How Does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencoding

Xi Chen; Aske Plaat; Niki van Stein

AAAI 2026technical0 citations

How Does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencoding

Abstract

Chain‑of‑thought (CoT) prompting boosts Large Language Models accuracy on multi‑step tasks, yet whether the generated ``thoughts

BibTeX

@inproceedings{aaai2026_howdoeschainofth,
  title = {How Does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencoding},
  author = {Xi Chen and Aske Plaat and Niki van Stein},
  booktitle = {AAAI 2026},
  year = {2026}
}