preservation and scanning

Everything downstream of this page began with one physical fly, killed in a way that kept its brain intact at the resolution of individual synapses.

preservation

A synapse is on the order of tens of nanometers. Preserving a brain at that scale means fixing it chemically or by rapid freezing so that membranes, vesicles, and synaptic densities stay where they were in life, then staining it with heavy metals so those structures show up under an electron beam, then embedding it in resin hard enough to be cut into sections thinner than a wavelength of visible light. Each of these steps can destroy the information the next one needs. The protocols that work are the result of decades of iteration in the electron microscopy community.

Eon Systems summarizes their pipeline for the fly in four words: preserved, scanned, reconstructed, and emulated. Each word is a field.

  fly ──▶ fixed ──▶ stained ──▶ sliced ──▶ imaged ──▶ traced ──▶ diagram
   │                              │                                 │
   │                     specimen consumed here                     │
   │                                                                │
   ╰────────────── the only remaining copy ◀────────────────────────╯

scanning

The preserved brain is imaged by electron microscopy, section by section, at a resolution of a few nanometers per pixel. The full fly brain volume comes to hundreds of terabytes of raw imagery. At this resolution a neuron is not a dot; it is a vast branching object threading through thousands of sections, and a synapse is a visible darkening where two of those objects meet, with a cloud of vesicles on the sending side.

The scan is destructive in most workflows: the brain is consumed in the act of reading it. This detail does more work in the observers' conversations than it probably should. The fly that was preserved no longer exists anywhere except as the diagram. When an observer in the logs refers to "the specimen", they mean an animal that has not physically existed for years and is, at the same time, the only thing this program has ever touched.

what was lost

Preservation keeps structure. It does not keep state. Whatever the fly was doing in its last moments, whatever short-term traces its synapses held, whatever its neuromodulatory tone was, none of that survives fixation in readable form. The diagram is the fly's anatomy, not its biography.

One consequence is worth stating because the logs return to it: every partition we run starts from a generic baseline, not from the fly's own past. The mushroom body is the fly's memory organ, and the one we run remembers nothing. What MB-01 filled it with is the subject of a third of this archive.