潜在 token 真的在思考吗？对 COCONUT 连续思维链的因果与对抗分析

View PDF HTML (experimental)

Abstract:Latent tokens are gaining attention for enhancing reasoning in large language models (LLMs), yet their internal mechanisms remain unclear. This paper examines the problem from a reliability perspective, uncovering fundamental weaknesses: latent tokens function as uninterpretable placeholders rather than encoding faithful reasoning. While resistant to perturbation, they promote shortcut usage over genuine reasoning. We focus on Chain-of-Continuous-Thought (COCONUT), which claims better efficiency and stability than explicit Chain-of-Thought (CoT) while maintaining performance. We investigate this through two complementary approaches. First, steering experiments perturb specific token subsets, namely COCONUT and explicit CoT. Unlike CoT tokens, COCONUT tokens show minimal sensitivity to steering and lack reasoning-critical information. Second, shortcut experiments evaluate models under biased and out-of-distribution settings. Results on MMLU and HotpotQA demonstrate that COCONUT consistently exploits dataset artifacts, inflating benchmark performance without true reasoning. These findings reposition COCONUT as a pseudo-reasoning mechanism: it generates plausible traces that conceal shortcut dependence rather than faithfully representing reasoning processes.

Comments:	Accepted to EMNLP 2026 Main Conference. 14 pages, 5 figures
Subjects:	Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
ACM classes:	I.2.7; K.6.5
Cite as:	arXiv:2512.21711 [cs.CL]
	(or arXiv:2512.21711v2 [cs.CL] for this version)
	https://doi.org/10.48550/arXiv.2512.21711 arXiv-issued DOI via DataCite

Submission history

From: Yuyi Zhang [view email]
[v1] Thu, 25 Dec 2025 15:14:53 UTC (930 KB)
[v2] Mon, 26 Oct 2026 02:40:17 UTC (921 KB)