[en] There is ample anatomical, physiological and psychophysical evidence that the mammilian retina contains networks that mediate interactions among neighboring receptors, resulting in intersecting transformations between input images and their corresponding neural output patterns. The almost universally accepted view is that the principal form of interaction involves lateral inhibition, resulting in an output pattern that is the convolution of the input with a ''Mexican hat'' or difference-of-Gaussians spread function, having a positive center and a negative surround. A closely related process is widely applied in digital image processing, and in photography as ''unsharp masking''. The authors show that a simple and fundamentally different process, involving no inhibitory or subtractive terms can also account for the physiological and psychophysical findings that have been attributed to lateral inhibition. This process also results in a number of fundamental effects that occur in mammalian vision and that would be of considerable significance in robotic vision, but which cannot be explained by lateral inhibitory interaction