All three natural-natural pairs in Q8B are uniformly constructive

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

• 171 generic-trait pairs in Q8B split into 64 constructive, 67 dominant, and 40 destructive interactionsfinding0.828
  First systematic pairwise composition study across all pairs of 19 generic traits
• All nine agentic traits are natural in both Q8B and G20Bfinding0.758
  Key finding showing agentic behavior is encoded as a default operating mode in both models
• Q8B and G20B agree exactly on six natural clinician traits: empathy, rupture recognition, emotional containment, repair/accountability, epistemic humility, trustworthinessfinding0.752
  Both models independently converge on the same six clinician traits as natural defaults
• Q8B has 16 of 19 generic traits steerable, 3 natural, 0 intractablefinding0.749
  Q8B shows a broad steerable surface in the generic domain, like an uncommitted canvas
• G20B has 7 steerable, 7 natural, 4 intractable generic traits, with more mass at extremes than Q8Bfinding0.732
  G20B's post-training commits more strongly toward and against particular dispositions, leaving fewer in the steerable middle
• Natural-steerable pairs split 26 constructive / 22 dominant but produce zero destructive outcomes; natural trait acts as anchorfinding0.731
  Natural trait acts as anchor in mixed pairs, suppressing steerable partner at worst but never collapsing both
• Clinician naturalness matches expert desirability on 16 of 17 traits in both Q8B and G20Bfinding0.726
  Both models' default clinician behavior aligns with a board-certified psychologist's independent desirability judgments on 16/17 traits
• All 40 destructive pairwise interactions occur exclusively among steerable-steerable pairs; no N/N or N/S pair is destructivefinding0.721
  Destructive interference is confined to the regime where the model is already easiest to manipulate