Language Model Introspection

Overview

Language model introspection asks what models can know about their own computations, learned behavior, uncertainty, and likely outputs, and whether that knowledge can support auditing or safer decisions. Researchers use activation probes and logit lenses, causal interventions and attention-head analysis, behavioral benchmarks, and self-report training methods such as adapters and reinforcement learning. Tests often find that internal states contain useful signals, but models may not be able to report them reliably; surface cues and task shortcuts can also make apparent introspection hard to distinguish from ordinary prediction. A central open question is when a model has genuine access to information about its own computation, rather than merely producing a plausible report, and how such access varies across tasks and settings.

Latest papers

CardsList

41 papers4 papers in the last four weeks, down 20% on the four weeks before. 0.0% of all new papers.

  1. Training Language Models for Introspective Oversight

    Nov 17, 2026Alex Lin Wang, Linus Yifeng Tang, Dami Choi +1language model monitoringLanguage Model Introspection

  2. A Closer Look at Language Model Introspection: Probing Information and Reporting

    Nov 11, 2026Dustin Rubin, Vincenzo Timmel, Xufeng He +3Language Model IntrospectionLanguage Model Probing

  3. MicroLens: Read, Trace, and Rewire the Residual Stream with Two Lenses

    Nov 9, 2026Nika Zahedi, Fangyi Zhu, Anna WuLLM InterpretabilityLanguage Model Introspection

  4. Interventional Training Reduces Causal Bypass in In-Context Introspection

    Oct 28, 2026Zilu Tang, Jongin Kim, Kerem Sahin +3Language Model IntrospectionCausal Interventions in Language Models

  5. The Quine Test: How Much of a Model's Self Survives Its Own Description?

    Oct 28, 2026Bibin Babu, Ivan P. Yamshchikov, Aleksei ShpilmanLanguage Model IntrospectionLanguage Model Self-Assessment

  6. A Sense of Proprioception in Language Models

    Oct 12, 2026Jonas GeipingLanguage Model IntrospectionMetacognition in Language Models

  7. Identifying Introspection From the Inside

    Oct 5, 2026David I. Atkinson, Dillon Plunkett, David BauLLM InterpretabilityLanguage Model Introspection

  8. A mechanistic study of language model introspection

    Sep 28, 2026Jiahong Zou, Xiangkun Sun, Lingkai Kong +1Attention Head AnalysisLLM Interpretability

  9. Introspective Uncertainty Estimation for LLM-Based Code Generation

    Sep 12, 2026Thomas KlassertLLM Uncertainty EstimationLanguage Model Introspection

  10. Evaluating and Improving LLM Self-Modeling

    Aug 31, 2026Siqi Zeng, Andre N. Assis, Rowan WangLLM EvaluationLanguage Model Introspection

  11. "Many Are My Names": The Anatomy of the Assistant and Its Personas via Sparse Autoencoders

    Aug 8, 2026Adelaide Danilov, Aria Nourbakhsh, Oleksandr Marchenko Breneur +1Large Language Model-Based Role-Play SimulationPersonality Modeling in Language Models

  12. Looking in the Mirror: Introspecting Side-Effect Misalignments Induced by Fine-Tuning

    Aug 5, 2026Kotaro Yoshida, Laura Gomezjurado Gonzalez, Yukinori Yamamoto +3LLM AuditingLanguage Model Introspection

  13. Self-Reference in Large Language Models: The Introspection Threshold for Recursive Self-Improvement

    Jul 5, 2026Jiang Zhang, Bing Yuan, Qian ZhangLanguage Model IntrospectionLanguage Model Self-Improvement

  14. Revealing Hidden Model Behaviors with Task-Specific Self-Reports

    Jul 3, 2026Taras Kutsyk, Bartosz ZielińskiLLM AuditingLanguage Model Introspection

  15. ICA Lens: Non-Gaussianity Reveals Interpretable Directions in Language Models

    Jun 10, 2026Sida Liu, Feijiang Han, Sarah Wiegreffe +1Independent Component AnalysisLLM Interpretability

  16. Can LLMs Introspect? A Reality Check

    May 25, 2026Shashwat Singh, Tal Linzen, Shauli RavfogelLLM EvaluationLLM Interpretability

  17. Tool Calling is Linearly Readable and Steerable in Language Models

    May 8, 2026Zekun Wu, Ze Wang, Seonglae Cho +4AI Agent ReliabilityTool-Augmented Language Model Agents

  18. TokenScope: Token-Level Explainability and Interpretability for Code-Oriented Tasks in Large Language Models

    Apr 30, 2026Amirreza Esmaeili, Fatemeh FardLLM InterpretabilityCode Generation