A neural network model of learning mathematical equivalence

Kevin W. MickeyStanford University, Stanford, CA, USA
James L. McClellandStanford University, Stanford, CA, USA

Abstract

The typical structure of equations influences how we learn the meaning of the equal sign. Previous studies have shown that as students gain experience with addition problems, they actually perform worse on certain problems, before eventually improving. We seek to explain this trajectory with gradual implicit learning, without explicit representation of strategies or principles. Our parallel distributed processing model is nevertheless able to simulate several phenomena observed in how children learn mathematical equivalence: not only how successful performance develops, but also what strategies are used and how equations are encoded.

Files

A neural network model of learning mathematical equivalence (864 KB)



Back to Table of Contents