Rule acquisition events in the discovery of problem-solving strategies

Abstract

3) This subject took o scientific approach to strategy discovery, even planning and conducting small experiments.' By "cognitive architecture," VANLEHN people solved novel problems using means-ends analysis, forward search, abstraction planning, and other weak methods.It was soon found that experts often used distinctly different problem solving strategies than novices, so strategy acquisition became an important topic.In the late 197Os, the first computational models of strategy acquisition began to appear, and the field of machine learning was born.Nowadays, many models exist and more are invented each year (130 papers were presented at the 1989 machine-learning workshop).However, strategy acquisition is not solved in the way that problem solving is, because the evidence connecting models to human data is weak.Most of the existing work reduces human data to a sequence of strategies, then demonstrates that the model can make these transitions too (Anderson,