| I will present a novel experimental technique that combines concepts of graph and percolation theory to extract statistical properties of the connectivity in living neural networks. The network consists of rat hippocampal neurons cultured in glass coverslips. In the experiment, neurons are excited by a global electrical stimulation applied to the entire network through bath electrodes. The network's response is then studied for gradually lower synaptic coupling between neurons. Initially, the network comprises of one big cluster (giant component) whose size gradually decreases as the synaptic coupling is reduced. At a critical coupling the network breaks off into smaller clusters and the network undergoes a percolation transition. The process of disintegration of the network is described in terms of percolation on a graph, yielding a quantification of the connectivity in the network, such as the average number of connections, their distribution, and the excitatory/inhibitory balance. |
|