@inproceedings{vural-etal-2026-learning,
title = "Learning to Think Like a Cartoon Captionist: Incongruity{--}Resolution Supervision for Multimodal Humor Understanding",
author = {Vural, Hatice Merve and
Kukul, Do{\u{g}}a and
{\"O}zl{\"u}, Ege Erdem and
Ar{\i}kan, Demir Ekin and
Mankoff, Bob and
Erdem, Erkut and
Erdem, Aykut},
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.1994/",
pages = "45402--45442",
ISBN = "979-8-89176-459-0",
abstract = "Humor is one of the few cognitive tasks where getting the reasoning right matters as much as getting the answer right. While recent work evaluates humor understanding on benchmarks such as the New Yorker Cartoon Caption Contest (NYCC), it largely treats it as black-box prediction, overlooking the structured reasoning processes underlying humor comprehension. We introduce IRS (Incongruity{--}Resolution Supervision), a framework that decomposes humor understanding into three components: Incongruity Modeling, which identifies mismatches in the visual scene; Resolution Modeling, which constructs coherent reinterpretations of these mismatches; and Preference Alignment, which evaluates candidate interpretations under human judgments. Grounded in incongruity-resolution theory and expert captionist practice, IRS supervises intermediate reasoning process through structured traces that make the path from visual perception to humorous interpretation explicit and learnable. Across 7B, 32B, and 72B models on NYCC, IRS improves performance across caption matching and ranking, with IRS-72B achieving the strongest model performance on ranking (76.10{\%}), surpassing both non-expert human performance and all evaluated open- and closed-source multimodal baselines. Zero-shot transfer further shows that IRS learns generalizable reasoning patterns."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="vural-etal-2026-learning">
<titleInfo>
<title>Learning to Think Like a Cartoon Captionist: Incongruity–Resolution Supervision for Multimodal Humor Understanding</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hatice</namePart>
<namePart type="given">Merve</namePart>
<namePart type="family">Vural</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Doğa</namePart>
<namePart type="family">Kukul</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ege</namePart>
<namePart type="given">Erdem</namePart>
<namePart type="family">Özlü</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Demir</namePart>
<namePart type="given">Ekin</namePart>
<namePart type="family">Arıkan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bob</namePart>
<namePart type="family">Mankoff</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Erkut</namePart>
<namePart type="family">Erdem</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aykut</namePart>
<namePart type="family">Erdem</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>Humor is one of the few cognitive tasks where getting the reasoning right matters as much as getting the answer right. While recent work evaluates humor understanding on benchmarks such as the New Yorker Cartoon Caption Contest (NYCC), it largely treats it as black-box prediction, overlooking the structured reasoning processes underlying humor comprehension. We introduce IRS (Incongruity–Resolution Supervision), a framework that decomposes humor understanding into three components: Incongruity Modeling, which identifies mismatches in the visual scene; Resolution Modeling, which constructs coherent reinterpretations of these mismatches; and Preference Alignment, which evaluates candidate interpretations under human judgments. Grounded in incongruity-resolution theory and expert captionist practice, IRS supervises intermediate reasoning process through structured traces that make the path from visual perception to humorous interpretation explicit and learnable. Across 7B, 32B, and 72B models on NYCC, IRS improves performance across caption matching and ranking, with IRS-72B achieving the strongest model performance on ranking (76.10%), surpassing both non-expert human performance and all evaluated open- and closed-source multimodal baselines. Zero-shot transfer further shows that IRS learns generalizable reasoning patterns.</abstract>
<identifier type="citekey">vural-etal-2026-learning</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1994/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>45402</start>
<end>45442</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Learning to Think Like a Cartoon Captionist: Incongruity–Resolution Supervision for Multimodal Humor Understanding
%A Vural, Hatice Merve
%A Kukul, Doğa
%A Özlü, Ege Erdem
%A Arıkan, Demir Ekin
%A Mankoff, Bob
%A Erdem, Erkut
%A Erdem, Aykut
%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 vural-etal-2026-learning
%X Humor is one of the few cognitive tasks where getting the reasoning right matters as much as getting the answer right. While recent work evaluates humor understanding on benchmarks such as the New Yorker Cartoon Caption Contest (NYCC), it largely treats it as black-box prediction, overlooking the structured reasoning processes underlying humor comprehension. We introduce IRS (Incongruity–Resolution Supervision), a framework that decomposes humor understanding into three components: Incongruity Modeling, which identifies mismatches in the visual scene; Resolution Modeling, which constructs coherent reinterpretations of these mismatches; and Preference Alignment, which evaluates candidate interpretations under human judgments. Grounded in incongruity-resolution theory and expert captionist practice, IRS supervises intermediate reasoning process through structured traces that make the path from visual perception to humorous interpretation explicit and learnable. Across 7B, 32B, and 72B models on NYCC, IRS improves performance across caption matching and ranking, with IRS-72B achieving the strongest model performance on ranking (76.10%), surpassing both non-expert human performance and all evaluated open- and closed-source multimodal baselines. Zero-shot transfer further shows that IRS learns generalizable reasoning patterns.
%U https://aclanthology.org/2026.emnlp-main.1994/
%P 45402-45442
Markdown (Informal)
[Learning to Think Like a Cartoon Captionist: Incongruity–Resolution Supervision for Multimodal Humor Understanding](https://aclanthology.org/2026.emnlp-main.1994/) (Vural et al., EMNLP 2026)
ACL