MetaMind: General and Cognitive World Models in Multi-Agent Systems by Meta-Theory of Mind

2026-02-28

AI Summary

This work proposes MetaMind, a cognitive world model grounded in meta-theory of mind (Meta-ToM), to address key limitations in multi-agent systems—namely, insufficient understanding of interdependent dynamics, poor trajectory prediction, and the absence of long-term coordinated planning without centralized supervision. MetaMind enables agents to perform zero-shot theory-of-mind reasoning and adapt to collective intentions without explicit communication, leveraging reflexive bidirectional inference. It introduces a novel metacognitive transfer mechanism that shifts perspective from first-person to third-person, integrated with self-supervised learning and analogical reasoning to achieve efficient few-shot generalization. Experimental results demonstrate that MetaMind significantly outperforms existing baselines across diverse multi-agent tasks, exhibiting both superior performance and robust generalization capabilities.

Technology Category

• Multiagent Systems: Multiagent Learning
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Agentic search

Abstract

A major challenge for world models in multi-agent systems is to understand interdependent agent dynamics, predict interactive multi-agent trajectories, and plan over long horizons with collective awareness, without centralized supervision or explicit communication. In this paper, MetaMind, a general and cognitive world model for multi-agent systems that leverages a novel meta-theory of mind (Meta-ToM) framework, is proposed. Through MetaMind, each agent learns not only to predict and plan over its own beliefs, but also to inversely reason goals and beliefs from its own behavior trajectories. This self-reflective, bidirectional inference loop enables each agent to learn a metacognitive ability in a self-supervised manner. Then, MetaMind is shown to generalize the metacognitive ability from first-person to third-person through analogical reasoning. Thus, in multi-agent systems, each agent with MetaMind can actively reason about goals and beliefs of other agents from limited, observable behavior trajectories in a zero-shot manner, and then adapt to emergent collective intention without an explicit communication mechanism. Extended simulation results on diverse multi-agent tasks demonstrate that MetaMind can achieve superior task performance and outperform baselines in few-shot multi-agent generalization.

Problem

• multi-agent systems
• world models
• theory of mind
• collective intention
• zero-shot reasoning

Innovation

• Meta-Theory of Mind
• World Model
• Multi-Agent Systems
• Metacognition
• Zero-Shot Generalization