A mathematical model for the self-organization of orientation columns in visual cortex

The visual cortex contains regular arrangements of neurons responding to specific types of visual stimulation, such as ocular dominance columns and orientation columns. These columnar structures can be considered as the functional architecture for early visual information processing. The model of activity-dependent self-organization reported here demonstrates that both response properties of neurons to visual stimulation and the related columnar organization can be reproduced based on the competition between ON- and OFF-center inputs. Furthermore, the relationships between anatomical neural organization and its physiological response properties are clarified. This model also predicts new columnar organization linking the symmetry of receptive fields, which has never been examined. A hypercolumn, composed of orientation columns and symmetry columns, is suggested to serve as a basis for the representation of visual information.