framework powered by reinforcement learning (RL) to develop strategic language agents, i.e., LLM-based agents with strategic thinking ability, for a popular language game, Werewolf. Werewolf is a social deduction game with hidden roles that involves both cooperation and competition and emphasizes deceptive communication and diverse gameplay. Our agent tackles this game by first using LLMs to reason about potential deceptions and generate a set of strategically diverse actions. Then an RL policy, which selects an action from the candidates, is learned by population-based training to enhance the agents' decision-making ability. By combining LLMs with the RL policy, our agent produces a variety of emergent strategies, achieves the highest win rate against other LLM-based agents, and stays robust against adversarial human players in the Werewolf game.