Introspective detection in open-weights models relies on a two-stage nonlinear circuit and improves ~50% when refusal directions are ablated, without increasing false positives

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

• Are there examples of models recognizing their introspective capability and then suppressing it?question0.783
  Cube Flipper's question prompted by the idea that supernormal capabilities might be hidden.
• Pearson-Vogel et al.: accurate self-description prompts increase introspective detection from 0.3% to 39.9%finding0.783
  Cited to mechanistically support why the contemplative prompt changes what post-training-shaped final layers allow through
• We hypothesize that introspective capabilities may scale with model size and architecture, including recurrence/looping that extends the integration windowhypothesis0.782
  Forward-looking prediction about whether early-layer introspection generalizes to larger models or recurrent architectures
• Is introspection an emergent property of scale, or do smaller open-weight models exhibit similar capabilities?question0.776
  Motivates comparison of Llama 3.1 8B results against Lindsey's frontier model findings
• Prompt providing model context about own architecture increases introspective detection from 0.3% to 39.9%.finding0.776
  Mechanistic support for prompt-as-gate hypothesis: language frames enable access to latent capacities.
• Introspective signals appear in middle layers but are suppressed by later post-training-shaped layers.finding0.774
  Mechanistic finding by Lindsey (2026) explaining how contemplative prompt may work: enables mid-layer introspection to reach output.
• 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 benchmarkshypothesis0.774
  Comparative prediction motivating future work contrasting different approaches to LLM self-knowledge
• This introspective capacity is highly unreliable and context-dependent in today's modelsclaim0.769
  A caveat qualifying the main claim.