| It is widely believed that the array of beta (X) cell dendritic field centers constitutes the sampling grid for the X cell component of the retinal "neural image", with the dendritic field as the sampling aperture and the cell spacing as the sampling distance. Dendrites of either ON- or OFF-center X cells cover the whole retina. Hence, the X cell part of the "neural image" is sampled by two sets of cells. This dissertation aimed to improve our understanding of the role of beta cells in image representation, and addressed the following three major questions: (1) What are the morphological characteristics of the dendritic trees, and how do they depend upon retinal location? (2) How are the dendritic trees of a homogeneous beta cell mosaic arranged relative to one another, and what does this imply about the spatial arrangement of retinal sampling? (3) Are the beta cell ON- and OFF-center subsets spatially dependent and, if so, what is the spatial relationship?; A combined approach including retrograde Neurobiotin transport, Voronoi-Delaunay analysis, and computer modeling was employed to answer these questions. The basic results are: (1) The shape of dendritic field and soma was found to be elliptical, with a major-to-minor axis ratio of 1.3{dollar}sim{dollar}1.4. (2) The orientation of the major dendritic axis was found to run with the gradient of local cell density, and point roughly towards the area centralis. (3) The dendritic coverage factor for either ON- or OFF-center beta cells was found to be about 1.7 across the retina. (4) Soma lattices were found to be always less regular than corresponding dendritic mosaics, and become more regular as the eccentricity increases. (5) The regularity of dendritic mosaics was found to remain fairly unchanged across the retina, with the CV of the adjacent neighbor distribution being 0.17 which is similar to the CV of cat cone lattices. (6) The ON- and OFF-center beta cell lattices have a precise relationship to one another in the form of a spatial shift of about one half cell spacing.; The results suggest that cat beta (X) cells, with two homogeneous dendritic mosaics as regular as cat cone lattices and positioned precisely with respect to one another, are optimized for coding the highest resolution spatial information. |
