Hypothetical Minds: Scaffolding Theory of Mind for Multi-Agent Tasks with Large Language Models

TL;DR

This paper introduces Hypothetical Minds, an LLM-based multi-agent system with a Theory of Mind module that improves online adaptation and strategic reasoning in multi-agent environments, outperforming previous baselines.

Contribution

It presents a cognitively-inspired architecture with hypothesis generation and refinement, enabling LLM agents to better handle non-stationarity and novel agents in multi-agent tasks.

Findings

Significant performance improvements on Melting Pot benchmark

Effective hypothesis evaluation and refinement mechanisms

Enhanced adaptability in diverse multi-agent scenarios

Abstract

Multi-agent reinforcement learning (MARL) methods struggle with the non-stationarity of multi-agent systems and fail to adaptively learn online when tested with novel agents. Here, we leverage large language models (LLMs) to create an autonomous agent that can handle these challenges. Our agent, Hypothetical Minds, consists of a cognitively-inspired architecture, featuring modular components for perception, memory, and hierarchical planning over two levels of abstraction. We introduce the Theory of Mind module that scaffolds the high-level planning process by generating hypotheses about other agents' strategies in natural language. It then evaluates and iteratively refines these hypotheses by reinforcing hypotheses that make correct predictions about the other agents' behavior. Hypothetical Minds significantly improves performance over previous LLM-agent and RL baselines on a range of competitive, mixed motive, and collaborative domains in the Melting Pot benchmark, including both dyadic and population-based environments. Additionally, comparisons against LLM-agent baselines and ablations reveal the importance of hypothesis evaluation and refinement for succeeding on complex scenarios.