潜在标记是否思考？对连续思维链的因果和对抗分析
Do Latent Tokens Think? A Causal and Adversarial Analysis of Chain-of-Continuous-Thought
AI总结：
本文通过因果和对抗分析揭示COCONUT作为伪推理机制的问题，指出其依赖捷径而非真实推理。
摘要：
潜在标记正在被关注以增强大语言模型（LLMs）的推理能力，但其内部机制仍不清楚。本文从可靠性角度出发，揭示了基本弱点：潜在标记作为不可解释的占位符而非编码忠实推理。尽管对扰动具有抵抗力，但它们促进捷径使用而非真正的推理。我们专注于连续思维链（COCONUT），它声称比显式连续思维（CoT）更高效和稳定，同时保持性能。我们通过两种互补的方法进行研究。首先，引导实验扰动特定标记子集，即COCONUT和显式CoT。与CoT标记不同，COCONUT标记对引导的敏感性较低，缺乏推理关键信息。其次，捷径实验评估模型在偏置和分布外设置下的表现。在MMLU和HotpotQA上的结果表明，COCONUT一致利用数据集的瑕疵，夸大基准性能而不具备真正的推理。这些发现将COCONUT重新定位为伪推理机制：它生成合理的轨迹，隐藏捷径依赖，而不是忠实代表推理过程。
英文摘要：
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.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
