@inproceedings{yang-etal-2026-acttraitbench,
title = "{A}ct{T}rait{B}ench: Quantifying the Knowledge-Decision Gap in Large Language Models via Human-Grounded Behavioral Validation",
author = "Yang, Yutong and
Miao, Chenxi and
Li, Weikang and
Wu, Yunfang",
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.2068/",
pages = "44018--44030",
ISBN = "979-8-89176-462-0",
abstract = "While Large Language Models (LLMs) can convincingly simulate personas in explicit self-reports, they often deviate in implicit behavioral decisions, revealing a substantial Knowledge{--}Decision Gap (GKD). Existing benchmarks struggle to measure this discrepancy due to limited construct validity, multidimensional entanglement, and distributional biases in LLM-based evaluation. To address these issues, we propose ActTraitBench, a human-grounded evaluation framework for measuring personality consistency in LLMs. Grounded in empirical human data, ActTraitBench establishes one-to-one mappings between psychometric facets and behavioral paradigms and applies Distributional Calibration via Quantile Mapping to reduce distributional mismatch between LLM-judge scores and human responses. Experiments on 14 mainstream LLMs reveal substantial knowledge{--}decision gaps and show that assigned personas are reflected more consistently in self-reports than in behavioral decisions for most evaluated models. To mitigate this gap, we further introduce the Chain of Cognitive Alignment (CoCA), an inference-time intervention that reduces GKD for 12 of the 13 models with paired results. Code and resources are available at https://github.com/Selina233/ActTraitBench."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yang-etal-2026-acttraitbench">
<titleInfo>
<title>ActTraitBench: Quantifying the Knowledge-Decision Gap in Large Language Models via Human-Grounded Behavioral Validation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yutong</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenxi</namePart>
<namePart type="family">Miao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weikang</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yunfang</namePart>
<namePart type="family">Wu</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>While Large Language Models (LLMs) can convincingly simulate personas in explicit self-reports, they often deviate in implicit behavioral decisions, revealing a substantial Knowledge–Decision Gap (GKD). Existing benchmarks struggle to measure this discrepancy due to limited construct validity, multidimensional entanglement, and distributional biases in LLM-based evaluation. To address these issues, we propose ActTraitBench, a human-grounded evaluation framework for measuring personality consistency in LLMs. Grounded in empirical human data, ActTraitBench establishes one-to-one mappings between psychometric facets and behavioral paradigms and applies Distributional Calibration via Quantile Mapping to reduce distributional mismatch between LLM-judge scores and human responses. Experiments on 14 mainstream LLMs reveal substantial knowledge–decision gaps and show that assigned personas are reflected more consistently in self-reports than in behavioral decisions for most evaluated models. To mitigate this gap, we further introduce the Chain of Cognitive Alignment (CoCA), an inference-time intervention that reduces GKD for 12 of the 13 models with paired results. Code and resources are available at https://github.com/Selina233/ActTraitBench.</abstract>
<identifier type="citekey">yang-etal-2026-acttraitbench</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2068/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>44018</start>
<end>44030</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ActTraitBench: Quantifying the Knowledge-Decision Gap in Large Language Models via Human-Grounded Behavioral Validation
%A Yang, Yutong
%A Miao, Chenxi
%A Li, Weikang
%A Wu, Yunfang
%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 yang-etal-2026-acttraitbench
%X While Large Language Models (LLMs) can convincingly simulate personas in explicit self-reports, they often deviate in implicit behavioral decisions, revealing a substantial Knowledge–Decision Gap (GKD). Existing benchmarks struggle to measure this discrepancy due to limited construct validity, multidimensional entanglement, and distributional biases in LLM-based evaluation. To address these issues, we propose ActTraitBench, a human-grounded evaluation framework for measuring personality consistency in LLMs. Grounded in empirical human data, ActTraitBench establishes one-to-one mappings between psychometric facets and behavioral paradigms and applies Distributional Calibration via Quantile Mapping to reduce distributional mismatch between LLM-judge scores and human responses. Experiments on 14 mainstream LLMs reveal substantial knowledge–decision gaps and show that assigned personas are reflected more consistently in self-reports than in behavioral decisions for most evaluated models. To mitigate this gap, we further introduce the Chain of Cognitive Alignment (CoCA), an inference-time intervention that reduces GKD for 12 of the 13 models with paired results. Code and resources are available at https://github.com/Selina233/ActTraitBench.
%U https://aclanthology.org/2026.findings-emnlp.2068/
%P 44018-44030
Markdown (Informal)
[ActTraitBench: Quantifying the Knowledge-Decision Gap in Large Language Models via Human-Grounded Behavioral Validation](https://aclanthology.org/2026.findings-emnlp.2068/) (Yang et al., Findings 2026)
ACL