Revac: A Social Deduction Reasoning Agent
Summary: arXiv:2604.19523v1 Announce Type: new
Abstract: Social deduction games such as Mafia present a unique AI challenge: players must reason under uncertainty, interpret incomplete and intentionally misleading information, evaluate human-like communication, and make strategic elimination decisions. Unlike deterministic board games, success in Mafia depends not on perfect information or brute-force search, but on inference, memory, and adaptability in the presence of deception.
This work presents the design and evaluation of Revac-8, an AI agent developed for the Social Deduction track of the MindGames Arena competition, where it achieved first place. The final agent evolved from a simple two-stage reasoning system into a multi-module architecture that integrates memory-based player profiling, social-graph analysis of accusations and defenses, and dynamic tone selection for communication. These results highlight the importance of structured memory and adaptive communication for achieving strong performance in high-stakes social environments.
Introduction to Social Deduction Games
Social deduction games, such as Mafia, involve multiple players working to uncover the identity of hidden roles within the game. These games are characterized by:
- Uncertainty: Players often lack complete information about other players’ roles.
- Deception: Players may provide misleading information to manipulate others.
- Communication: Effective interaction is crucial for forming alliances and making strategic decisions.
The Development of Revac-8
Revac-8 represents a significant advancement in AI capabilities for social deduction contexts. Its development process involved several key stages:
- Initial Design: The agent began with a two-stage reasoning model that focused primarily on basic decision-making.
- Evolution to Multi-Module Architecture: As the competition progressed, the agent was enhanced to include various modules that improved its functionality.
- Memory-Based Player Profiling: Revac-8 employs a system to track and analyze player behaviors over time, allowing it to adapt its strategies based on opponents’ past actions.
- Social-Graph Analysis: This feature enables the agent to evaluate the relationships between players based on accusations and defenses, helping it to identify potential alliances and threats.
- Dynamic Tone Selection: Revac-8 can adjust its communication style to influence other players, enhancing its ability to deceive and persuade.
Achieving Success in the MindGames Arena
Revac-8’s advanced reasoning capabilities were put to the test in the MindGames Arena competition, where it outperformed other participants. The agent’s success can be attributed to its:
- Robust Inference Skills: The ability to draw conclusions from limited information.
- Adaptive Memory Utilization: Leveraging past interactions to inform present decisions.
- Effective Communication Strategies: Tailoring messages based on the context to maximize influence.
Conclusion
The development of Revac-8 marks a significant stride in the realm of AI for social deduction games. Its innovative approach, combining memory, social analysis, and adaptive communication, sets a new standard for future agents in complex, high-stakes environments. As AI continues to evolve, the insights gained from Revac-8’s performance will undoubtedly inform the design of even more sophisticated systems capable of navigating the intricacies of human-like reasoning and interaction.