Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation

TL;DR

This paper addresses lazy agent behavior in multi-agent LLM reasoning systems, proposing methods to measure influence and encourage deliberation, thereby enhancing collaboration and reasoning performance.

Contribution

It introduces a causal influence measurement and a verifiable reward mechanism to mitigate lazy behavior and improve multi-agent reasoning collaboration.

Findings

Mitigates lazy agent behavior in multi-agent systems

Enhances reasoning accuracy through deliberation mechanisms

Improves collaboration and task performance in complex reasoning

Abstract

Large Language Models (LLMs) trained with reinforcement learning and verifiable rewards have achieved strong results on complex reasoning tasks. Recent work extends this paradigm to a multi-agent setting, where a meta-thinking agent proposes plans and monitors progress while a reasoning agent executes subtasks through sequential conversational turns. Despite promising performance, we identify a critical limitation: lazy agent behavior, in which one agent dominates while the other contributes little, undermining collaboration and collapsing the setup to an ineffective single agent. In this paper, we first provide a theoretical analysis showing why lazy behavior naturally arises in multi-agent reasoning. We then introduce a stable and efficient method for measuring causal influence, helping mitigate this issue. Finally, as collaboration intensifies, the reasoning agent risks getting lost in multi-turn interactions and trapped by previous noisy responses. To counter this, we propose a verifiable reward mechanism that encourages deliberation by allowing the reasoning agent to discard noisy outputs, consolidate instructions, and restart its reasoning process when necessary. Extensive experiments demonstrate that our framework alleviates lazy agent behavior and unlocks the full potential of multi-agent framework for complex reasoning tasks.