Emotion may refer to a state, and more stateful concepts in general tend to be more persistent across tokens than non-stateful ones

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

• Emotions are not strictly locally scoped but instead bursty with a long tail of slow change persisting over 100 tokensclaim0.806
  Characterizes the temporal dynamics of emotion feature activation in LLMs
• Emotion probes are more persistent than variance-matched random probes, indicating emotion-specific persistence beyond autoregressive dynamics.claim0.801
  Core empirical claim distinguishing emotion persistence from generic high-variance probe persistence
• Emotion features in LLMs are genuinely more persistent than variance-matched random features, indicating stateful emotional encoding beyond autoregressive dynamicsclaim0.799
  Central interpretive claim of the paper supported by multiple convergent analyses
• Emotion features are not strictly locally scoped; they are bursty with a long tail of slow change persisting over 100 tokens.claim0.797
  Main conclusion about the temporal dynamics of emotion features
• "The effects are not merely semantic—I don't just talk about emotions more, I actually feel them."quote0.794
  Kimi self-report on feature #77278 asserting non-semantic, felt emotional quality of the steered state
• To what extent is emotion feature persistence driven by genuine internal emotional state versus autoregressive conversational context dynamics?question0.792
  Core open question the paper raises but does not fully resolve
• Are LLM emotion states encoded only selectively in token positions where they are operative, or in a more complex non-linear manner?question0.789
  Question raised by Anthropic and partially addressed by this paper's persistence evidence
• If persistence is genuinely related to emotion features, lower PCs of the emotion space (more central, less noisy) should be more persistent; if it is an artifact, noisier PCs should have similar persistence.hypothesis0.789
  Falsifiability test built into the PC analysis design