This is the third volume in the Vancouver Studies in Cognitive Science Series. It is based on a conference that was held in 1990, which was sponsored by the Cognitive Science Program and Linguistics Department of Simon Fraser University. Over the last decade, there has emerged a paradigm of cognitive modeling that has been hailed by many researchers as a radically new and promising approach to cognitive science. This new paradigm has come to be known by a number of names, including connectionism, neural networks, and parallel distributed processing, (or PDP). This method of computation attempts to model the neural processes that are thought to underlie cognitive functions in human beings. Unlike the digital computation methods used by AI researchers, connectionist models claim to approximate the kind of spontaneous, creative and somewhat unpredicatable behavior of human agents. However, over the last few years, a heated controversy has arisen over the extent to which connectionist models are able to provide successful explanations for higher cognitive processes. A central theme of this book reviews the adequacy of recent attempts to implement higher cognitive processes in connectionist networks.