In this chapter we explore modeling of arithmetic fact retrieval within a connectionist or “neural net” framework. We first sketch briefly some of the major empirical phenomena emerging from research on arithmetic facts, and review previous attempts to develop distributed connectionist models of arithmetic fact retrieval. We then describe a new model, MATHNET, and present preliminary findings concerning the model's ability to simulate several phenomena of human fact retrieval. We conclude by discussing limitations of our initial modeling, and directions to be pursued in subsequent work.
ASJC Scopus subject areas