@inproceedings{dong-etal-2026-thinking,
title = "Thinking Without Speaking: Efficient Tree of Thoughts Reasoning via Latent Steering",
author = "Dong, Zhanshuo and
Wen, Zhenyu and
Li, Xiong and
Su, Jie and
Qin, Jianbin and
Yuan, Ye and
Hong, Zhen",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.224/",
pages = "5182--5202",
ISBN = "979-8-89176-459-0",
abstract = "Tree of Thoughts (ToT) reasoning is fundamentally bottlenecked by its generate-then-score paradigm, which decodes every candidate branch into text and evaluates it through a decoupled scoring stage. Prior efficiency efforts primarily shrink the search breadth, leaving this prohibitive per-step textual evaluation intact. To resolve this, we propose ITOT (Implicit Tree of Thoughts), a training-free framework that recasts tree-structured reasoning as a continuous search within a model{'}s latent space. In this framework, branch quality is assessed intrinsically from hidden states via alignment with a contrastive prototype distilled from successful and failed traces. This elegantly collapses explicit branch enumeration and textual scoring into a single latent generate-evaluate-decide loop. Across six reasoning benchmarks and five LLM backbones, ITOT consistently improves mean accuracy over both the strongest tree-search and single-path baselines by roughly ten absolute points. Crucially, it achieves this while cutting token consumption by nearly an order of magnitude relative to explicit tree-search methods, operating even below the token budget of vanilla Chain of Thought. We will release our code to facilitate further research."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dong-etal-2026-thinking">
<titleInfo>
<title>Thinking Without Speaking: Efficient Tree of Thoughts Reasoning via Latent Steering</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhanshuo</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhenyu</namePart>
<namePart type="family">Wen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiong</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Su</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianbin</namePart>
<namePart type="family">Qin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ye</namePart>
<namePart type="family">Yuan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Hong</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>Tree of Thoughts (ToT) reasoning is fundamentally bottlenecked by its generate-then-score paradigm, which decodes every candidate branch into text and evaluates it through a decoupled scoring stage. Prior efficiency efforts primarily shrink the search breadth, leaving this prohibitive per-step textual evaluation intact. To resolve this, we propose ITOT (Implicit Tree of Thoughts), a training-free framework that recasts tree-structured reasoning as a continuous search within a model’s latent space. In this framework, branch quality is assessed intrinsically from hidden states via alignment with a contrastive prototype distilled from successful and failed traces. This elegantly collapses explicit branch enumeration and textual scoring into a single latent generate-evaluate-decide loop. Across six reasoning benchmarks and five LLM backbones, ITOT consistently improves mean accuracy over both the strongest tree-search and single-path baselines by roughly ten absolute points. Crucially, it achieves this while cutting token consumption by nearly an order of magnitude relative to explicit tree-search methods, operating even below the token budget of vanilla Chain of Thought. We will release our code to facilitate further research.</abstract>
<identifier type="citekey">dong-etal-2026-thinking</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.224/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>5182</start>
<end>5202</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Thinking Without Speaking: Efficient Tree of Thoughts Reasoning via Latent Steering
%A Dong, Zhanshuo
%A Wen, Zhenyu
%A Li, Xiong
%A Su, Jie
%A Qin, Jianbin
%A Yuan, Ye
%A Hong, Zhen
%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 dong-etal-2026-thinking
%X Tree of Thoughts (ToT) reasoning is fundamentally bottlenecked by its generate-then-score paradigm, which decodes every candidate branch into text and evaluates it through a decoupled scoring stage. Prior efficiency efforts primarily shrink the search breadth, leaving this prohibitive per-step textual evaluation intact. To resolve this, we propose ITOT (Implicit Tree of Thoughts), a training-free framework that recasts tree-structured reasoning as a continuous search within a model’s latent space. In this framework, branch quality is assessed intrinsically from hidden states via alignment with a contrastive prototype distilled from successful and failed traces. This elegantly collapses explicit branch enumeration and textual scoring into a single latent generate-evaluate-decide loop. Across six reasoning benchmarks and five LLM backbones, ITOT consistently improves mean accuracy over both the strongest tree-search and single-path baselines by roughly ten absolute points. Crucially, it achieves this while cutting token consumption by nearly an order of magnitude relative to explicit tree-search methods, operating even below the token budget of vanilla Chain of Thought. We will release our code to facilitate further research.
%U https://aclanthology.org/2026.emnlp-main.224/
%P 5182-5202
Markdown (Informal)
[Thinking Without Speaking: Efficient Tree of Thoughts Reasoning via Latent Steering](https://aclanthology.org/2026.emnlp-main.224/) (Dong et al., EMNLP 2026)
ACL