ABSTRACT
This chapter describes an approach to modelling human memory that attempts to link current mathematical and connectionist approaches. It reviews some of the existing mathematical models of memory, focusing particularly on those that use convolution as the basic associative mechanism, and correlation as the basic retrieval operation. Current mathematical models provide a good account of a wide range of empirical data concerning adult human memory performance. The chapter explains Developmental Associative Recall Network (DARNET) is a connectionist-like architecture that "learns to learn". It therefore combines the ability to perform one-shot learning with a developmental aspect - the one-shot learning ability is not present from the start, but gradually develops. Thus the new architecture allows examination of the development of associative learning in a way that has not previously been possible. The approach describes intended to develop a link between connectionist learning algorithms and convolution-based models of memory, by showing that a connectionist-style learning algorithm.
