Me, Myself, and $\pi$: Evaluating and Explaining LLM Introspection

Atharv Naphade, Samarth Bhargav, Sean Lim, McNair Shah

Keywords: Large Language Models, LLM Introspection, Model Calibration, AI Safety, Mechanistic Interpretability, Benchmark, Red-teaming, Jailbreaking.

TL;DR: We introduce Introspect-Bench, a cognitively grounded benchmark for policy introspection in LLMs, and show that models learn to introspect through identifiable mechanisms.

Abstract: A hallmark of human intelligence is Introspection—the ability to assess and reason about one’s own cognitive processes. Introspection has emerged as a promising but contested capability in large language models (LLMs). However, current evaluations often fail to distinguish genuine meta-cognition from the mere application of general world knowledge or text-based self-simulation. In this work, we propose a principled taxonomy that formalizes introspection as the latent computation of specific operators over a model’s policy and parameters. To isolate the components of generalized introspection, we present $\textbf{Introspect-Bench}$, a multifaceted evaluation suite designed for rigorous capability testing. Our results show that frontier models exhibit privileged access to their own policies, outperforming peer models in predicting their own behavior. Furthermore, we provide causal, mechanistic evidence explaining both how LLMs learn to introspect without explicit training, and how the mechanism of introspection emerges via attention diffusion.

Paper Type: New Full Paper

Submission Number: 75