Can large language models introspect—that is, accurately detect perturbations to their own internal states?

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Do large language models monitor their own internal states?question0.874
  Framing question that motivates the entire paper
• Can language models genuinely introspect on internal states or only confabulate?question0.866
  Central research question animating the paper: distinguishing genuine introspection from illusion through causal manipulation of activations.
• Can 'Consciousness' Be Observed from Large Language Model (LLM) Internal States? Dissecting LLM Representations Obtained from Theory of Mind Test with Integrated Information Theory and Span Representation Analysisconcept0.816
  The primary paper being extracted — applies IIT 3.0 and 4.0 to LLM representation sequences derived from ToM test data to investigate whether consciousness phenomena can be observed.
• "Our findings demonstrate that LLMs can compute meaningful functions over perturbations to their internal states, establishing introspection as a real but layer-dependent phenomenon that merits further investigation."quote0.815
  Central thesis statement of the paper
• Large language models develop surprisingly coherent yet often rigid internal preferences as they scalefinding0.815
  Mazeika et al. finding reinforcing the need for emptiness-based flexible value architectures
• Our results demonstrate that modern language models possess at least a limited, functional form of introspective awareness.quote0.809
  Abstract's main conclusion.
• The inability for autoregressive large language models to maintain states of long-range order resembles tangential speech or derailment in formal thought disorder.claim0.807
  Analogy between LLM incoherence and schizophrenia symptoms
• Emergent Introspective Awareness in Large Language Models (Lindsey, 2025)concept0.806
  Related work demonstrating LLM introspective capabilities with scale-dependent pattern paralleling ESR