It may seem perverse to begin a chapter on spatial coding with an outline of Fourier analysis, but the reason is simple. If one describes stimuli in terms of the spatial profile of their luminance, then it is not clear how one can relate the coding of one profile to that of another. Fourier analysis allows one to describe any function in terms of sinusoidal functions, so one can describe the differences between functions in terms of their respective Fourier components. This approach to spatial vision was pioneered by Campbell and Robson (1968).