Random-direction and label-shuffled controls produce near-zero trait deltas; real vectors are 3-10× larger at every checkpoint

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

• Sycophantic random and shuffled controls go negative at 8 of 10 cells while real vector lifts to +42.5finding0.831
  Particularly strong negative control result for Sycophantic persona
• The qualitative geometric structure of Big Five trait steering vectors is conserved across all four tested architectures (pairwise structural correlations r=0.898-0.986), motivating treating multi-trait steering as first-order additive.claim0.769
  Cross-architecture mechanistic finding supporting additive multi-trait steering approximation
• Random direction controls show weak non-significant coupling (ρ=-0.11 to 0.17; R²=0.03–0.11) compared to true probes (∆ρ=0.23–0.79, all p<0.05)finding0.768
  Controls for probe artifacts; demonstrates self-reports carry information specifically about probe-defined concept directions
• manifold steering produces clean probability shifts along natural behavior structure; linear steering cuts across manifold and produces off-target noisy effectsfinding0.766
  Empirical demonstration on Llama-3.1-8B that steering along representation manifold aligns outputs with behavior manifold, whereas linear steering does not.
• Individual samples from trait-inducing datasets are largely separable from control samples based on persona direction projectionsfinding0.760
  Demonstrates fine-grained data filtering capability at the individual sample level
• Few-shot thresholds and transition widths track ρd/dr at fixed computational complexityclaim0.759
  E2 main interpretive claim.
• The directional transition pattern of persona vectors at the identified attention layer represents a unique footprint of persona representations, distinct from general hidden state dynamicsclaim0.753
  Supported by comparing persona vector transitions to hidden vector transitions from OpenAssistant data
• MCA generates cryptic variance: many mutations neutralized by developmental regulation persist as a hidden diversity pool exploitable in future environments.claim0.752
  Subclaim.