{
  "id": 7216,
  "url": "https://arxiv.org/abs/2603.22295v1",
  "title": "Whether, Not Which: Mechanistic Interpretability Reveals Dissociable Affect Reception and Emotion Categorization in LLMs",
  "summary": "Large language models appear to develop internal representations of emotion -- \"emotion circuits,\" \"emotion neurons,\" and structured emotional manifolds have been reported across multiple model families. But every study making these claims uses stimuli signalled by explicit emotion keywords, leaving a fundamental question unanswered: do these circuits detect genuine emotional meaning, or do they detect the word \"devastated\"? We present the first clinical validity test of emotion circuit claims u",
  "authors": "Michael Keeman",
  "category": "research",
  "topics": "safety-alignment,healthcare",
  "orgs": null,
  "regions": null,
  "published_at": "2026-03-15T15:11:45.000Z",
  "fetched_at": "2026-07-14T16:33:03.573Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/7216",
  "original_url": "https://arxiv.org/abs/2603.22295v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}