The human visual system is optimised for processing the spatial information in natural visual images

A fundamental tenet of visual science is that the detailed properties of vi sual systems are not capricious accidents, but are closely matched by evolu tion and neonatal experience to the environments and lifestyles in which th ose visual systems must work [1-5]. This has been shown most convincingly f or fish [6] and insects [7], For mammalian vision, however, this tenet is b ased more upon theoretical arguments [8-11] than upon direct observations [ 12,13], Here, we describe experiments that require human observers to discr iminate between pictures of slightly different faces or objects. These are produced by a morphing technique that allows small, quantifiable changes to be made in the stimulus images, The independent variable is designed to gi ve increasing deviation from natural visual scenes, and is a measure of the Fourier composition of the image (its second-order statistics). Performanc e in these tests was best when the pictures had natural second-order spatia l statistics, and degraded when the images were made less natural, Furtherm ore, performance can be explained with a simple model of contrast coding, b ased upon the properties of simple cells [14-17] in the mammalian visual co rtex. The findings thus provide direct empirical support for the notion tha t human spatial vision is optimised to the second-order statistics of the o ptical environment.