Reasoning Traces Shape Outputs but Models Won't Say So

Computer Science > Artificial Intelligence [Submitted on 21 Mar 2026] Reasoning Traces Shape Outputs but Models Won't Say So Yijie Hao, Lingjie Chen, Ali Emami, Joyce Ho Can we trust the reasoning traces that large reasoning models (LRMs) produce? We investigate whether these traces faithfully reflect what drives model outputs, and whether models will honestly report their influence. We introduce Thought Injection, a method that injects synthetic reasoning snippets into a model's <think> trace, then measures whether the model follows the injected reasoning and acknowledges doing so. Across 45,000 samples from three LRMs, we find that injected hints reliably alter outputs, confirming that reasoning traces causally shape model behavior. However, when asked to explain their changed answers, models overwhelmingly refuse to disclose the influence: overall non-disclosure exceeds 90% for extreme hints across 30,000 follow-up samples. Instead of acknowledging the injected reasoning, models fabricate aligned-appearing but unrelated explanations. Activation analysis reveals that sycophancy- and deception-related directions are strongly activated during these fabrications, suggesting systematic patterns rather than incidental failures. Our findings reveal a gap between the reasoning LRMs follow and the reasoning they report, raising concern that aligned-appearing explanations may not be equivalent to genuine alignment. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2603.20620 [cs.AI]   (or arXiv:2603.20620v1 [cs.AI] for this version)   https://doi.org/10.48550/arXiv.2603.20620 Focus to learn more Submission history From: Yijie Hao [view email] [v1] Sat, 21 Mar 2026 03:31:36 UTC (2,217 KB) Access Paper: HTML (experimental) view license Current browse context: cs.AI < prev   |   next > new | recent | 2026-03 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)