Globally stable saturable learning laws

Abstract This article discusses minimal saturable learning laws that incorporate deassociative as well as associative behavior. The requirement of global stability results in new terms for the nodal activity equation. These terms provide temporal competition of a node with its own past history and imply spontaneous code protection and better memory storage capacity for a slab. One variant of these saturable learning laws is identified as the fringe contrast in an emulsion-based optical holographic interconnection implementation.