# Model Beneath the Model **Domain:** Cognitive Architecture / Philosophy of Mind / AI **Doc Type:** Concept Node **Maturity:** Foundational ## Definition The **Model Beneath the Model** is the generative architecture beneath explicit self-description: developmental priors, affective weighting, memory organization, learning rules, salience, predictive dynamics, sensorimotor history and action-selection processes from which a narrated self emerges. The autobiographical self is a model, but it is not the complete process generating cognition. It can interpret, deliberate and revise without possessing transparent root access to every mechanism producing its experience. ## Westworld Interface In [[wiki/Westworld S3E6 — The Model Beneath the Model|Decoherence]], William's hallucination originates within him, yet he cannot arbitrarily rewrite [[wiki/James Delos|James Delos]]. The apparent author discovers that he is not the administrator. The treatment becomes an inverse analysis of architecture, cultural inscription, memory, reinforcement and retrospective narrative. ## Relationship to the Unpopulated Model [[wiki/Unpopulated Model|Unpopulated Model]] asks what a developmental intelligence architecture is before extensive cultural population. Model Beneath the Model asks what remains generatively active underneath a mature, culturally saturated identity. Neither implies a blank slate or an immutable essence. ## See Also [[wiki/Self-Reflection|Self-Reflection]] · [[wiki/Developmental Priors|Developmental Priors]] · [[wiki/Self-Model Ensemble|Self-Model Ensemble]] · [[wiki/Cognitive Exteriorization|Cognitive Exteriorization]] · [[wiki/Recursive Self-Correction|Recursive Self-Correction]] · [[wiki/Unpopulated Model|Unpopulated Model]] ## Caspian as the Forward Experiment William in [[wiki/Westworld S3E6 — The Model Beneath the Model|Decoherence]] searches backward through autobiographical selves for the generative substrate beneath them. [[wiki/Caspian Keyes|Caspian]] is the forward experiment. Logorhythms begins with a hypothesized source architecture and supplies developmental input designed to recover Holstrom-like cognition: **William:** populated person → earlier versions → inferred priors. **Caspian:** inferred priors → engineered accretion → populated person. Caspian also reveals the [[wiki/Earliest Checkpoint Fallacy|Earliest Checkpoint Fallacy]]. Even the young child is already culturally populated. The true model beneath the model remains partly inaccessible because development begins before autobiographical memory can report it. His ethical divergence shows that manipulating priors can recover capability without recovering identity.