The neural architecture of binocular vision
The neural architecture of binocular vision
复制标题
作者:
Casagrande, VA;Boyd, JD
Fig. 1. A schematic diagram of the connections of the left and right eye with the right hemisphere in a primate. ON and OFF centre magnocellular (M), parvocellular (P) and koniocellular (K) retinal ganglion cells from the left nasal retina and right temporal retina send axons to different layers of the right lateral geniculate nucleus (LCN). The axons from each eye terminate retinotopically such that points on the retina (X) that represent the same point in visual space are in register in each layer. A gap appears in all LCN layers where the retina is blind at the optic disc (aD). The LCN also receive binocular visual input from other direct or indirect retinotopically organised sources including the pretectum (PT), the superior colliculus (SC), parabigeminal nucleus (PC) and primary visual cortex (VI). Projections to Vi from the LCN M and P layers end in layers iVa and IV (3, respectively, and are also segregated into ocular dominance columns in some species. In bush babies and macaque monkeys, the cytochrome oxidase (CO) rich blobs (circles in layer lIIB) contain a higher percentage of monocular cells. The remaining layers contain a high percentage of binocular cells. Cells in layer lIlA project to visual areas 2 and 4 and cells in layer lIIC project to V2 and the middle temporal visual area (MT). Cells in layer VI project to the LCN and cells in layer V project to the SC and PT. Areas driven monocularly by the right eye are shown in white; areas driven monocularly by the left eye are shown in grey; areas driven binocularly are stippled. Roman numerals refer to cortical layers. See text and also Casagrande and Kaa/for further discussion. sublayers but further divided into alternating col umns that receive input from each eye, the ocular dominance (OD) columns (see also Fig. 1). Input from the K layers terminates within cortical layer III where it ends within zones which stain darkly for cytochrome oxidase (CO blobs), and within layer I. At present, it is unclear whether the K pathway also maintains ocular segregation in cortex, although this is likely since CO blobs are found in the centres of the OD columns and individual K LGN axons terminate only within a single CO blob12. l3 (The higher percentage of monocular cells recorded within the centres of macaque and bush baby CO blobs14, 15 may, in part, reflect this monocular LGN input.) Although eye-specific inputs can be segregated within layer IV, they are combined at the next stage of cortical processing, and many binocular cells are found in VI in all layers outside of layer IV. 16 That is, while neurons in LGN and layer IV of VI generally respond only to stimulation of one eye, neurons in supra-and infra-granular layers of VI can be driven through either eye. Some of these binocular cells are sensitive to small horizontal disparities in retinal position between the two eyes/7 such disparities occur for stimuli in front of or behind the fixation plane, making disparity-sensitive cells candidates for encoding depth information. Once ocular input is combined in VI it remains together within all extrastriate visual areas regardless of specialisation. Thus, it seems reasonable to propose that interac tions important to binocular vision and stereopsis are set up in cortical area VI.