Accounts of conceptual change tend to run together two operations: representational innovation — new concepts, mathematics and standards of correctness — and epistemic recoupling, a change in how a system is causally coupled to what it inquires into. This paper separates them by distinguishing what a system’s interface transmits, what its current procedures access, and what it represents. Representation can outrun a fixed interface; evidential discrimination cannot. The central result is an invariance theorem: if two candidate conditions induce the same distribution over every history an interface can produce, and nothing capable of grounding certification bypasses the interface, then no change confined downstream of it — including a change of representational law — can discriminate between them or certify the difference. The paper derives such equivalences from the interface’s causal architecture, not from what the system fails to distinguish. Refining the space of hypotheses a system entertains does not, by itself, refine the experiments through which the world discriminates among them: an untransmitted distinction cannot be recovered, only acquired through a recoupling that refines what the interface transmits. For Representational Sealing Theory (ReST), the result gives an architectural route to its non-certification claim that does not need law-closure as a premise. The mathematics is elementary and is not claimed as new.
No takes yet. Share an insight, caveat, or question.
Peter Kahl (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: