The Hypocrisy Gap: Quantifying Divergence Between Internal Belief and Chain-of-Thought Explanation via Sparse Autoencoders

• URL: http://arxiv.org/abs/2602.02496v1
• Date: Wed, 14 Jan 2026 00:40:40 GMT
• Title: The Hypocrisy Gap: Quantifying Divergence Between Internal Belief and Chain-of-Thought Explanation via Sparse Autoencoders
• Abstract summary: We introduce the Hypocrisy Gap to quantify the divergence between a model's internal reasoning and its final generation.<n>By mathematically comparing an internal truth belief, derived via sparse linear probes, to the final generated trajectory in latent space, we quantify and detect a model's tendency to engage in unfaithful behavior.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models (LLMs) frequently exhibit unfaithful behavior, producing a final answer that differs significantly from their internal chain of thought (CoT) reasoning in order to appease the user they are conversing with. In order to better detect this behavior, we introduce the Hypocrisy Gap, a mechanistic metric utilizing Sparse Autoencoders (SAEs) to quantify the divergence between a model's internal reasoning and its final generation. By mathematically comparing an internal truth belief, derived via sparse linear probes, to the final generated trajectory in latent space, we quantify and detect a model's tendency to engage in unfaithful behavior. Experiments on Gemma, Llama, and Qwen models using Anthropic's Sycophancy benchmark show that our method achieves an AUROC of 0.55-0.73 for detecting sycophantic runs and 0.55-0.74 for hypocritical cases where the model internally "knows" the user is wrong, consistently outperforming a decision-aligned log-probability baseline (0.41-0.50 AUROC).

Related papers

• THESIS-MoE: Trainable Hierarchical Extraction and SteerIng of Sycophancy in Mixture-of-Experts [0.300988853836121]
  Sycophancy, the tendency of a language model to change its answer to match a user's stated belief, is a common alignment failure.<n>We introduce a shared contrastive signal, built from matched prompts with and without a stated belief, that identifies where sycophancy lives.<n>Our results demonstrate that sycophancy resides in identifiable computational subcircuits and can be selectively steered.
  arXiv  Detail & Related papers  (2026-08-16T11:30:26Z)
• Probing the Misaligned Thinking Process of Language Models [32.29324298373053]
  Large language models exhibit a growing range of misaligned behaviors.<n>It is critical to reliably detect such behaviors to ensure safe and responsible use.<n>We propose to monitor misalignment by decomposing it into fine-grained cognitive processes.
  arXiv  Detail & Related papers  (2026-06-23T07:40:28Z)
• 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)
• Reliable Control-Point Selection for Steering Reasoning in Large Language Models [28.288321095634128]
  Steering vectors offer a training-free mechanism for controlling reasoning behaviors in large language models.<n>However, constructing effective vectors requires identifying genuine behavioral signals in the model's hidden states.<n>We show that current methods detect these behaviors through keyword matching in chain-of-thought traces, implicitly assuming that every detected boundary encodes a genuine behavioral signal.<n>We develop a probabilistic model that formalizes intrinsic reasoning behaviors as events with context-dependent trigger probabilities, and show that unstable boundaries dilute the steering signal.
  arXiv  Detail & Related papers  (2026-04-02T14:48:56Z)
• Closing the Confidence-Faithfulness Gap in Large Language Models [3.122242683664973]
  Large language models verbalize confidence scores that are largely detached from their actual accuracy.<n>We present a mechanistic interpretability analysis of verbalized confidence.<n>We introduce a two-stage adaptive steering pipeline that reads the model's internal accuracy estimate and steers verbalized output to match it.
  arXiv  Detail & Related papers  (2026-03-26T05:42:04Z)
• Decoding the Critique Mechanism in Large Reasoning Models [50.821607345799386]
  Large Reasoning Models (LRMs) exhibit backtracking and self-verification mechanisms that enable them to revise intermediate steps and reach correct solutions.<n>We investigate how current LRMs recover from errors by inserting arithmetic mistakes in their intermediate reasoning steps.<n>We find that despite the error propagating through the chain-of-thought, the model still reaches the correct final answer.
  arXiv  Detail & Related papers  (2026-03-17T10:03:30Z)
• Mitigating LLM Hallucination via Behaviorally Calibrated Reinforcement Learning [32.32593439144886]
  Behavior-calibrated reinforcement learning allows smaller models to surpass frontier models in uncertainty quantification.<n>Our model's log-scale Accuracy-to-Hallucination Ratio gain (0.806) exceeds GPT-5's (0.207) in a challenging in-domain evaluation.
  arXiv  Detail & Related papers  (2025-12-22T22:51:48Z)
• Which Sparse Autoencoder Features Are Real? Model-X Knockoffs for False Discovery Rate Control [0.0]
  We introduce Model-X knockoffs to SAE feature selection, using knock-off+ to control the false discovery rate (FDR)<n>Our method offers a re-producible and principled framework for reliable feature discovery by combining SAEs with multiple-testing-aware inference.
  arXiv  Detail & Related papers  (2025-11-12T17:12:45Z)
• User-Based Sequential Modeling with Transformer Encoders for Insider Threat Detection [0.005755004576310333]
  Insider threat detection presents unique challenges due to the authorized status of malicious actors.<n>Existing machine learning methods treat user activity as isolated events, thereby failing to leverage sequential dependencies in user behavior.<n>We propose a User-Based Sequencing (UBS) methodology, transforming the CERT insider threat dataset into structured temporal sequences suitable for deep sequential modeling.
  arXiv  Detail & Related papers  (2025-06-30T00:47:31Z)
• SEAL: Steerable Reasoning Calibration of Large Language Models for Free [58.931194824519935]
  Large Language Models (LLMs) have demonstrated compelling capabilities for complex reasoning tasks via the extended chain-of-thought (CoT) reasoning mechanism.<n>Recent studies reveal substantial redundancy in the CoT reasoning traces, which negatively impacts model performance.<n>We introduce SEAL, a training-free approach that seamlessly calibrates the CoT process, improving accuracy while demonstrating significant efficiency gains.
  arXiv  Detail & Related papers  (2025-04-07T02:42:07Z)
• Verify when Uncertain: Beyond Self-Consistency in Black Box Hallucination Detection [25.176984317213858]
  Large Language Models (LLMs) suffer from hallucination problems, which hinder their reliability in sensitive applications.<n>We propose a budget-friendly, two-stage detection algorithm that calls the verifier model only for a subset of cases.
  arXiv  Detail & Related papers  (2025-02-20T21:06:08Z)
• Disentangled Latent Spaces for Reduced Order Models using Deterministic Autoencoders [0.0]
  More interpretability can be gained by disentangling the latent variables and analyzing the resulting modes.<n>Probabilistic autoencoders ($beta$-VAEs) are frequently used in computational fluid dynamics.<n>We show that competitive results can be achieved using non-probabilistic autoencoder approaches.
  arXiv  Detail & Related papers  (2025-02-20T16:09:57Z)
• Self-Evaluation Guided Beam Search for Reasoning [61.523627290397556]
  We introduce a stepwise self-evaluation mechanism to guide and calibrate the reasoning process of Large Language Model (LLM) We propose a decoding algorithm integrating the self-evaluation guidance via beam search. Our approach surpasses the corresponding Codex-backboned baselines in few-shot accuracy by $6.34%$, $9.56%$, and $5.46%$ on the GSM8K, AQuA, and StrategyQA.
  arXiv  Detail & Related papers  (2023-05-01T02:37:59Z)
• Robustness and Accuracy Could Be Reconcilable by (Proper) Definition [109.62614226793833]
  The trade-off between robustness and accuracy has been widely studied in the adversarial literature. We find that it may stem from the improperly defined robust error, which imposes an inductive bias of local invariance. By definition, SCORE facilitates the reconciliation between robustness and accuracy, while still handling the worst-case uncertainty.
  arXiv  Detail & Related papers  (2022-02-21T10:36:09Z)
• On the Practicality of Differential Privacy in Federated Learning by Tuning Iteration Times [51.61278695776151]
  Federated Learning (FL) is well known for its privacy protection when training machine learning models among distributed clients collaboratively. Recent studies have pointed out that the naive FL is susceptible to gradient leakage attacks. Differential Privacy (DP) emerges as a promising countermeasure to defend against gradient leakage attacks.
  arXiv  Detail & Related papers  (2021-01-11T19:43:12Z)
• Autoencoding Variational Autoencoder [56.05008520271406]
  We study the implications of this behaviour on the learned representations and also the consequences of fixing it by introducing a notion of self consistency. We show that encoders trained with our self-consistency approach lead to representations that are robust (insensitive) to perturbations in the input introduced by adversarial attacks.
  arXiv  Detail & Related papers  (2020-12-07T14:16:14Z)
• Structural Causal Models Are (Solvable by) Credal Networks [70.45873402967297]
  Causal inferences can be obtained by standard algorithms for the updating of credal nets. This contribution should be regarded as a systematic approach to represent structural causal models by credal networks. Experiments show that approximate algorithms for credal networks can immediately be used to do causal inference in real-size problems.
  arXiv  Detail & Related papers  (2020-08-02T11:19:36Z)

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.