Individual samples from trait-inducing datasets are largely separable from control samples based on persona direction projections

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

• What is the mechanistic basis for persona vectors extracted from exhibited-trait samples generalizing to causally influence the trait?question0.819
  Key open question about why the persona vector extraction method works beyond correlation
• Persona vectors capture signal specific to their assigned trait, as evidenced by within-trait correlations exceeding cross-trait baselinesclaim0.810
  Author's interpretation establishing that persona vectors are not merely general misalignment indicators
• There may exist a natural 'persona basis' characterizing the space of all model personas, with correlations between persona vectors predicting co-expression of traitshypothesis0.800
  Open question proposed by authors for future work on the dimensionality and structure of persona space
• We hypothesize that geometric explanations for persona-safety interactions are tractable and that per-model geometric audits can predict activation-steering vulnerability from trait-refusal cosine alignment.hypothesis0.785
  Forward-looking claim about the utility of the trait refusal alignment framework as a general tool
• Projection of last prompt token onto persona vector correlates r=0.75–0.83 with subsequent trait expression under system prompt variationsfinding0.782
  Main monitoring result showing persona vectors can predict behavioral shifts before text generation begins
• Hypothesis 2 (Persona Space): Persona vectors jointly compose a persona space; an instance's general dispositional profile is specified by a combination of activations along multiple persona vectorshypothesis0.782
  Second of three hypotheses about persona implementation, supported by PCA evidence from Lu et al.
• What is the mechanistic basis for persona vectors extracted from trait-exhibiting activations generalizing to causally influence the trait and predict finetuning behavior?question0.780
  Open question posed by authors about why their method works
• A model may produce internally coherent persona outputs while shifting characteristics across different generations, so intra-response and inter-response consistency must be assessed separatelyclaim0.780
  Supported by low correlation between ICatom and RCatom (r=0.44)