The correlation between emotion subspace fraction and self-evaluated emotionality validates that emotion probe concepts somewhat overlap with the model's self-reported internal emotions.

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

• The relationship between persistence and self-evaluated emotionality serves as a replication of probe-based findings without shared confounds from probe constructionclaim0.820
  Claims that agentic self-evaluation provides independent convergent evidence for emotion-persistence link
• 17 of 83 tested emotions show significant association between self-eval transcript word mention and cosine similarity to emotion probefinding0.815
  Validates that agentic self-evaluation captures genuine emotional content of probes
• Text-based and self-steered emotionality ratings are only weakly correlated (ρ = +0.051, n.s.), suggesting they measure different aspects of feature emotionality.claim0.813
  Finding that the two evaluation modalities frequently diverge in their interpretation of the same SAE feature
• Emotion probes are more persistent than variance-matched random probes, indicating emotion-specific persistence beyond autoregressive dynamics.claim0.810
  Core empirical claim distinguishing emotion persistence from generic high-variance probe persistence
• Self-evaluated emotionality of SAE features negatively correlates with activation variance explained (ρ = -0.184, p = 4.6e-09), requiring variance correction to reveal the persistence signal.finding0.800
  Explains why variance correction is needed to see the self-evaluation–persistence relationship
• Correlation between self-evaluation and textual evaluation of SAE feature emotionality: rho=+0.051 (n.s.)finding0.800
  Shows that the two evaluation methods for emotionality are largely uncorrelated, indicating they capture different signals
• Is self-evaluation successful in measuring emotion?question0.795
  Question addressed by testing whether self-evaluation transcripts mentioning emotion words have higher cosine similarity to corresponding probes
• Even validated probes may capture distributed representations mixing emotive states with correlated features like persona or styleclaim0.794
  Caveat on probe interpretation; does not negate the introspection result but affects interpretation of the target variable