Analysis of real neuronal circuits in the vertebrate central nervous system

A method has been developed which permits the reconstruction of real neuronal systems using the time series of multiple-unit spiking. Two categories of spike activity are selected. The first category consists of high-amplitude, well defined spikes produced by the single leading cell of the system. The second category consists of spike activity well below the amplitude of the leading cell, but above the noise of the system. This category reflects the mass activity of neurons within the vicinity of the leading cell. Auto-correlation and cross-correlation are calculated between and within the two levels of spiking. Only nonredundant, statistically significant interspike intervals are selected for analysis. The intervals are then attributed to pairs of spikes in the original record in order to assemble a network of real neuronal interactions. Using this method, networks have been modeled from the following brain areas: the visual cortex of the cat, the nucleus raphe magnus of the rat, and the medial geniculate body of the rat