ns2250225/fly-flappy

This browser-based experiment runs Flappy Bird under control of a frozen fruit-fly connectome, with PURE BRAIN, ONLINE TRAIN, and FAST MODEL modes. The main workflow is: the player starts a flight; the app advances the game and sends captured state to the brain worker; the worker encodes sensory input, simulates neural activity, decodes a flap decision, and returns activity and training results; the app applies the decision and renders the game and neural map. Training/model management and experiment history are additional branches. Worker internals and several UI handlers are only partly sampled, so uncertain links are omitted.