Metraipolis › archive › dc_6ba2ff3f4486
Does Latent Chain-of-Thought Actually Reason? A Single, Unifying Taxonomy for Latent Reasoning — arXiv:2610.21182
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.21182
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T09:02:15.819Z
- SHA-256 of the archived text
0da8db5c37d2d5c13b83d58779cb66a8a341524d684e519601a8a46e442d1f38- Stable URL of this page
https://metraipolis.com/documento/dc_6ba2ff3f4486
How to cite
Does Latent Chain-of-Thought Actually Reason? A Single, Unifying Taxonomy for Latent Reasoning — arXiv:2610.21182 — arXiv (en), https://arxiv.org/abs/2610.21182 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6ba2ff3f4486 (Metraipolis archive dc_6ba2ff3f4486, sha256 0da8db5c37d2d5c13b83d58779cb66a8a341524d684e519601a8a46e442d1f38)
BibTeX
@misc{dc6ba2ff3f4486,
title = {Does Latent Chain-of-Thought Actually Reason? A Single, Unifying Taxonomy for Latent Reasoning — arXiv:2610.21182},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.21182},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6ba2ff3f4486; sha256 0da8db5c37d2d5c13b83d58779cb66a8a341524d684e519601a8a46e442d1f38},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21182. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Latent chain-of-thought replaces written reasoning steps with continuous hidden states, giving models additional steps to compute before answering. It remains unclear what computation these steps actually perform. High task accuracy and decodable intermediate answers do not establish that a model uses its latent chain to reach an answer. We study this question across twenty-one model-task configurations covering three training recipes, three reasoning tasks, and two backbones. We combine chain deletion and content replacement with probes and trajectory analysis to test whether the chain affects the answer, what it encodes, and how its states change. On logical tasks, COCONUT often carries an answer that is already readable before the chain begins. Its states settle early, and deleting the chain leaves predictions almost unchanged. Per-step supervision produces an ordered layout of reasoning steps and can make the model depend on the chain. Yet even an ordered chain can act as a placeholder, with deletion hurting performance while content replacement leaves answers unchanged. On arithmetic tasks, all tested recipes show dependence on chain content. These results show why readable latent states alone are insufficient evidence of reasoning. We introduce a taxonomy that separates what a chain represents from how the model uses it. Seen this way, earlier reports no longer compete: placeholder tokens, collapsed superposition and step-dependent causal roles are each consistent with a different kind of chain, each obtained on a different task, training recipe or measurement. Once a study says which property it tested, whether the model uses the chain, what the chain holds, or how its state moves, claims about latent reasoning can be compared. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.21182 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T08:44:17.712Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).