Artificial neural nets: dependence of the EEG amplitude's probability distribution on statistical parameters

The statistical laws governing the output of a population of unitary generators are not explicit with regard to the effect of population size and properties of the individual generators on the summed activity. Experimental work was therefore undertaken with artificial nerve nets, the activity of which simulates with a high degree of realism individual nerve cells and the electroencephalogram. It was found that the summed activity is not affected by the statistical properties of single generators even in nets of only 200-1000 elements. On the other hand, the output of the net is highly sensitive to the level of connectivity between individual generators. When connectivity is low, the summed output is distributed in normal (Gaussian) fashion. The output of the net becomes less and less Gaussian with increase in coupling between the generators.