摘要：潜在推理模型（Latent Reasoning Models，LRM）因其较低的推理成本（相对于显式推理模型）以及在理论上能够并行探索多条推理路径的能力，受到了研究界的广泛关注。然而，这些优势的代价是可解释性下降：由于 LRM 不使用自然语言进行推理，因此难以对其进行监控。本文通过考察两种最先进的 LRM，研究其可解释性。首先，我们发现，潜在推理 token 对 LRM 的预测通常并非必需；在逻辑推理数据集上，LRM 几乎总能在完全不使用潜在推理的情况下，生成相同的最终答案。对推理 token 的这种利用不足，或许部分解释了为何 LRM 并未始终优于显式推理方法，也对既有研究中这些 token 被赋予的作用提出了疑问。其次，我们证明，当潜在推理 token 对性能确有必要时，对于预测正确的样本，我们能够在 65% 至 93% 的情况下解码出标准推理轨迹。这表明，LRM 往往执行的是预期的解题过程，而非不可解释的推理过程。最后，我们提出了一种方法：无需事先知道标准推理轨迹，即可从潜在 token 中解码出经过验证的自然语言推理轨迹。结果显示，对于大多数正确预测，可以找到经过验证的推理轨迹；而对于错误预测，只能在少数情况下找到此类轨迹。我们的发现强调，当前 LRM 在很大程度上编码了可解释的过程，而可解释性本身可以成为预测正确性的一个信号。

Abstract:Latent reasoning models (LRMs) have attracted significant research interest due to their low inference cost (relative to explicit reasoning models) and theoretical ability to explore multiple reasoning paths in parallel. However, these benefits come at the cost of reduced interpretability: LRMs are difficult to monitor because they do not reason in natural language. This paper presents an investigation into LRM interpretability by examining two state-of-the-art LRMs. First, we find that latent reasoning tokens are often unnecessary for LRMs' predictions; on logical reasoning datasets, LRMs can almost always produce the same final answers without using latent reasoning at all. This underutilization of reasoning tokens may partially explain why LRMs do not consistently outperform explicit reasoning methods and raises doubts about the stated role of these tokens in prior work. Second, we demonstrate that when latent reasoning tokens are necessary for performance, we can decode gold reasoning traces up to 65-93% of the time for correctly predicted instances. This suggests LRMs often implement the expected solution rather than an uninterpretable reasoning process. Finally, we present a method to decode a verified natural language reasoning trace from latent tokens without knowing a gold reasoning trace a priori, demonstrating that it is possible to find a verified trace for a majority of correct predictions but only a minority of incorrect predictions. Our findings highlight that current LRMs largely encode interpretable processes, and interpretability itself can be a signal of prediction correctness.