Abstract
Depth perception is fundamental to vision. It can be achieved by comparing the left and right retinal images, an ability known as stereopsis. To do this the brain must first determine for each point in one retinal image which point in the other originated from the same part of the same object. This issue, known as the correspondence problem, is central to binocular vision since without its solution the input from the two eyes cannot be combined. Despite engendering considerable interest, its resolution by the visual system has remained enigmatic. Here we show that for a scene consisting of a single, long binocular bar some cells in the primary visual cortex (V1) solve this problem. We present evidence that such cells are always endstopped but although endstopping is clearly necessary, on its own it is insufficient to ensure a cell solves the correspondence problem. Furthermore the cells that do match corresponding points in the two retinal images respond to visual stimuli with a characteristic timecourse. Initially the cells do not solve the correspondence problem and so their initial responses are uninformative but after approximately 120ms a suppressive mechanism develops that ensures that the cells respond only to the horizontal disparity between elements that originated from the same point in the visual environment. Clearly such cells have solved a form of the correspondence problem and so would appear to play a crucial role in binocular vision.
We thank Tamara Chuprina and David Freeman for technical assistance, and Bevil Conway, Christopher Pack and Doris Tsao for comments and discussion. This work was supported by a Helen Hay Whitney Foundation grant to P.H. and NIH grant EY 13135 to M.L.