Looming Detector Neuron: Angular Size, Time-to-Collision and the LGMD Peak Rule
An object on a collision course expands on the retina of a locust. Watch the visual angle theta = 2 atan(l / vt) grow, the looming neuron respond with rate = kappa theta-dot exp(-alpha theta), and the peak arrive at the fixed time alpha l / v before collision. Change the size, speed, alpha, neural delay and motor latency in a Blender-built scene and see whether the escape trigger comes in time, with numerical checks of the peak rule built in.
This model maps a real visual field as a spherical cap: eccentricity from a central fixation point runs outward across the dome, colored by a live motion-sensitivity heatmap built from rod pooling, receptive-field size, speed tuning and light adaptation. Drag to orbit the field, adjust the stimulus and lighting sliders, and run the reaction-time trial to compare your own detection speed against the model's eccentricity-driven prediction.
An object on a collision course expands on the retina of a locust. Watch the visual angle theta = 2 atan(l / vt) grow, the looming neuron respond with rate = kappa theta-dot exp(-alpha theta), and the peak arrive at the fixed time alpha l / v before collision. Change the size, speed, alpha, neural delay and motor latency in a Blender-built scene and see whether the escape trigger comes in time, with numerical checks of the peak rule built in.
3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install