<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dey-etal-2026-palms">
    <titleInfo>
        <title>PALMs: Using Multi-Construct Grounded Rationales for Modeling Population Preferences in LLMs</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Priyanka</namePart>
        <namePart type="family">Dey</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Brihi</namePart>
        <namePart type="family">Joshi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Preyashi</namePart>
        <namePart type="family">Poddar</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jieyu</namePart>
        <namePart type="family">Zhao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Emilio</namePart>
        <namePart type="family">Ferrara</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>Large language models are being extensively used to simulate individual user behavior, yet faithfully representing a population requires capturing the systematic variation in values, beliefs, and cultural norms that distinguish one group from another. We introduce (), a suite of models each aligned to specific populations, covering five countries: USA, India, Brazil, France and Italy. are created by synthesizing rationales grounded in psychological and cultural constructs and using these as latent supervision during preference tuning for population-specific alignment. Evaluated across four dimensions: personality, values and beliefs, cultural norms, and morality, consistently outperform baselines, including culture-specialized models, achieving an average of 8.59% relative improvement over the best baseline across all five populations. Notably, construct-grounded rationales outperform both demographic prompting and survey-based fine-tuning, suggesting that grounding preference learning in psychology and culture provides a richer inductive signal than surface-level response distributions. We further demonstrate strong generalization to downstream applications without task-specific supervision: outperforming best baselines by 5.19% in personalized reward modeling, 6.34% in population simulation, and showing strong transfer to social reasoning tasks. Datasets and code are available at: \hrefhttps://github.com/limenlp/PALMshttps://github.com/limenlp/PALMs.</abstract>
    <identifier type="citekey">dey-etal-2026-palms</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.888/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>20475</start>
            <end>20517</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T PALMs: Using Multi-Construct Grounded Rationales for Modeling Population Preferences in LLMs
%A Dey, Priyanka
%A Joshi, Brihi
%A Poddar, Preyashi
%A Zhao, Jieyu
%A Ferrara, Emilio
%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 dey-etal-2026-palms
%X Large language models are being extensively used to simulate individual user behavior, yet faithfully representing a population requires capturing the systematic variation in values, beliefs, and cultural norms that distinguish one group from another. We introduce (), a suite of models each aligned to specific populations, covering five countries: USA, India, Brazil, France and Italy. are created by synthesizing rationales grounded in psychological and cultural constructs and using these as latent supervision during preference tuning for population-specific alignment. Evaluated across four dimensions: personality, values and beliefs, cultural norms, and morality, consistently outperform baselines, including culture-specialized models, achieving an average of 8.59% relative improvement over the best baseline across all five populations. Notably, construct-grounded rationales outperform both demographic prompting and survey-based fine-tuning, suggesting that grounding preference learning in psychology and culture provides a richer inductive signal than surface-level response distributions. We further demonstrate strong generalization to downstream applications without task-specific supervision: outperforming best baselines by 5.19% in personalized reward modeling, 6.34% in population simulation, and showing strong transfer to social reasoning tasks. Datasets and code are available at: \hrefhttps://github.com/limenlp/PALMshttps://github.com/limenlp/PALMs.
%U https://aclanthology.org/2026.emnlp-main.888/
%P 20475-20517

Markdown (Informal)

[PALMs: Using Multi-Construct Grounded Rationales for Modeling Population Preferences in LLMs](https://aclanthology.org/2026.emnlp-main.888/) (Dey et al., EMNLP 2026)

• PALMs: Using Multi-Construct Grounded Rationales for Modeling Population Preferences in LLMs (Dey et al., EMNLP 2026)

ACL

• Priyanka Dey, Brihi Joshi, Preyashi Poddar, Jieyu Zhao, and Emilio Ferrara. 2026. PALMs: Using Multi-Construct Grounded Rationales for Modeling Population Preferences in LLMs. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 20475–20517, Budapest, Hungary. Association for Computational Linguistics.