Two Regimes of Chain-of-Thought Unfaithfulness: Behavioral Detection Fails Where Models Are Wrong

• URL: http://arxiv.org/abs/2607.23458v1
• Date: Sun, 26 Jul 2026 04:43:53 GMT
• Title: Two Regimes of Chain-of-Thought Unfaithfulness: Behavioral Detection Fails Where Models Are Wrong
• Abstract summary: Black-box (behavioral) detection of unfaithful CoT against FaithCoT-Bench's human annotations.<n>We find answer correctness structures the problem at every level.
• Score: 1.2744523252873352
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• Tracing Provenance and Detecting Tampering with Complementary LLM Watermarks [68.64050157343334]
  Existing watermarks preserve provenance under editing, but this same robustness allows an adversary to alter critical content while retaining attribution.<n>We introduce an innovative watermark that jointly provides provenance and tamper evidence.<n>The signals share the same mechanism but use independent keys and different seeding windows over normalized text, making one resilient to edits and the other sensitive to reader-visible changes.
  arXiv  Detail & Related papers  (2026-08-13T01:55:16Z)
• Toward Auditable Fraud Detection: Combining Graph Features, Model Explanations, and Agentic Case Investigation [0.0]
  Fraud detection systems must scale with rising transaction volume while remaining explainable and reviewable.<n>We study a layered pipeline that combines a gradient-boosted classifier, graph-derived structural features, an autoencoder-based anomaly signal, and an investigation agent.<n>We conclude that each component contributes only under specific conditions, and that a plausible rationale from an investigation agent is not evidence of a better decision.
  arXiv  Detail & Related papers  (2026-07-21T16:37:41Z)
• The Knowing-Saying Gap: When Probes See Errors that Confidence Misses [20.46211358172975]
  Linear probes detect corrupted context in language models with near-perfect accuracy, yet this does not translate into reliable failure prediction.<n>Across multi-hop arithmetic chains, probes that detect corruption turn out to be uninformative about final answer correctness.<n>As a real-time monitor, probe-based interventions are sharply model and error-type dependent.
  arXiv  Detail & Related papers  (2026-07-21T12:13:10Z)
• Semantic Drift in Bug Resolution: How Behavioral Signals Propagate from Reports to Tests and Patches [18.727291516223115]
  Desc2Fix is a framework for measuring semantic alignment between bug reports, triggering tests, and developer-written fixes.<n>We analyze 2,857 report-test-patch triplets from Defects4J and SWT-Bench using GPT-4o and DeepSeek-Chat.
  arXiv  Detail & Related papers  (2026-07-20T22:32:23Z)
• Fantastic Adaptive Taxonomies and How to Use Them [100.20614173157708]
  AdaMAST converts execution traces into a compact, evidence-grounded failure taxonomy.<n>No code is hand-authored, no trace human-annotated.<n>In agent-system search, taxonomy-coded diagnoses of failed candidates outperform free-form.
  arXiv  Detail & Related papers  (2026-07-17T17:41:16Z)
• What Accuracy and Gradient Cosine Miss: Evaluating Feedback Alignment via Scale Stability, Reference Validity, and Depth Utility [48.10132234701036]
  We propose a diagnostic evaluation protocol based on three checks -- scale stability, reference validity, and depth utility -- together with per-layer rather than aggregate cosine reporting.<n>Across multiple architectures and methods, our protocol identifies all failures with wide calibration margins.
  arXiv  Detail & Related papers  (2026-06-19T06:04:17Z)
• Distributional Energy-Based Models for Uncertainty-Aware Structured LLM Reasoning [40.342912574072024]
  Large Language Models produce structured outputs such as travel plans or code solutions.<n>Individual reasoning steps may appear correct while the output as a whole violates budgets, fails test cases, or contradicts earlier deductions.<n>We propose a learned quality scorer with deterministic analytical constraint penalties for verifying structured LLM outputs.
  arXiv  Detail & Related papers  (2026-05-15T17:08:27Z)
• Hidden Error Awareness in Chain-of-Thought Reasoning: The Signal Is Diagnostic, Not Causal [6.908637308550535]
  Chain-of-thought assumes that generated reasoning reflects a model's internal computation.<n>We show this assumption is wrong in a specific, measurable way.<n>Models internally detect their own reasoning errors but outwardly express confidence in them.
  arXiv  Detail & Related papers  (2026-05-10T12:26:46Z)
• 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)
• SELFDOUBT: Uncertainty Quantification for Reasoning LLMs via the Hedge-to-Verify Ratio [0.038379177968040606]
  We propose SELFDOUBT, a single-pass uncertainty framework for reasoning language models.<n>Our key signal, the Hedge-to-Verify Ratio (HVR), detects whether a reasoning trace contains uncertainty markers and, if so, whether they are offset by explicit selfchecking behavior.<n> SELFDOUBT operates on a single observed reasoning trajectory, making it suitable for latency- and cost-constrained deployment over any proprietary API.
  arXiv  Detail & Related papers  (2026-04-07T19:19:29Z)
• CoVerRL: Breaking the Consensus Trap in Label-Free Reasoning via Generator-Verifier Co-Evolution [52.691495954442985]
  CoVerRL is a framework where a single model alternates between generator and verifier roles, with each capability bootstrapping the other.<n> Experiments across Qwen and Llama model families demonstrate that CoVerRL outperforms label-free baselines by 4.7-5.9% on mathematical reasoning benchmarks.<n>Self-verification accuracy improves from around 55% to over 85%, confirming that both capabilities genuinely co-evolve.
  arXiv  Detail & Related papers  (2026-03-18T14:38:55Z)
• CARE What Fails: Contrastive Anchored-REflection for Verifiable Multimodal [84.71254539482369]
  Group-relative reinforcement learning with verifiable rewards (RLVR) often wastes the most informative data it already has the failures.<n>We present CARE, a failure-centric post-training framework for multimodal reasoning that turns errors into supervision.<n> CARE improves accuracy and training smoothness while explicitly increasing the share of learning signal that comes from failures.
  arXiv  Detail & Related papers  (2025-12-22T16:34:21Z)
• Reference-Free Rating of LLM Responses via Latent Information [53.463883683503106]
  We study the common practice of asking a judge model to assign Likert-scale scores to free-text responses.<n>We then propose and evaluate Latent Judges, which derive scalar ratings from internal model signals.<n>Across a broad suite of pairwise and single-rating benchmarks, latent methods match or surpass standard prompting.
  arXiv  Detail & Related papers  (2025-09-29T12:15:52Z)

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.