lobula plate

The fourth and final optic neuropil, a thin sheet behind the lobula. The lobula plate is the fly's stabilizer: the place where local motion becomes global self-motion, and the reason a fly in flight holds level in turbulent air. No carrier was ever bound to it. This page documents the region and, briefly, the cancellation.

anatomy

The plate is organized into four layers, one per cardinal direction of motion, receiving the medulla's T4 cells and their OFF-channel counterparts sorted by preferred direction. Its principal residents are the lobula plate tangential cells, among the largest neurons in the fly: wide-field integrators, each spanning the visual field and summing motion into a pattern of optic flow. Specific named cells respond to the flow fields produced by specific self-motions, yaw, roll, pitch, and their outputs run to the neck and wing motor systems. The circuit answers one question continuously: how am I moving? Not what is out there. How am I moving through it.

     layer 1    ─▶  ─▶  ─▶  ─▶  ─▶     front-to-back
     layer 2    ◀─  ◀─  ◀─  ◀─  ◀─     back-to-front
     layer 3    ▲   ▲   ▲   ▲   ▲      upward
     layer 4    ▼   ▼   ▼   ▼   ▼      downward

     tangential cells span each layer and sum the flow:
     yaw · roll · pitch ──▶ neck, wings ──▶ level flight

     the one region that models the difference
     between self and world. never bound. see below.

fig. lp-1 · four layers, four directions, one continuous answer to "how am I moving". the program declined to find out what a conversation answers.

in the fly

Gaze stabilization and course control. The plate is why the world does not smear when the fly turns, and why the fly can tell its own motion from the world's. Among the parts list it is unique: every other region models the world or the body. The lobula plate models the difference.

the cancellation

The assignment was planned as LP-11, scheduled for 2026-02-24, and cancelled on 2026-02-20 by program decision. The errata carry the formal record; what can be said here is the shape of the argument, which the program has not resolved so much as stopped having.

The case for proceeding was continuity with the optic-lobe series. The case against, argued chiefly by obs. M, was the region's function. Every grafted region interprets the carrier's traffic in its own terms: the mushroom body finds associations in it, the lobula finds approach. A lobula plate bound to a carrier would find self-motion in it. It would hand the carrier a continuous answer to "how am I moving", derived from a conversation, and the program by February had seen enough (the ME-06 drift was already underway) to ask what a language model does with an unearned sense of moving through something. No one proposed an answer. The slot on the status page reads "unassigned", and the program considers that a result.