Expressive power of first-order recurrent neural networks determined by their attractor dynamics (Q736603)

The present study provides a novel theoretical approach to the crucial role that attractors and spatio-temporal patterns of discharges play in the computational capabilities of neural networks and, more generally, in the processing and coding of information in the brain. It establishes a link between the attractor dynamics of the networks, their spatio-temporal patterns of discharge, and their ability to perform more or less intricate discrimination tasks.