ABSTRACT
The detailed structure of the human central nervous system began to be unravelled only a century ago. In the nineteenth century there were two schools contending for scientific acceptance. The dynamical behaviour and rigorous mathematical foundation of Hopfield-type neural networks with randomly generating interconnection weights was first proposed by Amari using a statistical neurodynamical approach rather than the Ising spin system of statistical mechanics. The most important property of synchronous AMNs is that the networks always converge to stable states within 40 iterations. This property is quite in agreement with the fact that it takes only a few steps for a human to recall specific information. The essential difference between synchronous AMNs and Hopfield neural networks is that the Hopfield model is an asynchronous system.
