@inproceedings{floro-benedetto-2026-cultural,
title = "Cultural Binding Heads in Language Models",
author = "Floro, Avrile and
Benedetto, Luca",
editor = "Arad, Dana and
Belinkov, Yonatan and
Chen, Hanjie and
Feldhus, Nils and
Mueller, Aaron and
Mohebbi, Hosein and
Sarti, Gabriele and
Tutek, Martin",
booktitle = "Proceedings of the 9th {B}lackbox{NLP} Workshop: Analyzing and Interpreting Neural Networks for {NLP}",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.blackboxnlp-1.3/",
pages = "34--48",
ISBN = "979-8-89176-476-7",
abstract = "LLMs often default to equal treatment across cultural groups, even though context warrants differentiation: this is a lack of *difference awareness*. Using mechanistic interpretability and a factorial design on the N4 cultural appropriation benchmark from Wang et al. (2025), we identify 2{--}3 mid-layer attention heads per model that contribute causally to cultural binding across eight models (base and instruct versions of four architectures). Cultural binding is the process of associating a cultural item with its related identity. Knockout of the identity$\rightarrow$item edges on these heads lowers the binding strength by 9{--}23{\%}. The identified heads transfer from instruct to base models, suggesting that cultural binding is created during pre-training. An $\alpha$-scaling shows a graded dose{--}response. Moderate amplification steering at generation ($\alpha = 2${--}3) increases cultural differentiation accuracy by 1{--}3 pp while leaving reasoning on culturally neutral questions mostly intact. A knowledge probing task shows that models know 3{--}$6\times$ more than they act upon, indicating that the bottleneck lies in routing and not knowledge."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="floro-benedetto-2026-cultural">
<titleInfo>
<title>Cultural Binding Heads in Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Avrile</namePart>
<namePart type="family">Floro</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Luca</namePart>
<namePart type="family">Benedetto</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 9th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP</title>
</titleInfo>
<name type="personal">
<namePart type="given">Dana</namePart>
<namePart type="family">Arad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yonatan</namePart>
<namePart type="family">Belinkov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hanjie</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nils</namePart>
<namePart type="family">Feldhus</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aaron</namePart>
<namePart type="family">Mueller</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hosein</namePart>
<namePart type="family">Mohebbi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gabriele</namePart>
<namePart type="family">Sarti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Martin</namePart>
<namePart type="family">Tutek</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">979-8-89176-476-7</identifier>
</relatedItem>
<abstract>LLMs often default to equal treatment across cultural groups, even though context warrants differentiation: this is a lack of *difference awareness*. Using mechanistic interpretability and a factorial design on the N4 cultural appropriation benchmark from Wang et al. (2025), we identify 2–3 mid-layer attention heads per model that contribute causally to cultural binding across eight models (base and instruct versions of four architectures). Cultural binding is the process of associating a cultural item with its related identity. Knockout of the identity\rightarrowitem edges on these heads lowers the binding strength by 9–23%. The identified heads transfer from instruct to base models, suggesting that cultural binding is created during pre-training. An α-scaling shows a graded dose–response. Moderate amplification steering at generation (α = 2–3) increases cultural differentiation accuracy by 1–3 pp while leaving reasoning on culturally neutral questions mostly intact. A knowledge probing task shows that models know 3–6\times more than they act upon, indicating that the bottleneck lies in routing and not knowledge.</abstract>
<identifier type="citekey">floro-benedetto-2026-cultural</identifier>
<location>
<url>https://aclanthology.org/2026.blackboxnlp-1.3/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>34</start>
<end>48</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Cultural Binding Heads in Language Models
%A Floro, Avrile
%A Benedetto, Luca
%Y Arad, Dana
%Y Belinkov, Yonatan
%Y Chen, Hanjie
%Y Feldhus, Nils
%Y Mueller, Aaron
%Y Mohebbi, Hosein
%Y Sarti, Gabriele
%Y Tutek, Martin
%S Proceedings of the 9th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-476-7
%F floro-benedetto-2026-cultural
%X LLMs often default to equal treatment across cultural groups, even though context warrants differentiation: this is a lack of *difference awareness*. Using mechanistic interpretability and a factorial design on the N4 cultural appropriation benchmark from Wang et al. (2025), we identify 2–3 mid-layer attention heads per model that contribute causally to cultural binding across eight models (base and instruct versions of four architectures). Cultural binding is the process of associating a cultural item with its related identity. Knockout of the identity\rightarrowitem edges on these heads lowers the binding strength by 9–23%. The identified heads transfer from instruct to base models, suggesting that cultural binding is created during pre-training. An α-scaling shows a graded dose–response. Moderate amplification steering at generation (α = 2–3) increases cultural differentiation accuracy by 1–3 pp while leaving reasoning on culturally neutral questions mostly intact. A knowledge probing task shows that models know 3–6\times more than they act upon, indicating that the bottleneck lies in routing and not knowledge.
%U https://aclanthology.org/2026.blackboxnlp-1.3/
%P 34-48
Markdown (Informal)
[Cultural Binding Heads in Language Models](https://aclanthology.org/2026.blackboxnlp-1.3/) (Floro & Benedetto, BlackboxNLP 2026)
ACL
- Avrile Floro and Luca Benedetto. 2026. Cultural Binding Heads in Language Models. In Proceedings of the 9th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP, pages 34–48, Budapest, Hungary. Association for Computational Linguistics.