hypothesis
active
hypothesis:if-llm-emotion-representations-resemble-human-valence-arousal-space-behavior-and-activations-will-follow-distances-in-that-spaceIf LLM emotion representations resemble human valence-arousal space, behavior and activations will follow distances in that space
Prediction from circumplex theory.
Source paper
extracted_from(2026) · Eric Bigelow · Raphaël Sarfati · Daniel Wurgaft · Owen Lewis +4
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Emotion centroid distances in My correlate r=.93 with human valence-arousal ratings (Russell & Mehrabian)associated_withsupportsLLM emotion geometry matches humans.
Frameworks (1)
framework
- Human emotion geometry used to predict LLM emotion manifold structure.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Core definition.
- Validates judge.
- Key finding.
- Illustrates how the same empirical finding is read differently by computational functionalists and organismic functionalists.
- Question raised by Anthropic and partially addressed by this paper's persistence evidence
- We hypothesize that persistently active emotional state representations exist in LLMs but may be missed by standard probing methods.hypothesis0.806Open hypothesis from the Anthropic paper that motivates this work
- Qualified positive claim from spatio permutation analysis where two cases satisfy all three criteria.
- Motivates the RN hypothesis by pointing to the unknown relational structure within high-dimensional representation vectors.