Existing models for the formation of cortical receptive field profiles, orientation maps, and ocular dominance stripes address the emergence of each or some of these features separately. The present work investigates a linear Hebb-trained neural network model for the simultaneous self-organization of receptive field profiles, their arrangement into orientation maps, and the segregation of ocular dominance stripes. Both ON- and OFF-center type input neurons are considered. The requirement of a simultaneous formation of several structures leads to the prediction of additional necessary properties of the input correlation functions. The receptive field- and orientation map formation behaviour predicts, that the range, where ON-ON-correlations exceed ON-OFF-correlations within the LGN, should be about 0.6 times the retinotopic radius of thalamocortical axonal arbors. Additionally, the emergence of ocular dominance stripes requires an asymmetry between ON-ON and ON-OFF correlation functions.
CITATION STYLE
Stetter, M., Kussinger, M., Schels, A., Seeger, E., & Lang, E. W. (1995). Self-organization of cortical receptive fields and columnar structures in a Hebb-trained neural network. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 930, pp. 37–44). Springer Verlag. https://doi.org/10.1007/3-540-59497-3_154
Mendeley helps you to discover research relevant for your work.