Seeing Isn't Believing: Mitigating Belief Inertia via Active Intervention in Embodied Agents

TL;DR

This paper introduces the EVU mechanism to actively manage and update agents' beliefs during embodied tasks, significantly improving success rates by mitigating belief inertia.

Contribution

It proposes a novel EVU intervention method that explicitly predicts, verifies, and updates beliefs, enhancing robustness of embodied agents against belief inertia.

Findings

EVU consistently improves task success rates across benchmarks.

The approach effectively reduces belief inertia in embodied agents.

EVU can be integrated into various reasoning frameworks.

Abstract

Recent advancements in large language models (LLMs) have enabled agents to tackle complex embodied tasks through environmental interaction. However, these agents still make suboptimal decisions and perform ineffective actions, as they often overlook critical environmental feedback that differs from their internal beliefs. Through a formal probing analysis, we characterize this as belief inertia, a phenomenon where agents stubbornly adhere to prior beliefs despite explicit observations. To address this, we advocate active belief intervention, moving from passive understanding to active management. We introduce the Estimate-Verify-Update (EVU) mechanism, which empowers agents to predict expected outcomes, verify them against observations through explicit reasoning, and actively update prior beliefs based on the verification evidence. EVU is designed as a unified intervention mechanism that generates textual belief states explicitly, and can be integrated into both prompting-based and training-based agent reasoning methods. Extensive experiments across three embodied benchmarks demonstrate that EVU consistently yields substantial gains in task success rates. Further analyses validate that our approach effectively mitigates belief inertia, advancing the development of more robust embodied agents. Our code is available at https://github.com/WangHanLinHenry/EVU.