Neural network modeling of brain cognitive functions: review of basic ideas

Abstract

A review of basic ideas of neural network modeling of brain cognitive functions is given. Several models of the neuron are described: threshold neuron of McCulloch and Pitts, neuron with sigmoid activation function, neuron with non-monotone activation function, stochastic neuron, impulse neuron. As well different types of network architectures are described: perceptron, back-propagation network, Hopfield network, Boltzman machine. Structured models composed of more than one neural networks and used for modeling particular brain systems (hyppocampus, hippocampus – neocortex system, prefrontal cortex – basal ganglia system) are considered. In conclusuion, general principles of brain modeling are discussed.