UserHarness: Harnessing User Minds for Stronger Agent Theory-of-Mind

2026-05-26

AI Summary

Existing agents struggle to explicitly model users’ mental states in theory-of-mind (ToM) tasks, often relying on indirect behavioral inference. This work proposes UserHarness, a novel framework that, for the first time, explicitly decomposes user mental states into observable and trackable components, establishing a closed-loop mechanism for belief updating, intention generation, and environmental interaction that supports nested belief reasoning. Built with a modular architecture, UserHarness enables end-to-end reconstruction of psychological states from the user’s perspective without requiring complex pipelines. Evaluated across five benchmark tasks, the framework achieves a macro accuracy of up to 95.94%, yielding relative performance improvements exceeding 15% over current reasoning-based methods and over 20% compared to the strongest prompting-based approaches.

Abstract

Understanding what a user believes and intends is central to building effective agent assistants. This ability is often evaluated through Theory-of-Mind (ToM) tasks, where success requires reasoning from the user's perspective. However, many existing approaches address ToM with complex pipelines that model behavior indirectly, without explicitly reconstructing the user's mental state. This misses the core structure of the problem: users act based on their beliefs, which are updated through observations of the environment; beliefs and intentions jointly determine actions, which in turn change the environment; and social reasoning often requires nested beliefs about what others believe or intend. We propose UserHarness, a simple framework that reframes ToM reasoning as explicit user-mind reconstruction. UserHarness decomposes the user's mental state, its relation to the external environment, and the actions that follow from it, enabling agents to track what the user observes, believes, intends, and does. Across five benchmarks, UserHarness reaches up to 95.94% macro accuracy, improving over existing inference methods by more than 15% relative and over the strongest prompt-only harness by about 20% relative. These results suggest that robust user understanding requires reasoning from the roots of the user's mind, positioning user harnessing as a promising foundation for more adaptive future assistants.

Problem

• Theory-of-Mind
• user mental state
• belief modeling
• intention understanding
• social reasoning

Innovation

• Theory-of-Mind
• user-mind reconstruction
• belief-intention-action modeling
• nested beliefs
• agent reasoning