Attention-Aligned Reasoning for Large Language Models

Hongxiang Zhang; Yuan Tian; Tianyi Zhang

doi:10.18653/2026.findings-emnlp.1185

Attention-Aligned Reasoning for Large Language Models

Hongxiang Zhang, Yuan Tian, Tianyi Zhang

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use ... for bold, ... for italic, ... for underline, <sc>...</sc> for small-caps, <tt>...<tt> for typewriter text, <url>...</url> for URLs, <a href=...> for hyperlinks, and <par/> for paragraph breaks.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

Large Language Models (LLMs) tend to generate a long reasoning chain when solving complex tasks. However, as the reasoning chain extends, critical intermediate steps and the original prompt will be buried in the context, receiving insufficient attention and leading to errors. In this work, we present ATAR, a novel reasoning method that leverages the inherent reasoning structure to steer LLM attention. Our experiments show that ATAR outperforms SOTA methods across six benchmarks, achieving up to 15.39% absolute improvement. Furthermore, with ATAR, “non-reasoning” models achieve comparable or even better performance compared to reasoning models of the same size in most benchmarks. Finally, our ablation studies show that the attention alignment component contributes significantly, and that these improvements are persist under different attention-steering backends.

Anthology ID:: 2026.findings-emnlp.1185
Volume:: Findings of the Association for Computational Linguistics: EMNLP 2026
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Hung-Yi Lee, Sunipa Dev, Jessy Li, Desmond Elliott
Venue:: Findings
SIG:: SIGDAT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 25046–25064
Language:
URL:: https://aclanthology.org/2026.findings-emnlp.1185/
DOI:: 10.18653/2026.findings-emnlp.1185
Bibkey:
Cite (ACL):: Hongxiang Zhang, Yuan Tian, and Tianyi Zhang. 2026. Attention-Aligned Reasoning for Large Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 25046–25064, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: Attention-Aligned Reasoning for Large Language Models (Zhang et al., Findings 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.findings-emnlp.1185.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{zhang-etal-2026-attention,
    title = "Attention-Aligned Reasoning for Large Language Models",
    author = "Zhang, Hongxiang  and
      Tian, Yuan  and
      Zhang, Tianyi",
    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.1185/",
    doi = "10.18653/2026.findings-emnlp.1185",
    pages = "25046--25064",
    ISBN = "979-8-89176-462-0",
    abstract = "Large Language Models (LLMs) tend to generate a long reasoning chain when solving complex tasks. However, as the reasoning chain extends, critical intermediate steps and the original prompt will be buried in the context, receiving insufficient attention and leading to errors. In this work, we present ATAR, a novel reasoning method that leverages the inherent reasoning structure to steer LLM attention. Our experiments show that ATAR outperforms SOTA methods across six benchmarks, achieving up to 15.39{\%} absolute improvement. Furthermore, with ATAR, ``non-reasoning'' models achieve comparable or even better performance compared to reasoning models of the same size in most benchmarks. Finally, our ablation studies show that the attention alignment component contributes significantly, and that these improvements are persist under different attention-steering backends."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-attention">
    <titleInfo>
        <title>Attention-Aligned Reasoning for Large Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Hongxiang</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yuan</namePart>
        <namePart type="family">Tian</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Tianyi</namePart>
        <namePart type="family">Zhang</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>Large Language Models (LLMs) tend to generate a long reasoning chain when solving complex tasks. However, as the reasoning chain extends, critical intermediate steps and the original prompt will be buried in the context, receiving insufficient attention and leading to errors. In this work, we present ATAR, a novel reasoning method that leverages the inherent reasoning structure to steer LLM attention. Our experiments show that ATAR outperforms SOTA methods across six benchmarks, achieving up to 15.39% absolute improvement. Furthermore, with ATAR, “non-reasoning” models achieve comparable or even better performance compared to reasoning models of the same size in most benchmarks. Finally, our ablation studies show that the attention alignment component contributes significantly, and that these improvements are persist under different attention-steering backends.</abstract>
    <identifier type="citekey">zhang-etal-2026-attention</identifier>
    <identifier type="doi">10.18653/2026.findings-emnlp.1185</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.1185/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>25046</start>
            <end>25064</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T Attention-Aligned Reasoning for Large Language Models
%A Zhang, Hongxiang
%A Tian, Yuan
%A Zhang, Tianyi
%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 zhang-etal-2026-attention
%X Large Language Models (LLMs) tend to generate a long reasoning chain when solving complex tasks. However, as the reasoning chain extends, critical intermediate steps and the original prompt will be buried in the context, receiving insufficient attention and leading to errors. In this work, we present ATAR, a novel reasoning method that leverages the inherent reasoning structure to steer LLM attention. Our experiments show that ATAR outperforms SOTA methods across six benchmarks, achieving up to 15.39% absolute improvement. Furthermore, with ATAR, “non-reasoning” models achieve comparable or even better performance compared to reasoning models of the same size in most benchmarks. Finally, our ablation studies show that the attention alignment component contributes significantly, and that these improvements are persist under different attention-steering backends.
%R 10.18653/2026.findings-emnlp.1185
%U https://aclanthology.org/2026.findings-emnlp.1185/
%U https://doi.org/10.18653/2026.findings-emnlp.1185
%P 25046-25064

Download as File

Markdown (Informal)

[Attention-Aligned Reasoning for Large Language Models](https://aclanthology.org/2026.findings-emnlp.1185/) (Zhang et al., Findings 2026)

Attention-Aligned Reasoning for Large Language Models (Zhang et al., Findings 2026)

ACL

Hongxiang Zhang, Yuan Tian, and Tianyi Zhang. 2026. Attention-Aligned Reasoning for Large Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 25046–25064, Budapest, Hungary. Association for Computational Linguistics.