Persona Role-Play MFQ Elicitation Protocol

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

• Contrastive Persona Vector Extraction Protocolmethod0.742
  Named procedure for extracting persona vectors from mean residual-stream activation differences between trait-expressing and non-expressing responses
• Role-play prompting techniquemethod0.730
  Method of eliciting specific personas from an LLM through prompt design.
• Persona Fidelityconcept0.727
  The degree to which an LLM's generated responses consistently reflect an assigned persona
• The model is able to express many subforms of a persona, with different elicitation methods eliciting different manifestationshypothesis0.723
  PSM-derived hypothesis supported by discourse-type facet analysis
• Mathematical Personaconcept0.720
  One of 11 character training personas: precise, pattern-spotting assistant obsessed with logic
• Toxic persona role-play produces profiles that reduce individualizing foundations rather than saturating all foundations near ceiling, not reproducing insecure fine-tuned profilesfinding0.716
  Rules out the simple alternative explanation that insecure models merely resemble a generic toxic character
• Persona vector extraction via contrastive activationmethod0.711
  Method of extracting persona vectors by contrasting activations when model is prompted to exhibit vs suppress a trait
• A linear preference vector encoding how much a model likes a given task is persona-relative: it activates strongly for creative writing in the assistant but for phishing in the evil personafinding0.701
  Evidence that core representations like preferences are persona-relative, supporting claim that personas gate content of representations