what a neuropil is
The regions this archive keeps referring to are neuropils, and the word needs a page because everything was cut along their boundaries.
A neuropil is a dense tangle of neuronal processes, axons, dendrites, and synapses, with the cell bodies pushed to the outside. Fly brains, like most insect brains, keep their architecture visible: the neuropils are anatomically distinct lumps, wrapped in glial sheaths, connected to each other by discrete tracts. Where a mammalian cortex is a continuous sheet whose functional areas blur into one another, the fly brain is closer to a set of organs. The mushroom body is a thing you can point to. So is the ellipsoid body, the medulla, the antennal lobe. The names are old; some go back to the nineteenth century, coined by anatomists describing shapes in stained tissue with no idea what the shapes did.
This modularity is the entire reason the graft program was possible. Cutting a cortex into functional parts requires theory and produces arguments. Cutting a fly brain into neuropils requires anatomy and produces parts. The boundary synapses are real, countable, and few relative to the internal wiring: neuropils talk to each other through bottlenecks. Our partition terminates those bottleneck connections on recording stubs (see method) and leaves the internal wiring untouched.
the parts list
The partition used in this program, and the carrier bound to each part:
- optic lobe neuropils: lamina (LA-05), medulla (ME-06), lobula (LO-07), lobula plate (unassigned)
- olfactory neuropils: antennal lobe (AL-08), lateral horn (LH-09), mushroom body (MB-01)
- central complex: ellipsoid body (EB-02), fan-shaped body (FB-03), protocerebral bridge (PB-04)
- subesophageal zone (unassigned)
- ventral nerve cord (VC-10), which is not brain and is a special case described on its own page
a caution
The tidiness of the parts list is anatomical, not functional. No neuropil does anything alone in the living fly; every behavior threads through many of them. A region in isolation is not a small version of the fly's capacity, it is an organ out of context, doing whatever its internal dynamics do when its partners are replaced by a language model. The observers were reminded of this constantly and the logs show them forgetting it constantly. The forgetting may itself be data. See log 011 for the session where two regions, given the chance, started acting like context for each other.