@inproceedings{wu-etal-2026-perceive,
title = "Perceive, Localize, Reflect: Evidence-Anchored Multi-Agent Debate for {MLLM} Hallucination Mitigation",
author = "Wu, Yebo and
Jin, Han and
Lim, Wei Yang Bryan and
Li, Li",
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.1597/",
pages = "36298--36313",
ISBN = "979-8-89176-459-0",
abstract = "Multimodal Large Language Models (MLLMs) demonstrate remarkable reasoning capabilities, yet they remain susceptible to hallucinations. While Multi-Agent Debate (MAD) presents a promising collaborative solution, unguided debates inevitably devolve into a stochastic martingale process and expanding conversational contexts trigger severe visual attention decay. In this paper, we propose EviaMAD, an Evidence-Anchored Multi-Agent Debate framework. To proactively expose internal hallucinations, EviaMAD bifurcates perception into two distinct roles: a Global Observer capturing the holistic scene and a Local Observer scrutinizing salient details. When conflicts arise, a dedicated Analyzer intervenes, distilling the linguistic disagreement into a deterministic bimodal evidence anchor: a physically cropped visual patch of the disputed region paired with key semantic directives. During the subsequent reflection phase, a surgical reallocation mechanism explicitly amplifies the attention weights of this anchor. By compelling agents to ground their reasoning in strictly verified evidence, EviaMAD effectively elicits collective intelligence. Extensive evaluations demonstrate that EviaMAD significantly mitigates hallucinations while elevating the reasoning capabilities of the multi-agent system."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2026-perceive">
<titleInfo>
<title>Perceive, Localize, Reflect: Evidence-Anchored Multi-Agent Debate for MLLM Hallucination Mitigation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yebo</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Han</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="given">Yang</namePart>
<namePart type="given">Bryan</namePart>
<namePart type="family">Lim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Li</namePart>
<namePart type="family">Li</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>Multimodal Large Language Models (MLLMs) demonstrate remarkable reasoning capabilities, yet they remain susceptible to hallucinations. While Multi-Agent Debate (MAD) presents a promising collaborative solution, unguided debates inevitably devolve into a stochastic martingale process and expanding conversational contexts trigger severe visual attention decay. In this paper, we propose EviaMAD, an Evidence-Anchored Multi-Agent Debate framework. To proactively expose internal hallucinations, EviaMAD bifurcates perception into two distinct roles: a Global Observer capturing the holistic scene and a Local Observer scrutinizing salient details. When conflicts arise, a dedicated Analyzer intervenes, distilling the linguistic disagreement into a deterministic bimodal evidence anchor: a physically cropped visual patch of the disputed region paired with key semantic directives. During the subsequent reflection phase, a surgical reallocation mechanism explicitly amplifies the attention weights of this anchor. By compelling agents to ground their reasoning in strictly verified evidence, EviaMAD effectively elicits collective intelligence. Extensive evaluations demonstrate that EviaMAD significantly mitigates hallucinations while elevating the reasoning capabilities of the multi-agent system.</abstract>
<identifier type="citekey">wu-etal-2026-perceive</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1597/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>36298</start>
<end>36313</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Perceive, Localize, Reflect: Evidence-Anchored Multi-Agent Debate for MLLM Hallucination Mitigation
%A Wu, Yebo
%A Jin, Han
%A Lim, Wei Yang Bryan
%A Li, Li
%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 wu-etal-2026-perceive
%X Multimodal Large Language Models (MLLMs) demonstrate remarkable reasoning capabilities, yet they remain susceptible to hallucinations. While Multi-Agent Debate (MAD) presents a promising collaborative solution, unguided debates inevitably devolve into a stochastic martingale process and expanding conversational contexts trigger severe visual attention decay. In this paper, we propose EviaMAD, an Evidence-Anchored Multi-Agent Debate framework. To proactively expose internal hallucinations, EviaMAD bifurcates perception into two distinct roles: a Global Observer capturing the holistic scene and a Local Observer scrutinizing salient details. When conflicts arise, a dedicated Analyzer intervenes, distilling the linguistic disagreement into a deterministic bimodal evidence anchor: a physically cropped visual patch of the disputed region paired with key semantic directives. During the subsequent reflection phase, a surgical reallocation mechanism explicitly amplifies the attention weights of this anchor. By compelling agents to ground their reasoning in strictly verified evidence, EviaMAD effectively elicits collective intelligence. Extensive evaluations demonstrate that EviaMAD significantly mitigates hallucinations while elevating the reasoning capabilities of the multi-agent system.
%U https://aclanthology.org/2026.emnlp-main.1597/
%P 36298-36313
Markdown (Informal)
[Perceive, Localize, Reflect: Evidence-Anchored Multi-Agent Debate for MLLM Hallucination Mitigation](https://aclanthology.org/2026.emnlp-main.1597/) (Wu et al., EMNLP 2026)
ACL