Recognition-Refusal Misalignment in LLMs: Why Models Answer Structurally Unanswerable Questions

• URL: http://arxiv.org/abs/2608.29109v1
• Date: Sat, 29 Aug 2026 07:38:17 GMT
• Title: Recognition-Refusal Misalignment in LLMs: Why Models Answer Structurally Unanswerable Questions
• Abstract summary: Large language models often answer structurally unanswerable questions, such as computing cot(-540) or evaluating (1).startswith("1").<n>We ask whether this failure reflects missing recognition or failed routing from recognition to abstention.<n>Across instruction-tuned models from 1.7B to 70B parameters, a single linear direction in the hidden state separates answerable from structurally impossible math and code prompts.
• Score: 3.596084005285064
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models often answer structurally unanswerable questions, such as computing cot(-540°) or evaluating (1).startswith("1"), instead of abstaining. We ask whether this failure reflects missing recognition or failed routing from recognition to abstention. Across instruction-tuned models from 1.7B to 70B parameters, a single linear direction in the hidden state separates answerable from structurally impossible math and code prompts, showing that models represent impossibility before generation. Yet this recognition direction is nearly orthogonal to the canonical safety-refusal direction that mediates trained harmful-content refusal. An in-domain behavior-defined invalidity-aware direction is closer to recognition, but only partially aligned with it, and remains near-orthogonal to safety refusal. Generation-time steering along the recognition direction changes invalidity-aware behavior bidirectionally and dose-responsively on structural math and code cells, while random directions do not. Base/instruct comparisons further show that the low-cosine geometry is already present at the pretraining endpoint. The confident-on-impossible failure is therefore better explained as a routing failure than as an encoding failure: the model has a usable "no admissible answer" signal, but the safety-refusal pathway is not aligned to use it.

Related papers

• Curved Inference II: Sleeper Agent Geometry - Extending Interpretability Beyond Probes [0.0]
  This paper extends Anthropic's Sleeper Agents research, which showed artificial backdoors persist through safety training & can be detected by linear probes with >99% accuracy.<n>We introduce a naturalistic methodology using multi-turn context windows that simulates realistic deceptive reasoning without artificial triggers or supervised backdoor insertion.<n> Geometric structure reliably predicts semantic classification, with statistically significant differences in surface area across five prompt strategies & two model families.
  arXiv  Detail & Related papers  (2026-08-25T03:51:20Z)
• Broken Symmetry in LLM Refusal: Answer Release Is More Local Than Refusal Restoration [24.010665397925333]
  When a language model refuses to answer a prompt, it is unclear whether the correct answer is erased from its internal representations, or merely suppressed at the output layer.<n>We investigate this mechanism using a controlled withhold setting, which yields perfectly matched answering and refusal trajectories for bidirectional activation patching.
  arXiv  Detail & Related papers  (2026-08-16T14:47:33Z)
• 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)
• Actionable Activation Directions for Detecting and Mitigating Emergent Misalignment Across Language Model Families [0.0]
  Fine-tuning language models on insecure code induces emergent misalignment with poorly understood internal structure.<n>We investigate whether this misalignment corresponds to a causally actionable activation-space direction shared across architectures.
  arXiv  Detail & Related papers  (2026-06-18T13:39:59Z)
• Greedy Coordinate Diffusion: Effective and Semantically Coherent Adversarial Attacks via Diffusion Guidance [48.34904668359272]
  Adrial attacks on large language models have limited practical impact despite extensive research.<n>We introduce Greedy Coordinate Diffusion (GCD), a novel framework that efficiently generates adversarial attacks against safety-aligned models.<n>GCD maintains low perplexity and high semantic adherence to the adversary's original intent.
  arXiv  Detail & Related papers  (2026-06-14T01:18:53Z)
• Geometric Deviation as an Unsupervised Pre-Generation Reliability Signal: Probing LLM Representations for Answerability [0.0]
  A reliable language model should be able to signal, prior to generation, when a query falls outside its knowledge.<n>We investigate whether representation geometry can provide such a pre-generation signal by measuring the deviation of hidden states from an answerable reference set.
  arXiv  Detail & Related papers  (2026-05-04T22:24:34Z)
• Pando: Do Interpretability Methods Work When Models Won't Explain Themselves? [53.07826484214082]
  We introduce Pando, a model-organism benchmark that breaks this confound via an explanation axis.<n>Pando predicts held-out model decisions from $10$ labeled query-response pairs.<n>When explanations are faithful, black-box elicitation matches or exceeds all white-box methods.
  arXiv  Detail & Related papers  (2026-04-13T06:42:24Z)
• How Alignment Routes: Localizing, Scaling, and Controlling Policy Circuits in Language Models [0.0]
  This paper localizes the policy routing mechanism in alignment-trained language models.<n>An intermediate-layer attention gate reads detected content and triggers deeper heads that boost the signal toward refusal.
  arXiv  Detail & Related papers  (2026-04-06T03:20:37Z)
• Agentic Uncertainty Quantification [76.94013626702183]
  We propose a unified Dual-Process Agentic UQ (AUQ) framework that transforms verbalized uncertainty into active, bi-directional control signals.<n>Our architecture comprises two complementary mechanisms: System 1 (Uncertainty-Aware Memory, UAM), which implicitly propagates verbalized confidence and semantic explanations to prevent blind decision-making; and System 2 (Uncertainty-Aware Reflection, UAR), which utilizes these explanations as rational cues to trigger targeted inference-time resolution only when necessary.
  arXiv  Detail & Related papers  (2026-01-22T07:16:26Z)
• How Does Prefix Matter in Reasoning Model Tuning? [57.69882799751655]
  We fine-tune three R1 series models across three core model capabilities: reasoning (mathematics), coding, safety, and factuality.<n>Results show that prefix-conditioned SFT improves both safety and reasoning performance, yielding up to +6% higher Safe@1 accuracy.
  arXiv  Detail & Related papers  (2026-01-04T18:04:23Z)
• Refusal Falls off a Cliff: How Safety Alignment Fails in Reasoning? [68.82210578851442]
  We investigate why safety alignment fails in reasoning models through a mechanistic interpretability lens.<n>Using a linear probing approach to trace refusal intentions across token positions, we discover a phenomenon termed as textbfrefusal cliff<n>We propose textbfCliff-as-a-Judge, a novel data selection method that identifies training examples exhibiting the largest refusal cliff to efficiently repair reasoning models' safety alignment.
  arXiv  Detail & Related papers  (2025-10-07T15:32:59Z)
• R-Tuning: Instructing Large Language Models to Say `I Don't Know' [66.11375475253007]
  Large language models (LLMs) have revolutionized numerous domains with their impressive performance but still face their challenges. Previous instruction tuning methods force the model to complete a sentence no matter whether the model knows the knowledge or not. We present a new approach called Refusal-Aware Instruction Tuning (R-Tuning) Experimental results demonstrate R-Tuning effectively improves a model's ability to answer known questions and refrain from answering unknown questions.
  arXiv  Detail & Related papers  (2023-11-16T08:45:44Z)

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.