On Information Self-Locking in Reinforcement Learning for Active Reasoning of LLM agents

TL;DR

This paper investigates the problem of information self-locking in reinforcement learning-trained LLM agents during active reasoning, identifies core causes, and proposes a method to improve information exploration, resulting in significant performance gains.

Contribution

The paper introduces a decomposition of active reasoning into Action Selection and Belief Tracking, and proposes a critique-based approach to mitigate information self-locking in RL-trained LLM agents.

Findings

Up to 60% improvement in mitigating information self-locking

Identified feedback loop limiting information exploration

Effective approach across 7 datasets

Abstract

Reinforcement learning (RL) with outcome-based rewards has achieved significant success in training large language model (LLM) agents for complex reasoning tasks. However, in active reasoning where agents need to strategically ask questions to acquire task-relevant information, we find that LLM agents trained with RL often suffer from information self-locking: the agent ceases to ask informative questions and struggles to internalize already-obtained information. To understand the phenomenon, we decompose active reasoning into two core capabilities: Action Selection (AS), which determines the observation stream through queries, and Belief Tracking (BT), which updates the agent's belief based on collected evidence. We show that deficient AS and BT capabilities will limit the information exploration during RL training. Furthermore, insufficient exploration in turn hinders the improvement of AS and BT, creating a feedback loop that locks the agent in a low-information regime. To resolve the issue, we propose a simple yet effective approach that reallocates the learning signal by injecting easy- to-obtain directional critiques to help the agent escape self-locking. Extensive experiments with 7 datasets show that our approach significantly mitigates the information self-locking, bringing up to 60% improvements.