# Meta-Moderator: Empowering Multi-Agent Debate with Meta-Cognition

arXiv: https://arxiv.org/abs/2608.23029 · HTML: https://mlsanity.com/paper/2608.23029/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 23 of its 37 references that are in the map.
- First posted on arXiv 24 Aug 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM agents and tool use / [Social simulation and multi-agent debate](https://mlsanity.com/agents/social-simulation-and-multi-agent-debate/)

## Builds on (most cited first, 23 of 23)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Qwen Technical Report](https://mlsanity.com/paper/2309.16609/) (2023)
- [Improving Factuality and Reasoning in Language Models through Multiagent Debate](https://mlsanity.com/paper/2305.14325/) (2023)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate](https://mlsanity.com/paper/2305.19118/) (2023)
- [Did Aristotle Use a Laptop? A Question Answering Benchmark with Implicit Reasoning Strategies](https://mlsanity.com/paper/2101.02235/) (2021)
- [ToolRL: Reward is All Tool Learning Needs](https://mlsanity.com/paper/2504.13958/) (2025)
- [Metacognitive Prompting Improves Understanding in Large Language Models](https://mlsanity.com/paper/2308.05342/) (2023)
- [Language Models Are Capable of Metacognitive Monitoring and Control of Their Internal Activations](https://mlsanity.com/paper/2505.13763/) (2025)
- [Adaptive Tool Use in Large Language Models with Meta-Cognition Trigger](https://mlsanity.com/paper/2502.12961/) (2025)
- [Removal of Hallucination on Hallucination: Debate-Augmented RAG](https://mlsanity.com/paper/2505.18581/) (2025)
- [Free-MAD: Consensus-Free Multi-Agent Debate](https://mlsanity.com/paper/2509.11035/) (2025)
- [Multi-Agent Large Language Models for Conversational Task-Solving](https://mlsanity.com/paper/2410.22932/) (2024)
- [Multi-Agent Debate for LLM Judges with Adaptive Stability Detection](https://mlsanity.com/paper/2510.12697/) (2025)
- [A Picture Is Worth a Graph: A Blueprint Debate Paradigm for Multimodal Reasoning](https://mlsanity.com/paper/2403.14972/) (2024)
- [SyncLoop: A Multimodal Dual-Loop Framework for Self-Improving Mathematical Reasoning](https://mlsanity.com/paper/2507.16518/) (2025)
- [Meta-Reasoning Improves Tool Use in Large Language Models](https://mlsanity.com/paper/2411.04535/) (2024)
- [QualBench: Benchmarking Chinese LLMs with Localized Professional Qualifications for Vertical Domain Evaluation](https://mlsanity.com/paper/2505.05225/) (2025)
- [Multi-Agent Collaborative Intelligence: Dual-Dial Control for Reliable LLM Reasoning](https://mlsanity.com/paper/2510.04488/) (2025)

## Abstract (arXiv)

Multi-agent debate can improve large language model reasoning by eliciting diverse hypotheses and critiques, yet its performance is often constrained by weak moderation. Common pipelines rely on fixed budgets, agreement-based stopping, or untrained judges, leading to redundant deliberation and unreliable evidence aggregation. We cast moderation as a meta-cognitive process, monitoring debate utility, controlling deliberation, and adjudicating a final answer, and introduce Meta-Moderator, a learnable framework that dynamically regulates debate and decides when to finalize an answer. Meta-Moderator is trained independently of the debaters via outcome-driven policy optimization, making debate regulation an explicit capability rather than an incidental effect of prompting. Across five benchmarks, Meta-Moderator outperforms widely used decision layers and transfers across tasks and system configurations. Further analyses show that it allocates debate more selectively and reduces mis-aggregation after informative hypotheses appear.