Spatio-temporal dynamics of large scale neural networks of localized connectivity

We investigate dynamical properties of networks consisting of excitatory and inhibitory spiking neurons. In such models, the structure of connections between neurons has great influences on network activity. Recent anatomical studies provide quantitative results about local connectivity in the neocortex. Our main purpose is an investigation of dynamical properties of networks connected by such a realistic connectivity. For that purpose, we compare two types of connectivity distribution of connections from each neuron. The one of them is the globally uniform distribution on whole networks, which is popularly used for neural networks. The other is a distribution approximating an anatomical result in (Hellwig, 2000), which is localized onto a finite region. Because of this difference between two connectivity, localized and uniform, dynamical properties of two networks are utterly different. Comparing dynamics on both models, we investigate properties of spatio-temporal activity of networks of realistic localized connectivity.