A dynamic model for neural networks is proposed, which explicitly takes into account the existence of several time scales without discretizing the time. The dynamics is neither totally synchronous nor totally asynchronous, and may be regarded as more realistic. A simple but unphysical case of symmetric couplings has been analyzed to display desirable features similar to those of the Hopfield model. This supports the "universality" hypothesis that the properties are insensitive to the details of the model.

• Received 12 July 1988

DOI:https://doi.org/10.1103/PhysRevLett.61.2809