Recursive Self-Revising State Systems (RSRS): A Dynamical Framework for Identity, Memory, and Belief

We propose a minimal dynamical framework in which identity, memory, and belief are treated not as separate faculties but as observations, at different timescales, of a single recursive self-revising state system (RSRS). An RSRS carries a state on a structured landscape and updates it through a context-dependent operator that—this is its defining and only non-standard feature—can reshape the landscape itself, so that a revision alters which future states are reachable. We state the framework as exactly one term added to a standard adaptive system: where predictive processing, reconsolidation, and dynamicalsystems psychology each model the updating of a state, RSRS models the updating of the transition structure that governs future states. We give the formal apparatus the claim requires—a slow–fast stochastic model, an estimable hidden-Markov shadow, an explicit null model in which the added term is zero, and a set of stability and transition results—and a small number of falsifiable predictions. The framework's value rests on one measurement, and we state the result that would falsify it: if landscape revision cannot be distinguished empirically from ordinary state updating, RSRS reduces to a relabeling of established theory. We specify the experiment that would show this.