Modeling the development of orientation selectivity, maps, and the associated recurrent circuit

ABSTRACT Much evidence suggests that the development of orientation selectivity and maps in primary visual cortex (V1) is instructed by spontaneous patterns of input activity, without the necessity of visual experience. Evidence also suggests that this development is driven by a competition between ON- and OFF-center inputs, that leads individual cortical cells to become "simple cells", meaning they receive inputs from retinotopically adjacent subregions of ON-center and OFF-center inputs. A prominent model proposed statistics of spontaneous input activity that would lead to that outcome through "Hebbian" plasticity, in which "neurons that fire together wire together". But experiments have shown that those statistics are not present. We have discovered how the observed statistics can lead to development of simple cells. We now propose to build on that model to account for development of map organization, ON-dominance and anchoring of receptive fields, feature-specific recurrent connectivity, co-tuning of excitation and inhibition received by cells, and other observed phenomena. We will build on our own discoveries of cortical connectivity that produces many nonlinear cortical response properties, and the discoveries of others on the outcomes of co-development of feedforward and recurrent connections. In so doing we aim to build a comprehensive account of the mechanisms leading to the V1 circuitry that yields orientation selectivity and orientation maps as well as a variety of nonlinear response properties.