# Mechanisms of Introspective Awareness

arXiv: https://arxiv.org/abs/2603.21396 · HTML: https://mlsanity.com/paper/2603.21396/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability.
- Builds on 22 of its 30 references that are in the map.
- First posted on arXiv 22 Mar 2026.

## Placed in

- Mechanistic interpretability / [Activation steering methods](https://mlsanity.com/mechinterp/activation-steering-methods/)

## Builds on (most cited first, 22 of 22)

- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Representation Engineering: A Top-Down Approach to AI Transparency](https://mlsanity.com/paper/2310.01405/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Refusal in Language Models Is Mediated by a Single Direction](https://mlsanity.com/paper/2406.11717/) (2024)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets](https://mlsanity.com/paper/2310.06824/) (2023)
- ["Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Models](https://mlsanity.com/paper/2308.03825/) (2023)
- [Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models](https://mlsanity.com/paper/2403.19647/) (2024)
- [Reasoning Models Don't Always Say What They Think](https://mlsanity.com/paper/2505.05410/) (2025)
- [Persona Vectors: Monitoring and Controlling Character Traits in Language Models](https://mlsanity.com/paper/2507.21509/) (2025)
- [HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models](https://mlsanity.com/paper/2305.11747/) (2023)
- [Transcoders Find Interpretable LLM Feature Circuits](https://mlsanity.com/paper/2406.11944/) (2024)
- [Looking Inward: Language Models Can Learn About Themselves by Introspection](https://mlsanity.com/paper/2410.13787/) (2024)
- [Tell me about yourself: LLMs are aware of their learned behaviors](https://mlsanity.com/paper/2501.11120/) (2025)
- [Emergent Introspective Awareness in Large Language Models](https://mlsanity.com/paper/2601.01828/) (2026)
- [Convergent Linear Representations of Emergent Misalignment](https://mlsanity.com/paper/2506.11618/) (2025)
- [Latent Introspection: Models Can Detect Prior Concept Injections](https://mlsanity.com/paper/2602.20031/) (2026)
- [Simple Mechanistic Explanations for Out-Of-Context Reasoning](https://mlsanity.com/paper/2507.08218/) (2025)
- [Steering Awareness: Detecting Activation Steering from Within](https://mlsanity.com/paper/2511.21399/) (2025)
- [The Consciousness Cluster: Emergent preferences of Models that Claim to be Conscious](https://mlsanity.com/paper/2604.13051/) (2026)

## Built on by (most cited first, 7 of 7)

- [Can LLMs Introspect? A Reality Check](https://mlsanity.com/paper/2605.26242/) (2026)
- [Reported Confidence in LLMs Tracks Commitment More Than Correctness](https://mlsanity.com/paper/2606.29490/) (2026)
- [Phase Transitions in Driven Informational Systems: A Two-Field Perspective on Learning Theory and Non-Equilibrium Chemistry](https://mlsanity.com/paper/2605.16325/) (2026)
- [Unsupervised Features Mining via Activation Geometry](https://mlsanity.com/paper/2607.04222/) (2026)
- [Self-Referential Induction Increases Response Instability Relative to Unresolvable and Verifiable Questions in Large Language Models](https://mlsanity.com/paper/2608.13258/) (2026)
- [From cacophony to hierarchy: a principled framework for assessing AI consciousness](https://mlsanity.com/paper/2609.35618/) (2026)
- [Anosognosia in LLMs: Probing Self-Awareness of Quantized Computational Substrate](https://mlsanity.com/paper/2610.06174/) (2026)

## Abstract (arXiv)

Recent work has shown that LLMs can sometimes detect when steering vectors are injected into their residual stream and identify the injected concept -- a phenomenon termed "introspective awareness." We investigate the mechanisms underlying this capability in open-weights models. First, we find that it is behaviorally robust: models detect injected steering vectors at moderate rates with 0% false positives across diverse prompts and dialogue formats. Notably, this capability emerges specifically from post-training; we show that preference optimization algorithms like DPO can elicit it, but standard supervised finetuning does not. We provide evidence that detection cannot be explained by simple linear association between certain steering vectors and directions promoting affirmative responses. We trace the detection mechanism to a two-stage circuit in which "evidence carrier" features in early post-injection layers detect perturbations monotonically along diverse directions, suppressing downstream "gate" features that implement a default negative response. This circuit is absent in base models and robust to refusal ablation. Identification of injected concepts relies on largely distinct later-layer mechanisms that only weakly overlap with those involved in detection. Finally, we show that introspective capability is substantially underelicited: ablating refusal directions improves detection by +53%, and a trained bias vector improves it by +75% on held-out concepts, both without meaningfully increasing false positives. Our results suggest that this introspective awareness of injected concepts is robust and mechanistically nontrivial, and could be substantially amplified in future models. Code: https://github.com/safety-research/introspection-mechanisms.