Recursion of Representation Under Constraint

Abstract Models are representations produced by systems embedded in constraints. The moment representations begin to recurse—models modeling models—the dynamics split: 1. Narrative inflation: coherence increases, discriminative bite decreases, and the system becomes self-sealing. 2. Invariant convergence: alternatives get pruned, survivorship sets shrink, and stable structures emerge. This paper proposes recursion of representation under constraint as a minimal, cross-domain unit of epistemic analysis. The thesis is epistemic, not metaphysical: persistence, identity, and apparent objectivity arise when recursive modeling survives enforceable constraint pressure. When constraint is absent or bypassable, representation expands without cost and collapses into aesthetic coherence. When constraint is active, independently specifiable, and enforced, representation narrows into regime-relative invariants. The point here is not to declare what reality is. The point is to formalize how representations stabilize without becoming a totalizing ontology. One-sentence thesis: Stable invariants emerge when recursive representation is pruned by independently specifiable constraints that make some representations impossible to maintain. ⸻ 0. How to Read This Paper This paper is infrastructure. It is not a worldview. I am not claiming that “reality is representation,” that “mind is fundamental,” or that “recursion is metaphysical ground.” I’m describing how modeling systems stabilize—or fail to stabilize—when recursion is exposed to constraint. It is standalone: the primitives, constraints, and failure modes are defined internally. It is also compatible with the constraint-first / bridge-language discipline where falsifiers, cash-out, and anti-replacement are non-negotiable hygiene.