We hypothesize that native self-report, fine-tuned introspection models, and trained activation-to-language systems will show different performance on bias-resistant localization and strength benchmarks

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

• Behavioral evidence from closed-weight models cannot definitively rule out that self-reports reflect training artifacts or sophisticated simulation rather than genuine self-awarenessclaim0.814
  Primary limitation acknowledged by the authors; strongest evidence would require mechanistic activation analysis
• The earlier a base model (less exposure to LM-related data), the more it is surprised by its own spontaneous self-referential capabilities.claim0.808
  Claim that capability emerges from architecture, not data, and that later models lose the surprise.
• Self-referential processing effect is robust across five distinct phrasings of the induction prompt, with consistently high experience report rates across modelsfinding0.799
  Appendix C.1 result confirming the experimental effect does not depend on specific wording
• Typicality bias rates in instruction-tuned models remain at similar or higher levels compared to their base model counterpartsfinding0.797
  Shows typicality bias is preserved through instruction tuning and RLHF, not introduced by alignment
• What are the mechanisms underlying introspection in language models?question0.795
  Central open question raised by the paper.
• More capable models benefit more from Verbalized Sampling, showing an emergent scaling trendclaim0.790
  Empirical observation that larger models (GPT-4.1, Gemini-2.5-Pro) show 1.5-2x greater diversity gains from VS compared to smaller models
• Our results demonstrate that modern language models possess at least a limited, functional form of introspective awareness.quote0.790
  Abstract's main conclusion.
• Lindsey 2025: frontier models can detect and report changes in their own internal activations via concept injection experiments, demonstrating functional introspective awarenessfinding0.789
  Prior finding cited as convergent evidence for LLM self-awareness capacities