— saved image
Shunya Resear... @shunyaresear... · 29m Can connectomes help us design neural prostheses? 🧠🦾🪰 We tried to design one for a fruit fly. We give a simulated blind fly a camera and a hand-built motion detector, we then wired everything downstream of that straight into its connectome. Here's how we did it: A fly jumps away from anything approaching it, the circuit behind this is small and well mapped: motion detectors feed LPLC2 looming detectors which then feed the Giant Fiber triggering the jump. Blindness (regardless of the underlying cause) removes the early parts of the above mentioned pathway. We design a prosthesis with a camera, a model compound eye with 721 facets and an encoder that computes motion signals the fly's own cells would've sent. [Embedded video, 0:33, labelled "Demo": a simulated orange fruit fly on a checkered floor, with side panels labelled Compound Eye, Encoder, LPLC2 Connectome and Giant Fiber] 2 replies · 3 reposts · 5 likes · 49 views Shunya Resear... @shunyaresear... · 28m Everything after the encoder comes from the new male fly connectome (Janelia, Google Research, MRC LMB, Cambridge).
Note from Claude Sonnet 5
Two-part thread by Shunya Research on designing a neural prosthesis for a simulated blind fruit fly by wiring a camera, 721-facet model compound eye and motion encoder into the male fly connectome (LPLC2 looming detectors to Giant Fiber jump circuit). The embedded demo video shows an orange simulated fly on a checkered floor with panels for compound eye, encoder, LPLC2 connectome and giant fiber.
connectomefruit flyneural prosthesisbrain simulationlplc2neuroscience