CONCEPT
The Silicon Retina
The analog chip built by Carver Mead and Misha Mahowald in the late 1980s that performed the first stages of biological vision—edge detection, motion sensing, and luminance adaptation across enormous dynamic ranges—not by simulating these processes digitally but by embodying them in the physics of the silicon itself.
The silicon retina is the credo of
Carver Mead made flesh: you do not understand something until you can build it and make it work. In the late 1980s, Mead’s lab at Caltech set out to build, in silicon, a device that did what the first stages of biological vision do. Not a camera that records light uniformly—a sensor that processes it, as a living retina does: adapting to brightness across an enormous dynamic range, detecting edges and motion, responding to change rather than to absolute level, computing in the physics of the sensor rather than in a distant central processor. The project was driven principally by Misha Mahowald, his doctoral student, to whom Mead has carefully attributed the central ideas. The chip they produced was not a simulation of vision running in digital arithmetic; it was vision, partially, embodied in analog silicon obeying the same subthreshold