# Dissociation of Faithful and Unfaithful Reasoning in LLMs

arXiv: https://arxiv.org/abs/2405.15092 · HTML: https://mlsanity.com/paper/2405.15092/

## Key facts

- Cited by 23 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 1 of its 38 references that are in the map.
- First posted on arXiv 23 May 2024.

## 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, 1 of 1)

- [On Measuring Faithfulness or Self-consistency of Natural Language Explanations](https://mlsanity.com/paper/2311.07466/) (2023)

## Built on by (most cited first, 23 of 23)

- [Chain-of-Thought Reasoning In The Wild Is Not Always Faithful](https://mlsanity.com/paper/2503.08679/) (2025)
- [FaithCoT-Bench: Benchmarking Instance-Level Faithfulness of Chain-of-Thought Reasoning](https://mlsanity.com/paper/2510.04040/) (2025)
- [A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Models](https://mlsanity.com/paper/2509.03871/) (2025)
- [Measuring the Faithfulness of Thinking Drafts in Large Reasoning Models](https://mlsanity.com/paper/2505.13774/) (2025)
- [Learning to Self-Verify Makes Language Models Better Reasoners](https://mlsanity.com/paper/2602.07594/) (2026)
- [How does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencoding](https://mlsanity.com/paper/2507.22928/) (2025)
- [Improving Chain-of-Thought Reasoning via Quasi-Symbolic Abstractions](https://mlsanity.com/paper/2502.12616/) (2025)
- [CoMAT: Chain of Mathematically Annotated Thought Improves Mathematical Reasoning](https://mlsanity.com/paper/2410.10336/) (2024)
- [Can Aha Moments Be Fake? Towards Quantifying Decorative and True Thinking in Chain-of-Thought](https://mlsanity.com/paper/2510.24941/) (2025)
- [Learning to Ponder: Adaptive Reasoning in Latent Space](https://mlsanity.com/paper/2509.24238/) (2025)
- [State over Tokens: Characterizing the Role of Reasoning Tokens](https://mlsanity.com/paper/2512.12777/) (2025)
- [Detecting and Suppressing Reward Hacking with Gradient Fingerprints](https://mlsanity.com/paper/2604.16242/) (2026)
- [ReasoningFlow: Discourse Structures for Understanding LLM Reasoning Traces](https://mlsanity.com/paper/2606.05402/) (2026)
- [Unspoken Hints: Accuracy Without Acknowledgement in LLM Reasoning](https://mlsanity.com/paper/2509.26041/) (2025)
- [Measuring Faithfulness Depends on How You Measure: Classifier Sensitivity in LLM Chain-of-Thought Evaluation](https://mlsanity.com/paper/2603.20172/) (2026)
- [Correlation or Causation: Analyzing the Causal Structures of LLM and LRM Reasoning Process](https://mlsanity.com/paper/2509.17380/) (2025)
- [Detecting Unfaithful Chain-of-Thought via Circuit-Guided Internal-External Discrepancy](https://mlsanity.com/paper/2605.25603/) (2026)
- [Bypassing the Rationale: Causal Auditing of Implicit Reasoning in Language Models](https://mlsanity.com/paper/2602.03994/) (2026)
- [Deep Reasoning in General Purpose Agents via Structured Meta-Cognition](https://mlsanity.com/paper/2605.11388/) (2026)
- [Explanation-Bound Tool Execution for AI Agents: Server-Verified Action Claims Without Trusting Model Rationales](https://mlsanity.com/paper/2607.25364/) (2026)
- [Do Biased Models Have Biased Thoughts?](https://mlsanity.com/paper/2508.06671/) (2025)
- [Forecasting Future Behavior as a Learning Task](https://mlsanity.com/paper/2606.11445/) (2026)
- [Beyond Local Accuracy: A Protocol-Level Identifiability Audit for Controlled LLM Reasoning Evaluation](https://mlsanity.com/paper/2608.13326/) (2026)

## Abstract (arXiv)

Large language models (LLMs) often improve their performance in downstream tasks when they generate Chain of Thought reasoning text before producing an answer. We investigate how LLMs recover from errors in Chain of Thought. Through analysis of error recovery behaviors, we find evidence for unfaithfulness in Chain of Thought, which occurs when models arrive at the correct answer despite invalid reasoning text. We identify factors that shift LLM recovery behavior: LLMs recover more frequently from obvious errors and in contexts that provide more evidence for the correct answer. Critically, these factors have divergent effects on faithful and unfaithful recoveries. Our results indicate that there are distinct mechanisms driving faithful and unfaithful error recoveries. Selective targeting of these mechanisms may be able to drive down the rate of unfaithful reasoning and improve model interpretability.