Does a label survive the loop? A screen of necessary physical conditions for a self-measuring system, with a pre-registered test on fly connectomes
Theories that tie experience to a system that re-reads its own state rarely say what a physical substrate must supply for such a loop to run. We state necessary conditions as measurable bounds in one currency: the number of rounds, each one write, hold and re-read, that a written label survives.A floor follows from fundamental constants. Tier 1 asks whether a carrier can hold the label; it is the published quality-factor screen, restated in rounds and extended to a population code. Tier 2 bounds how many rounds a quantum resource survives in a loop that keeps a record. Tier 3 asks whether a network contains a compartment, a set of neurons that a random walk leaves more slowly than it mixes inside. Tier 4 compares what the loop and an outside observer can recover. We implement Tiers 3 and 4 on three connectomes of Drosophila. Exploration of two animals singled out the heading system of the ellipsoid body. Five criteria were then registered and run once on a third animal, and all were met with all connections counted. One was met narrowly, and two fail in that data set when connections of one or two synapses are removed, which we trace in part to synapse detection. As a population code the heading label holds about two bits for seven seconds. The screen can exclude substrates. Passing it is not evidence of experience.