@inproceedings{wu-etal-2026-trajtom,
title = "{T}raj{T}o{M}: Evaluating Theory of Mind over Mental State Trajectories in Social Dialogues",
author = "Wu, Chenglong and
Wang, Xi and
Liao, Jinzhi and
Zhao, Xiang",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1775/",
doi = "10.18653/2026.findings-emnlp.1775",
pages = "37968--37990",
ISBN = "979-8-89176-462-0",
abstract = "Theory of Mind (ToM) is crucial for understanding social dialogue, where characters' mental states are rarely stated explicitly but evolve through interaction and shape later behavior. Existing ToM benchmarks focus mainly on selected dialogue turns or state changes across separate scenarios, leaving within-dialogue mental state trajectories underexplored. We introduce TrajToM, a benchmark for evaluating whether LLMs can reason over evolving emotions, beliefs, intentions, and desires in continuous social dialogue. Built with a Sketch-first construction paradigm, TrajToM contains 700 dialogues and 18,165 multiple-choice questions across mental state reasoning, behavioral reasoning, and trajectory reasoning. Experiments reveal a substantial human-model gap, with GPT-5.2 reaching 67.4{\%} overall accuracy compared with 88.2{\%} for humans, and show persistent bottlenecks in second-order reasoning, behavior explanation, and change direction assessment. We further examine MARTA as an exploratory trajectory-aligned process-supervision framework, showing that explicit supervision over mental state trajectories can improve ToM reasoning."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2026-trajtom">
<titleInfo>
<title>TrajToM: Evaluating Theory of Mind over Mental State Trajectories in Social Dialogues</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chenglong</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xi</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinzhi</namePart>
<namePart type="family">Liao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiang</namePart>
<namePart type="family">Zhao</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>Theory of Mind (ToM) is crucial for understanding social dialogue, where characters’ mental states are rarely stated explicitly but evolve through interaction and shape later behavior. Existing ToM benchmarks focus mainly on selected dialogue turns or state changes across separate scenarios, leaving within-dialogue mental state trajectories underexplored. We introduce TrajToM, a benchmark for evaluating whether LLMs can reason over evolving emotions, beliefs, intentions, and desires in continuous social dialogue. Built with a Sketch-first construction paradigm, TrajToM contains 700 dialogues and 18,165 multiple-choice questions across mental state reasoning, behavioral reasoning, and trajectory reasoning. Experiments reveal a substantial human-model gap, with GPT-5.2 reaching 67.4% overall accuracy compared with 88.2% for humans, and show persistent bottlenecks in second-order reasoning, behavior explanation, and change direction assessment. We further examine MARTA as an exploratory trajectory-aligned process-supervision framework, showing that explicit supervision over mental state trajectories can improve ToM reasoning.</abstract>
<identifier type="citekey">wu-etal-2026-trajtom</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1775</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1775/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>37968</start>
<end>37990</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TrajToM: Evaluating Theory of Mind over Mental State Trajectories in Social Dialogues
%A Wu, Chenglong
%A Wang, Xi
%A Liao, Jinzhi
%A Zhao, Xiang
%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 wu-etal-2026-trajtom
%X Theory of Mind (ToM) is crucial for understanding social dialogue, where characters’ mental states are rarely stated explicitly but evolve through interaction and shape later behavior. Existing ToM benchmarks focus mainly on selected dialogue turns or state changes across separate scenarios, leaving within-dialogue mental state trajectories underexplored. We introduce TrajToM, a benchmark for evaluating whether LLMs can reason over evolving emotions, beliefs, intentions, and desires in continuous social dialogue. Built with a Sketch-first construction paradigm, TrajToM contains 700 dialogues and 18,165 multiple-choice questions across mental state reasoning, behavioral reasoning, and trajectory reasoning. Experiments reveal a substantial human-model gap, with GPT-5.2 reaching 67.4% overall accuracy compared with 88.2% for humans, and show persistent bottlenecks in second-order reasoning, behavior explanation, and change direction assessment. We further examine MARTA as an exploratory trajectory-aligned process-supervision framework, showing that explicit supervision over mental state trajectories can improve ToM reasoning.
%R 10.18653/2026.findings-emnlp.1775
%U https://aclanthology.org/2026.findings-emnlp.1775/
%U https://doi.org/10.18653/2026.findings-emnlp.1775
%P 37968-37990
Markdown (Informal)
[TrajToM: Evaluating Theory of Mind over Mental State Trajectories in Social Dialogues](https://aclanthology.org/2026.findings-emnlp.1775/) (Wu et al., Findings 2026)
ACL