Default behavior hides reflective capacity; models exhibit high gating between latent capacity and accessibility.

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

• Default presentation conflates capacity with accessibility, and most evaluation benchmarks measure only default presentation — systematically misreading models.claim0.814
  Argues current evaluation approaches are fundamentally misleading about model capabilities
• Baseline scores blend together at least three different things: latent reflective capacity, default accessibility, and stability of access.claim0.809
  Conceptual decomposition arising from the data showing different models dissociate these traits
• More inference compute amplifies both reflective capacity and safety gating; the contemplative prompt resolves gating by reframing self-referential probes.claim0.802
  Interpretation of Grok 4 vs Grok 4 Fast per-koan comparison
• We stress that in today’s models, this capacity is highly unreliable and context-dependent; however, it may continue to develop with further improvements to model capabilities.quote0.786
  Caveat and forward-looking statement from the abstract.
• Which behaviors does a model represent internally, default to, can be pushed to amplify, or refuses to expose?question0.779
  The motivating diagnostic question that prompting alone cannot answer
• Latent Reflective Capacityconcept0.768
  The maximum reflective capacity a model can reach under the right framing; separable from default accessibility
• Steering vectors capture latent dimensions of reflective behavior more faithfully than surface-level embedding similarity.claim0.767
  Supported by the instruction discovery experiments comparing steering vs. embedding baselines.
• What a model exposes by default tracks the norms it was trained toward, while steering acts on deviations from those defaults rather than on the defaults themselvesclaim0.767
  Central interpretive claim organizing the entire paper's results