Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought Reasoning

• URL: http://arxiv.org/abs/2605.24286v1
• Date: Fri, 22 May 2026 23:37:29 GMT
• Title: Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought Reasoning
• Abstract summary: Chain-of-thought (CoT) reasoning is useful for monitoring language models.<n>Models can rely on prompt-to-answer shortcuts that bypass the CoT.<n>We study CoT faithfulness through a structural information-flow perspective.
• Score: 10.87972575497941
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Chain-of-thought (CoT) reasoning is useful for monitoring language models only when the reasoning trace faithfully reflects the computation that produces the final answer. However, models can rely on prompt-to-answer shortcuts that bypass the CoT, making the visible reasoning trace misleading even when it appears plausible. We study CoT faithfulness through a structural information-flow perspective: faithful reasoning should route answer-relevant information through the mediated path from prompt to CoT to answer, rather than through a direct prompt-to-answer shortcut. This perspective yields a task-agnostic framework based on three complementary properties, sufficiency, completeness, and necessity, which we instantiate with entropy-based, masked-KL, and gradient-based diagnostics. We show that these metrics recover externally judged faithfulness differences in hinted reasoning, and identify a low-entropy failure mode of KL-based diagnostics where gradient-based measures remain more stable. Building on this analysis, we introduce update-time interventions for verifier-based on-policy RL, including attention masking, backward-only gradient masking, CoT gradients, and adversarial perturbations of prompt representations. Across hinted arithmetic, reward-hackable code repair, and DAPO-Math models trained without hints but evaluated under wrong-hint injection, our interventions shift behavioral and structural indicators toward stronger CoT mediation. In particular, they make shortcut and reward-hacking behavior more transparent in the CoT and improve task-agnostic faithfulness metrics, while in some settings also reducing wrong-hint susceptibility. Our results suggest that controlling information flow during training is a practical route toward more faithful and monitorable CoT reasoning. Code is available at https://github.com/safety-research/faithful-cot.

Related papers

• What Makes Effective Supervision in Latent Chain-of-Thought: An Information-Theoretic Analysis [24.026935801991744]
  We analyze Latent Chain-of-Thought (CoT) from an information-theoretic perspective and identify this failure as a dual collapse.<n>To measure these effects, we introduce the Unified Latent Probe (ULP), which quantifies the mutual information between latent trajectories and explicit reasoning steps.
  arXiv  Detail & Related papers  (2026-06-18T10:51:53Z)
• Detecting Unfaithful Chain-of-Thought via Circuit-Guided Internal-External Discrepancy [43.942072229910515]
  Chain-of-thought (CoT) reasoning improves problem-solving ability of large language models.<n> generated reasoning traces may not faithfully reflect the model's actual decision process.<n>We propose Circuit-guided Internal-External Discrepancy Scorer (CIE-Scorer) for instance-level CoT unfaithfulness detection.
  arXiv  Detail & Related papers  (2026-05-25T08:54:55Z)
• FACT-E: Causality-Inspired Evaluation for Trustworthy Chain-of-Thought Reasoning [49.65751420291115]
  Chain-of-Thought (CoT) prompting has improved LLM reasoning, but models often generate explanations that appear coherent while containing unfaithful intermediate steps.<n>We propose FACT-E, a causality-inspired framework for evaluating CoT quality.<n>We show that FACT-E improves reasoning-trajectory selection and yields stronger in-context learnings.
  arXiv  Detail & Related papers  (2026-04-12T15:35:08Z)
• Balancing Faithfulness and Performance in Reasoning via Multi-Listener Soft Execution [79.98699884805636]
  Reasoning Execution by Multiple Listeners (REMUL) is a multi-party reinforcement learning approach.<n>REMUL builds on the hypothesis that reasoning traces which other parties can follow will be more faithful.<n>Speakers are rewarded for producing reasoning that is clear to listeners.
  arXiv  Detail & Related papers  (2026-02-18T02:55:55Z)
• Constraint-Rectified Training for Efficient Chain-of-Thought [60.52883907721588]
  Chain-of-Thought (CoT) has significantly enhanced the reasoning capabilities of Large Language Models (LLMs)<n>While longer reasoning traces can improve answer quality and unlock abilities such as self-correction, they also incur high inference costs and often introduce redundant steps, known as overthinking.<n>Recent research seeks to develop efficient reasoning strategies that balance reasoning length and accuracy.
  arXiv  Detail & Related papers  (2026-02-13T02:13:45Z)
• Are Reasoning LLMs Robust to Interventions on Their Chain-of-Thought? [79.86483056611105]
  Reasoning LLMs generate step-by-step chains of thought before giving an answer.<n>How robust are these reasoning traces to disruptions that occur within them?<n>We introduce a controlled evaluation framework that perturbs a model's own CoT at fixed timesteps.
  arXiv  Detail & Related papers  (2026-02-07T10:02:58Z)
• Stop Rewarding Hallucinated Steps: Faithfulness-Aware Step-Level Reinforcement Learning for Small Reasoning Models [59.6715047267181]
  Small reasoning models (SRMs) are prone to hallucinations, especially in intermediate reasoning steps.<n>Existing mitigation methods based on online reinforcement learning rely on outcome-based rewards or coarse-grained chain-of-thought evaluation.<n>We propose Faithfulness-Aware Step-Level Reinforcement Learning (FaithRL), introducing step-level supervision via explicit faithfulness rewards from a process reward model.
  arXiv  Detail & Related papers  (2026-02-05T17:15:12Z)
• Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity [3.117948413097524]
  Chain-of-thought (CoT) outputs let us read a model's step-by-step reasoning.<n>We evaluate instruction-tuned and reasoning models on BBH, GPQA, and MMLU.
  arXiv  Detail & Related papers  (2025-10-31T11:14:39Z)
• How Chain-of-Thought Works? Tracing Information Flow from Decoding, Projection, and Activation [9.455881608413137]
  Chain-of-Thought (CoT) prompting significantly enhances model reasoning, yet its internal mechanisms remain poorly understood.<n>We analyze CoT's operational principles by reversely tracing information flow across decoding, projection, and activation phases.
  arXiv  Detail & Related papers  (2025-07-28T12:11:16Z)
• Interpretable Traces, Unexpected Outcomes: Investigating the Disconnect in Trace-Based Knowledge Distillation [14.489157453882767]
  This work aims to address the challenge of evaluating reasoning traces and their correlation with the final performance.<n>We employ a KD method leveraging rule-based problem decomposition to generate interpretable traces.<n>Specifically, we demonstrate this approach on Open Book QA, decomposing the problem into a Classification step and an Information Retrieval step.
  arXiv  Detail & Related papers  (2025-05-20T00:49:19Z)
• Ladder-of-Thought: Using Knowledge as Steps to Elevate Stance Detection [73.31406286956535]
  We introduce the Ladder-of-Thought (LoT) for the stance detection task. LoT directs the small LMs to assimilate high-quality external knowledge, refining the intermediate rationales produced. Our empirical evaluations underscore LoT's efficacy, marking a 16% improvement over GPT-3.5 and a 10% enhancement compared to GPT-3.5 with CoT on stance detection task.
  arXiv  Detail & Related papers  (2023-08-31T14:31:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.