<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="luo-etal-2026-tia">
    <titleInfo>
        <title>TIA-VL: Diagnosing and Correcting Template-Induced Ambiguity in Medical Vision-Language Decisions</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Yifan</namePart>
        <namePart type="family">Luo</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jiaran</namePart>
        <namePart type="family">Shi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Guoming</namePart>
        <namePart type="family">Wang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Siliang</namePart>
        <namePart type="family">Tang</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
    </relatedItem>
    <abstract>Medical vision-language models (VLMs) are often evaluated by report retrieval, grounding, and explanation quality. We show that these metrics can create a metric illusion: a model may appear clinically grounded while exploiting acquisition template, crop, orientation, or protocol rather than patient anatomy. We formalize this failure as template-induced ambiguity and introduce TIA-VL, a counterfactual benchmark with source, template-family, severity, and compound protocol shifts. We also propose Rationale-Grounded Template Ambiguity Calibration (R-GTAC), a plug-in calibration layer that discounts template-dominated evidence while preserving anatomy and textual rationale support. Across strong medical VLM-style baselines, including BioViL-T-style frozen features, adapter features, and anatomy-teacher features, R-GTAC reduces wrong same-template errors to 0.0000 on the main protocol splits. We then test whether systems know when to abstain through readiness and counterfactual rationale deferral. Counterfactual rationale readiness filtering further identifies a reliable low-risk 25% subset reaching 0.8500 Top-1 on the hardest compound shift, while revealing that coupled protocol shifts pose critical frontiers for global risk calibration. Thus, TIA-VL supports a conservative message for medical VLM evaluation: models should be judged not only by whether they answer, but by whether their apparent rationales survive anti-shortcut counterfactuals. The code and benchmark will be available at https://github.com/YifanLuo0103/TIA-VL.</abstract>
    <identifier type="citekey">luo-etal-2026-tia</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.538/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>11303</start>
            <end>11319</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T TIA-VL: Diagnosing and Correcting Template-Induced Ambiguity in Medical Vision-Language Decisions
%A Luo, Yifan
%A Shi, Jiaran
%A Wang, Guoming
%A Tang, Siliang
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F luo-etal-2026-tia
%X Medical vision-language models (VLMs) are often evaluated by report retrieval, grounding, and explanation quality. We show that these metrics can create a metric illusion: a model may appear clinically grounded while exploiting acquisition template, crop, orientation, or protocol rather than patient anatomy. We formalize this failure as template-induced ambiguity and introduce TIA-VL, a counterfactual benchmark with source, template-family, severity, and compound protocol shifts. We also propose Rationale-Grounded Template Ambiguity Calibration (R-GTAC), a plug-in calibration layer that discounts template-dominated evidence while preserving anatomy and textual rationale support. Across strong medical VLM-style baselines, including BioViL-T-style frozen features, adapter features, and anatomy-teacher features, R-GTAC reduces wrong same-template errors to 0.0000 on the main protocol splits. We then test whether systems know when to abstain through readiness and counterfactual rationale deferral. Counterfactual rationale readiness filtering further identifies a reliable low-risk 25% subset reaching 0.8500 Top-1 on the hardest compound shift, while revealing that coupled protocol shifts pose critical frontiers for global risk calibration. Thus, TIA-VL supports a conservative message for medical VLM evaluation: models should be judged not only by whether they answer, but by whether their apparent rationales survive anti-shortcut counterfactuals. The code and benchmark will be available at https://github.com/YifanLuo0103/TIA-VL.
%U https://aclanthology.org/2026.findings-emnlp.538/
%P 11303-11319

Markdown (Informal)

[TIA-VL: Diagnosing and Correcting Template-Induced Ambiguity in Medical Vision-Language Decisions](https://aclanthology.org/2026.findings-emnlp.538/) (Luo et al., Findings 2026)

• TIA-VL: Diagnosing and Correcting Template-Induced Ambiguity in Medical Vision-Language Decisions (Luo et al., Findings 2026)

ACL

• Yifan Luo, Jiaran Shi, Guoming Wang, and Siliang Tang. 2026. TIA-VL: Diagnosing and Correcting Template-Induced Ambiguity in Medical Vision-Language Decisions. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 11303–11319, Budapest, Hungary. Association for Computational Linguistics.