# Two Regimes of Chain-of-Thought Unfaithfulness: Metric-Based Detection Fails Where Models Are Wrong

arXiv: https://arxiv.org/abs/2607.23458 · HTML: https://mlsanity.com/paper/2607.23458/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning.
- Builds on 33 of its 39 references that are in the map.
- First posted on arXiv 26 Jul 2026.

## Placed in

- LLM reasoning / [Prompting and chain-of-thought reasoning](https://mlsanity.com/lr/prompting-and-chain-of-thought-reasoning/)

## Builds on (most cited first, 30 of 33)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://mlsanity.com/paper/2305.04388/) (2023)
- [Towards Understanding Sycophancy in Language Models](https://mlsanity.com/paper/2310.13548/) (2023)
- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [Measuring Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2307.13702/) (2023)
- [Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problems](https://mlsanity.com/paper/1705.04146/) (2017)
- [Designing and Interpreting Probes with Control Tasks](https://mlsanity.com/paper/1909.03368/) (2019)
- [Reasoning Models Don't Always Say What They Think](https://mlsanity.com/paper/2505.05410/) (2025)
- [ProcessBench: Identifying Process Errors in Mathematical Reasoning](https://mlsanity.com/paper/2412.06559/) (2024)
- [Chain-of-Thought Reasoning In The Wild Is Not Always Faithful](https://mlsanity.com/paper/2503.08679/) (2025)
- [Analyzing the Generalization and Reliability of Steering Vectors](https://mlsanity.com/paper/2407.12404/) (2024)
- [Llama Scope: Extracting Millions of Features from Llama-3.1-8B with Sparse Autoencoders](https://mlsanity.com/paper/2410.20526/) (2024)
- [A Chain-of-Thought Is as Strong as Its Weakest Link: A Benchmark for Verifiers of Reasoning Chains](https://mlsanity.com/paper/2402.00559/) (2024)
- [Chain-of-Thought Unfaithfulness as Disguised Accuracy](https://mlsanity.com/paper/2402.14897/) (2024)
- [FaithCoT-Bench: Benchmarking Instance-Level Faithfulness of Chain-of-Thought Reasoning](https://mlsanity.com/paper/2510.04040/) (2025)
- [Premise-Augmented Reasoning Chains Improve Error Identification in Math reasoning with LLMs](https://mlsanity.com/paper/2502.02362/) (2025)
- [Is Chain-of-Thought Really Not Explainability? Chain-of-Thought Can Be Faithful without Hint Verbalization](https://mlsanity.com/paper/2512.23032/) (2025)
- [A Causal Lens for Evaluating Faithfulness Metrics](https://mlsanity.com/paper/2502.18848/) (2025)
- [VeriCoT: Neuro-symbolic Chain-of-Thought Validation via Logical Consistency Checks](https://mlsanity.com/paper/2511.04662/) (2025)
- [The Impact of Off-Policy Training Data on Probe Generalisation](https://mlsanity.com/paper/2511.17408/) (2025)
- [Graph of Verification: Structured Verification of LLM Reasoning with Directed Acyclic Graphs](https://mlsanity.com/paper/2506.12509/) (2025)
- [Post-Hoc Reasoning in Chain of Thought: Decoding and Steering Pre-Committed Answers](https://mlsanity.com/paper/2603.01437/) (2026)
- [Catching rationalization in the act: detecting motivated reasoning before and after CoT via activation probing](https://mlsanity.com/paper/2603.17199/) (2026)

## Built on by (most cited first, 1 of 1)

- [Hallucination Across the Reasoning Lifecycle: Interface Visibility, Causal Evidence, and Release Control in Large Reasoning Models](https://mlsanity.com/paper/2610.05472/) (2026)

## Abstract (arXiv)

Chain-of-thought (CoT) explanations support oversight only if they are faithful: the stated reasoning must actually produce the answer. Auditing black-box (behavioral) detection of unfaithful CoT against FaithCoT-Bench's human annotations, we find answer correctness structures the problem at every level. Answer incorrectness alone (an oracle diagnostic, not a deployable detector) outperforms every purpose-built signal (AUROC 0.696), because 69% of annotated unfaithfulness occurs on incorrect answers. Stratifying by correctness splits detection into two regimes: on correct answers, behavioral signals moderately separate faithful from post-hoc reasoning (0.63-0.67); on incorrect answers, where most unfaithfulness lives, no tested signal is detectably above chance (replicated on all four models for benchmark-wide signals). The standard step-removal metric anti-correlates with human labels; this inversion reproduces on the benchmark's released scores and on hint-dependent counterfactually labeled traces. Linear probes decode the behaviorally blind regime in Llama-3.1-8B and the correct-answer regime in Qwen-2.5-7B, with no shared, positively aligned direction detected across regimes; instructed answer-first traces (7 models) transfer to neither annotated regime, while hint-induced unverbalized answer flips do, in model- and source-dependent settings. We also independently verify and resolve a documentation-data mismatch in the benchmark's label semantics.