{
  "slug": "evaluation-state",
  "url": "/papers/evaluation-state/",
  "title": "Evaluation State in Language Models",
  "type": "working-paper",
  "typeLabel": "Working paper",
  "lifecycle": "proposal",
  "epistemicStatus": "conceptual-framework",
  "statusLabel": "Research programme · five hypotheses, experiments proposed, none run",
  "programme": "cognition",
  "question": "What evidence would show that a model represents evaluation as a shared internal state rather than reacting to surface cues?",
  "date": "2026-07-10",
  "version": "v0.1",
  "authors": [
    "Latent Minds Institute"
  ],
  "methods": [
    "layer-wise probes vs lexical baselines (proposed)",
    "cross-family and cross-task transfer (proposed)",
    "activation steering with capability controls (proposed)",
    "Jacobian-lens readouts (proposed)"
  ],
  "models": [
    "open-weight instruction-tuned models (proposed)"
  ],
  "evidence": "literature synthesis and experimental framework; no original empirical results",
  "code": null,
  "data": null,
  "related": [
    "model-entrenchment",
    "interpretability-map"
  ],
  "provenance": null,
  "canonicalUrl": "https://latentmindsinstitute.com/papers/evaluation-state/",
  "apiUrl": "https://latentmindsinstitute.com/api/research-objects/evaluation-state.json"
}