The reasoning you can't see: filler tokens and illegible chains of thought

What a chain of thought does and does not reveal.

1. Jacob Pfau, William Merrill, and Samuel R. Bowman, "Let's Think Dot by Dot: Hidden Computation in Transformer Language Models," arXiv:2404.15758 (2024). https://arxiv.org/abs/2404.15758
2. Vatsal Baherwani, Tom Goldstein, and Ashwinee Panda, "Not All LLM Reasoning is Visible in the Chain-of-Thought," arXiv:2607.22925 (2026). https://arxiv.org/abs/2607.22925
3. Kaley Brauer, Claudio Mayrink Verdun, and Samuel Marks, "Reading Between the Dots: Decoding Hidden Computation across Filler Tokens," arXiv:2607.03502 (2026). https://arxiv.org/abs/2607.03502
4. Arun Jose, "Reasoning Models Sometimes Output Illegible Chains of Thought," arXiv:2510.27338 (2025). https://arxiv.org/abs/2510.27338