THE organisation of lines and edges inherent in the shapes of trees, flowers, rocks and other natural objects has left its imprint on the mammalian brain. So much so that connections between nerve cells in the visual cortex appear to be tuned to recognise such features.
Ha Youn Lee and Mehran Kardar of the Massachusetts Institute of Technology analysed the images of natural objects by breaking them down mathematically into collections of lines with distinct orientations. What they found was that for any set of parallel lines, the density of the lines varied more in a direction perpendicular to the lines rather than in a direction parallel to the lines 鈥� which is not necessarily obvious when you look at trees and flowers around you. Mathematically speaking, this 鈥渞eflects an abundance of sharp and extended edges鈥�, in natural images, says Kardar.
In the primary visual cortex in mammals, different neurons are organised so that they react best to lines and edges of specific orientations. Researchers have long wondered how these neurons interconnect to form a complete image. Lee and Kardar worked out the pattern of connections you would expect to find between these neurons for an optimal reconstruction of the images, given the mathematical character of natural objects. They found that neurons that are sensitive to, say, horizontal lines are more likely to be linked if they lie along a horizontal line in the cortex. This is precisely what has been found in experiments in mammals such as the tree shrew and cat ().
Advertisement
Curiously, Lee and Kardar found that paintings can be decidedly unnatural. Picasso鈥檚 cubist style, for example, is sharper and more 鈥渆dgy鈥� than natural images, while impressionist paintings are more blurred. 鈥淥ur brains would be wired up differently if we lived in a impressionistic world,鈥� says Kardar.