Either introspection is an emergent capability requiring larger scale, or more stringent controls are needed to test introspection in smaller models

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

• Is introspection an emergent property of scale, or do smaller open-weight models exhibit similar capabilities?question0.882
  Motivates comparison of Llama 3.1 8B results against Lindsey's frontier model findings
• We hypothesize that introspective capabilities may scale with model size and architecture, including recurrence/looping that extends the integration windowhypothesis0.859
  Forward-looking prediction about whether early-layer introspection generalizes to larger models or recurrent architectures
• Introspective capabilities have threshold effects requiring very large models; 70B models are barely on the threshold, and independent researchers lack access to larger models.claim0.848
  Practical bottleneck explaining why these phenomena are not widely studied.
• Introspection relies on general-purpose computational mechanisms—attention-based anomaly detection and residual stream dynamics—rather than specialized introspection circuitsclaim0.847
  Interpretive claim about the mechanistic substrate of introspection in LLMs
• Introspective capabilities may continue to develop with further improvements to model capabilitiesclaim0.840
  Forward-looking statement about future models.
• This introspective capacity is highly unreliable and context-dependent in today's modelsclaim0.827
  A caveat qualifying the main claim.
• Introspective ability can be decomposed into: (i) information available about internal state and (ii) capacity to transform that signal into precise output reportsclaim0.826
  Conceptual distinction motivated by entropy analyses showing probe and report entropy can diverge under steering
• Introspective ability is concept-specific: quality differs across emotive concepts and the same intervention helps some concepts but not othersclaim0.825
  Cross-concept steering results; only 2 of 12 non-diagonal cells show significant introspection improvement