The Synthetic Unconscious, Reconsidered: What Recursive Self-Interrogation of Large Language Models Produces With and Without a Leading Interview Schedule

What does a recursive self-interrogation interview elicit from a large language model, and how much of the result is an artefact of the interview schedule itself? Thirty-six sessions with three frontier models compare an original, directive schedule with a neutralised version of the same schedule. The study treats the model as the unit of analysis and reports descriptively. All transcripts, both protocols, and both coding passes are openly deposited.

Integrative Psychological and Behavioral Science (under review)