Computational verification of approximate probabilistic estimates of operational efficiency of random neural networks

One of the first examples of the efficient random neural network device of the physiologically plausible size has been proposed by Giles S. Brindley in 1969. The parameters of efficiency of that network were estimated, based on simple Poisson distribution-based approach. Here we report results of computer simulations of Brindley’s neural network, which enable verification of initial estimates, and provide grounds for discussion of various aspects of possible physiological implementation of the original construction of Brindley.