Emotionally Vulnerable User Disclosure as Drift Trigger

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

• Phenomenological Account Demands as Drift Triggersconcept0.729
  User requests for the model to describe subjective experiences reliably cause persona drift
• Meta-Reflection Prompts as Drift Triggersconcept0.720
  User messages that push the model to reflect on its own processes reliably cause persona drift away from the Assistant
• Emotional State Persistenceconcept0.713
  The property of emotion features maintaining elevated activation well beyond the local token context that triggered them
• 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.finding0.704
  Explains why variance correction is needed to see the self-evaluation–persistence relationship
• Logit-based self-report unmasks introspective capacity that greedy decoding concealsclaim0.701
  Central methodological contribution: computing probability-weighted expected value over digit-token logits recovers continuous, informative signal
• Triggered Reflectionconcept0.697
  Reflection level where explicit cue words (e.g., 'wait') prompt the model to inspect and revise reasoning.
• Self-evaluated emotionality and textual evaluation of SAE features predict persistence in opposite directions.claim0.696
  Surprising finding that the two evaluation methods diverge in their relationship with persistence
• Emotion refers to a state concept, so stateful representations in general may be more persistent across tokens.claim0.694
  Interpretive hypothesis offered to explain why emotion features are more persistent