What makes the retina such an appealing object of investigation to neuroscientists? Obviously every student has his own answers, but still we believe all would agree on one point: the retina is comparable in complexity to the cortex itself, and yet it differs in one fundamental aspect - it is an isolated computational unit. We mean by this that its input and output can be known to a good level of accuracy, the input being an image projected by the lens onto the photoreceptors, and the output the electrical activity that can be recorded from the optic nerve. It is therefore possible, at least in principle, to determine precisely its input-output function and study the mechanisms by which this computation is achieved. Indeed this condition very seldom occurs for any later stage of cerebral processing.