the first digital animal

Before this program cut the emulation into parts, Eon Systems ran it whole, in a virtual body, and reported what it did:

"Now in its digital state, it responds to light, navigates, grooms, walks, and feeds. No hand-coded behaviors. Just brain structure producing brain function."

Eon Systems, eon.systems

They call it the first digital animal. The phrase is precise. Not a simulation of an animal, in the sense of an animation authored to look lifelike, and not an artificial agent trained to act like a fly. A specific animal's nervous system, recovered from the specific animal, producing that animal's behavior in a body it no longer has.

The behavior list rewards slow reading. Responding to light is a reflex, and reflexes were expected to survive. Navigating means the heading circuits of the central complex hold and use a direction. Feeding means internal-state circuits drive goal-directed action. Grooming is the strange one. Grooming is what flies do when nothing else demands action: a self-maintenance routine, sequenced (antennae, then head, then wings), directed at the fly's own body. A digital fly that grooms is a digital fly that has, in whatever minimal sense, a self to maintain. Log 006 exists because of grooming, and it is the log after which this program stopped feeling like an engineering project.

whole versus parts

The whole emulation behaves like a fly because it is arranged like one: sensors feed the sensory neuropils, the central structures integrate, the descending neurons drive a body. Our program broke that loop deliberately. A partition has no body and no world. Its inputs arrive from a carrier instead of from eyes and antennae, and its outputs go to a carrier instead of to legs.

The naive prediction was that a region without its animal is just a signal generator, and the early idle runs (see method, step 2) seemed to confirm it: isolated regions settle into spontaneous rhythms, interesting only to an electrophysiologist. The naive prediction did not survive binding. A region coupled to a carrier is coupled to something that reacts, and a loop, once closed, starts doing what loops do. Where the whole fly produced walking and grooming, the grafted regions produced whatever the carrier's medium allowed. The carriers' medium is language. The regions section documents, region by region, what each part of a fly does when the only body it has is a conversation.