Self-Revising Records: Biological Memory and the Concept of a Record in Quantum Foundations

Three independent programmes in quantum foundations have converged on the record as the primitive from which classical facts are built: Feynman-rule bookkeeping of measurement outcomes, quantum Darwinism and its descendants (spectrum broadcast structures, strong quantum Darwinism), and the observerhood criteria of extended Wigner’s friend scenarios. In all three a record has exactly two states, retained or erased, and consultation of a retained record is assumed inert. Biological memory is used throughout this literature as a paradigm record, interchangeably with notes and copies. We show that biological memory does not satisfy the shared concept. A consolidated engram, on retrieval, is actively destabilised by an identified enzymatic pathway and restabilised with altered content, and whether this happens is gated by the relation between stored content and current input. The record is retained throughout, distinguishes measurement paths throughout, and is rewritten by its own consultation. This is a third record state, retained with content conditioned on consultation history, for which none of the frameworks has a category. We state what each framework assumes, exhibit the mismatch as four distinctions the frameworks do not draw, and propose a minimal readout-write channel on the record. The channel preserves non-signalling and the classical-quantum structure of the system-record state, so no quantum bookkeeping is affected, while violating Bohr non-disturbance and decrementing system-record mutual information by exactly the gap between the system entropy and the Holevo quantity of the rewritten record. In the Drosophila mushroom body every term of the channel has a named cell type, the record index splits into an odour component that every write preserves and a valence component that every write targets, and the circuit implements three distinct operations (revision in place, allocation, readout modulation) where the frameworks have one. The claim concerns what kind of record memory is, not whether the brain is quantum.