A mechanistic study of language model introspection

arXiv:2609.35108 · cs.CL, cs.AI · Submitted 2026-09-28 · Read on arXiv

cs.CL, cs.AI

Submitted: 2026-09-28

Updated: 2026-09-28

Code: https://github.com/Zethan06/introspection

Terminology

• Accuracy — 58× in this paper · explained in 1 episode
• Calibration — 16× in this paper · explained in 7 episodes
• Alignment — 15× in this paper · explained in 1 episode
• Introspection — 11× in this paper · explained in 2 episodes
• Steering — 10× in this paper · explained in 2 episodes
• Ablation — 8× in this paper · explained in 1 episode
• Activation Patching — 8× in this paper · explained in 2 episodes
• Transformer — 8× in this paper · explained in 2 episodes
• Interpretability — 6× in this paper · explained in 15 episodes
• In International Conference on Learning Representations — 5× in this paper
• Clustering — 3× in this paper · explained in 2 episodes
• Activation Steering — 2× in this paper · explained in 4 episodes
• Adam — 2× in this paper · explained in 2 episodes
• Empirical Methods in Natural Language Processing — 2× in this paper
• Grounding — 2× in this paper · explained in 3 episodes
• In Conference on Language Modeling — 2× in this paper
• LLM judge — 2× in this paper · explained in 2 episodes
• Large Language Models (LLMs) — 2× in this paper · explained in 6 episodes
• Mechanistic interpretability — 2× in this paper · explained in 2 episodes
• Pruning — 2× in this paper · explained in 4 episodes
• Abstraction — 1× in this paper · explained in 1 episode
• Activation Addition — 1× in this paper · explained in 1 episode
• Chain-of-Thought Prompting — 1× in this paper · explained in 2 episodes
• Consistency — 1× in this paper · explained in 3 episodes
• Cosine Similarity — 1× in this paper · explained in 3 episodes
• Generalization — 1× in this paper · explained in 13 episodes
• In-Context Learning — 1× in this paper · explained in 4 episodes
• Meeting of the Association for Computational Linguistics — 1× in this paper
• Modularity — 1× in this paper · explained in 9 episodes
• Representation Engineering — 1× in this paper · explained in 2 episodes
• Singular Value Decomposition (SVD) — 1× in this paper · explained in 2 episodes

Sources

• Understanding intermediate layers using linear classifier probes
• Refusal in Language Models Is Mediated by a Single Direction
• Probing Classifiers: Promises, Shortcomings, and Advances
• Finding Transformer Circuits with Edge Pruning
• Looking Inward: Language Models Can Learn About Themselves by Introspection
• Discovering Latent Knowledge in Language Models Without Supervision
• Low-Complexity Probing via Finding Subnetworks
• Towards Automated Circuit Discovery for Mechanistic Interpretability
• Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Masks
• How do Decisions Emerge across Layers in Neural Models? Interpretation with Differentiable Masking
• Causal Abstractions of Neural Networks
• Dissecting Recall of Factual Associations in Auto-Regressive Language Models
• Detecting the Disturbance: A Nuanced View of Introspective Abilities in LLMs
• How to use and interpret activation patching
• Inference-Time Intervention: Eliciting Truthful Answers from a Language Model
• Emergent Introspective Awareness in Large Language Models
• Mechanisms of Introspective Awareness
• The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets
• Locating and Editing Factual Associations in GPT
• Are Sixteen Heads Really Better than One?

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering