Dereification as Dynamics: Event Selectivity, Timescales and Identifiability in a Minimal Self-Representation Model

Abstract

This preprint presents a minimal stochastic model of dereification: reducing discursive elaboration while preserving perception and functional self-representation. The model distinguishes perceptual content, self-representation and elaboration, and compares interventions matched for their reduction of elaboration following ambiguous events. The study combines mathematical analysis with computational experiments. It establishes conditions under which reducing internal elaboration and reducing its reported expression are observationally equivalent. It also examines event selectivity, intrinsic timescales and the practical identification of intervention mechanisms. In the tested population, a nonlinear gate driven by relevance evidence preserved more elaboration after relevant events than a linear controller using the same evidence, but achieved lower decision accuracy. Mechanism identification depended on the available observations and modelling assumptions, and was sensitive to between-session baseline drift. Under specified nonlinear conditions, kurtosis improved classification beyond the tested second-order features. The tested mechanisms did not reproduce the targeted empirical timescale pattern under a shared observation model. These findings concern a computational model. They do not establish mechanisms in humans or validate consciousness, Rigpa, nondual awareness or compassion. The accompanying materials provide mathematical proofs, numbered pseudocode, experimental specifications, simulation and analysis code, stored results, reproducibility instructions and integrity manifests. The separate, unfinished LLM pilot is not evidence for this study.