Teaching a topic in cybernetics with APL

Abstract

This paper describes a curriculum unit in which APL is used to introduce models of neural networks. It begins with a simple model for transmitting impulses among a vector of logical neurons; then a more sophisticated model is developed with thresholds, decay, and inhibition in a matrix of neurons; then a general model is offered for higher order arrays of neurons, with dynamic input and automated impulse propagation; and, finally, the APL functions are embodied in an easy-to-use simulation package. Possible extensions of these models are proposed as student programming projects; some other related topics are identified as potentially suitable for teaching with APL; and implications for computer science education are mentioned.