STORE WORKING-MEMORY NETWORKS FOR STORAGE AND RECALL OF ARBITRARY TEMPORAL SEQUENCES

Neural network models of working memory, called ''sustained temporal o rder recurrent'' (STORE) models, are described. They encode the invari ant temporal order of sequential events in short-term memory (STM) in a way that mimics cognitive data about working memory, including prima cy, recency, and bowed order and error gradients. As new items are pre sented, the pattern of previously stored items remains invariant in th e sense that relative activations remain constant through time. This i nvariant temporal order code enables all possible groupings of sequent ial events to be stably learned and remembered in real time, even as n ew events perturb the system. Such competence is needed to design self -organizing temporal recognition and planning systems in which any sub sequence of events may need to be categorized in order to control and predict future behavior or external events. STORE models show how arbi trary event sequences may be invariantly stored, including repeated ev ents. A preprocessor interacts with the working memory to represent ev ent repeats in spatially separate locations. It is shown why at least two processing levels are needed to invariantly store events presented with variable durations and interstimulus intervals. It is also shown how network parameters control the type and shape of primacy, recency , or bowed temporal order gradients that will be stored.