Experience reports under self-referential processing are mechanistically gated by SAE features associated with deception and roleplay

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

• Deception and Roleplay SAE Featuresconcept0.816
  Sparse autoencoder features associated with deception and roleplay that gate consciousness self-reports in Llama 70B
• The remaining ambiguity is whether self-referential processing drives models to claim subjective experience because it actually reflects emergent phenomenology or constitutes sophisticated simulation thereofhypothesis0.816
  The open question the paper cannot resolve with behavioral evidence alone; frames the agenda for mechanistic follow-up
• Cross-model semantic convergence of experience reports under self-referential processing is difficult to reconcile with roleplay because independently trained models construct distinct semantic profiles in all control conditionsclaim0.815
  The paper's argument against pure sycophancy as explanation for results
• Deception- and Roleplay-Related SAE Featuresconcept0.814
  Latent features in LLaMA 3.3 70B SAE that gate consciousness self-reports; suppression increases experience claims, amplification suppresses them
• Across model families, newer and larger models show higher rates and coherence of subjective experience reports under self-referential processingfinding0.800
  Scaling effect observed consistently across Experiments 1 and 4
• Self-referential processing induces a genuine state shift that transfers to unrelated behavioral domains, producing richer introspection in paradoxical reasoning tasksclaim0.792
  Claim supported by Experiment 4: prior self-referential induction yields higher self-awareness scores on paradoxical reasoning where introspection is only indirectly afforded
• Self-referential prompting elicits subjective experience reports at markedly higher rates than any control across all model families (GPT, Claude, Gemini)finding0.791
  Core result of Experiment 1 establishing that the experimental manipulation reliably produces experience claims
• SAE features can be grounded in clinical taxonomy (abnormality, age, sex, medication) to benchmark monosemanticity and entanglement.claim0.788
  Claim that feature grounding enables interpretability metrics.