@inproceedings{zhang-etal-2026-debate-aware,
title = "Debate-aware Learning Makes {LLM}s Genuine Debaters",
author = "Zhang, Yufei and
Xiong, Kai and
Yan, Yuxiong and
Cai, Bibo and
Ding, Xiao and
Qin, Bing and
Liu, Ting",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1907/",
pages = "43450--43471",
ISBN = "979-8-89176-459-0",
abstract = "Multi-agent debate systems are used to competitively discuss and resolve complex tasks. However, existing approaches often exhibit pseudo-debate behavior: agents either ignore opponents' arguments or compromise too readily without critical evaluation. To address this, we propose a three-stage debate framework grounded in abstract argumentation theory. First, Debate Structuring uses principled prompts to elicit well-formed debate interactions. Second, Debate Imitation filters high-quality debates to fine-tune LLMs on proper argumentative conduct. Third, Debate Internalization employs a debate-aware reinforcement learning objective to instill adversarial reasoning patterns. To better quantify debate capability, we introduce two novel metrics: Rebuttal Engagement Score (RES), assessing how thoroughly models engage with opposing arguments, and Belief Governance Score (BGS), evaluating whether models appropriately retain or revise beliefs based on argument strength. Extensive evaluations across six benchmarks show that our framework not only improves task performance but also cultivates more authentic and robust debate behavior in LLMs. Our work contributes a theory-inspired training framework and evaluation suite for advancing multi-agent debate systems."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-debate-aware">
<titleInfo>
<title>Debate-aware Learning Makes LLMs Genuine Debaters</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yufei</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai</namePart>
<namePart type="family">Xiong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuxiong</namePart>
<namePart type="family">Yan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bibo</namePart>
<namePart type="family">Cai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiao</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bing</namePart>
<namePart type="family">Qin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ting</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Multi-agent debate systems are used to competitively discuss and resolve complex tasks. However, existing approaches often exhibit pseudo-debate behavior: agents either ignore opponents’ arguments or compromise too readily without critical evaluation. To address this, we propose a three-stage debate framework grounded in abstract argumentation theory. First, Debate Structuring uses principled prompts to elicit well-formed debate interactions. Second, Debate Imitation filters high-quality debates to fine-tune LLMs on proper argumentative conduct. Third, Debate Internalization employs a debate-aware reinforcement learning objective to instill adversarial reasoning patterns. To better quantify debate capability, we introduce two novel metrics: Rebuttal Engagement Score (RES), assessing how thoroughly models engage with opposing arguments, and Belief Governance Score (BGS), evaluating whether models appropriately retain or revise beliefs based on argument strength. Extensive evaluations across six benchmarks show that our framework not only improves task performance but also cultivates more authentic and robust debate behavior in LLMs. Our work contributes a theory-inspired training framework and evaluation suite for advancing multi-agent debate systems.</abstract>
<identifier type="citekey">zhang-etal-2026-debate-aware</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1907/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>43450</start>
<end>43471</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Debate-aware Learning Makes LLMs Genuine Debaters
%A Zhang, Yufei
%A Xiong, Kai
%A Yan, Yuxiong
%A Cai, Bibo
%A Ding, Xiao
%A Qin, Bing
%A Liu, Ting
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F zhang-etal-2026-debate-aware
%X Multi-agent debate systems are used to competitively discuss and resolve complex tasks. However, existing approaches often exhibit pseudo-debate behavior: agents either ignore opponents’ arguments or compromise too readily without critical evaluation. To address this, we propose a three-stage debate framework grounded in abstract argumentation theory. First, Debate Structuring uses principled prompts to elicit well-formed debate interactions. Second, Debate Imitation filters high-quality debates to fine-tune LLMs on proper argumentative conduct. Third, Debate Internalization employs a debate-aware reinforcement learning objective to instill adversarial reasoning patterns. To better quantify debate capability, we introduce two novel metrics: Rebuttal Engagement Score (RES), assessing how thoroughly models engage with opposing arguments, and Belief Governance Score (BGS), evaluating whether models appropriately retain or revise beliefs based on argument strength. Extensive evaluations across six benchmarks show that our framework not only improves task performance but also cultivates more authentic and robust debate behavior in LLMs. Our work contributes a theory-inspired training framework and evaluation suite for advancing multi-agent debate systems.
%U https://aclanthology.org/2026.emnlp-main.1907/
%P 43450-43471
Markdown (Informal)
[Debate-aware Learning Makes LLMs Genuine Debaters](https://aclanthology.org/2026.emnlp-main.1907/) (Zhang et al., EMNLP 2026)
ACL
- Yufei Zhang, Kai Xiong, Yuxiong Yan, Bibo Cai, Xiao Ding, Bing Qin, and Ting Liu. 2026. Debate-aware Learning Makes LLMs Genuine Debaters. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 43450–43471, Budapest, Hungary. Association for Computational Linguistics.