Association by synaptic facilitation in highly damped neural nets.

Cognitive functions are sought in a homogeneous, randomly connected net of neuron-like elements. Information is assumed to be contained in the instantaneous states of the system, which specify the firing states (off or on) of each neuron in the net. The hypothesis of synaptic facilitation is assumed to be the basis of learning and memory. Owing to the high degree of damping no reverberations occur in the net. However, close analogies can be found between the performance of the net and known association functions of the cerebral cortex, among them various types of conditioned reflexes. The data are obtained by a combination of mathematical analysis and computer simulation. It is emphasized that the biological entity simulated by this model is at best a limited component of the cerebral cortex.