A neural network model for the formation and for the spatial structure of retinotopic maps, orientation- and ocular dominance columns
K. Obermayer, G. G. Blasdel, K. Schulten
We demonstrate that important features in the spatial structure of retinotopic maps, orientation- and ocular dominance columns in the primary visual cortex, can be explained as the result of a mapping from a five-dimensional feature space onto a two-dimensional cortical surface under the constraint that (i) the mapped features vary smoothly along the cortical surface, and (ii) the mapping is established by an activity-based self-organizing process. We generate our model maps by using the self-organizing feature map algorithm [1,2], which is known to implement the above mentioned principles in a biologically plausible way. We characterize the spatial structure of the model maps by their Fourier transform and correlation functions, and we study the interaction between both (model-) column systems, and between them and the retinotopic map. Numerical simulations are supplemented by a mathematical analysis. Results are compared with experimental data obtained from area 17 of the macaque.