<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lewis-lim-etal-2025-analysing">
    <titleInfo>
        <title>Analysing Chain of Thought Dynamics: Active Guidance or Unfaithful Post-hoc Rationalisation?</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Samuel</namePart>
        <namePart type="family">Lewis-Lim</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xingwei</namePart>
        <namePart type="family">Tan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Zhixue</namePart>
        <namePart type="family">Zhao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Nikolaos</namePart>
        <namePart type="family">Aletras</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-11</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Christos</namePart>
            <namePart type="family">Christodoulopoulos</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Tanmoy</namePart>
            <namePart type="family">Chakraborty</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Carolyn</namePart>
            <namePart type="family">Rose</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Violet</namePart>
            <namePart type="family">Peng</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Suzhou, China</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-332-6</identifier>
    </relatedItem>
    <abstract>Recent work has demonstrated that using chain of thought (CoT), on soft-reasoning problems such as analytical and commonsense reasoning, often yields limited or even negative performance gains. CoT can also be unfaithful to the model’s actual reasoning. This paper investigates dynamics and unfaithfulness of CoT in soft-reasoning tasks across instruction-tuned, reasoning and reasoning-distilled models. Our findings show that distilled-reasoning models rely heavily on CoT for these tasks, while instruction-tuned and reasoning models often use it post-hoc. Additionally, we find that CoT can steer model predictions without faithfully reflecting reasoning, indicating a disconnect between CoT influence and faithfulness.</abstract>
    <identifier type="citekey">lewis-lim-etal-2025-analysing</identifier>
    <location>
        <url>https://aclanthology.org/2025.emnlp-main.1516/</url>
    </location>
    <part>
        <date>2025-11</date>
        <extent unit="page">
            <start>29826</start>
            <end>29841</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Analysing Chain of Thought Dynamics: Active Guidance or Unfaithful Post-hoc Rationalisation?
%A Lewis-Lim, Samuel
%A Tan, Xingwei
%A Zhao, Zhixue
%A Aletras, Nikolaos
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-332-6
%F lewis-lim-etal-2025-analysing
%X Recent work has demonstrated that using chain of thought (CoT), on soft-reasoning problems such as analytical and commonsense reasoning, often yields limited or even negative performance gains. CoT can also be unfaithful to the model’s actual reasoning. This paper investigates dynamics and unfaithfulness of CoT in soft-reasoning tasks across instruction-tuned, reasoning and reasoning-distilled models. Our findings show that distilled-reasoning models rely heavily on CoT for these tasks, while instruction-tuned and reasoning models often use it post-hoc. Additionally, we find that CoT can steer model predictions without faithfully reflecting reasoning, indicating a disconnect between CoT influence and faithfulness.
%U https://aclanthology.org/2025.emnlp-main.1516/
%P 29826-29841

Markdown (Informal)

[Analysing Chain of Thought Dynamics: Active Guidance or Unfaithful Post-hoc Rationalisation?](https://aclanthology.org/2025.emnlp-main.1516/) (Lewis-Lim et al., EMNLP 2025)

• Analysing Chain of Thought Dynamics: Active Guidance or Unfaithful Post-hoc Rationalisation? (Lewis-Lim et al., EMNLP 2025)

ACL

• Samuel Lewis-Lim, Xingwei Tan, Zhixue Zhao, and Nikolaos Aletras. 2025. Analysing Chain of Thought Dynamics: Active Guidance or Unfaithful Post-hoc Rationalisation?. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 29826–29841, Suzhou, China. Association for Computational Linguistics.