RATIONAL ANALYSIS OF LEARNING AND COGNITION

点击翻译按钮获取中文摘要

The principle methodology of this research has been to develop computer simulations of the representations and mechanisms of human cognition. My research has taken a new direction which is characterized as being focused on how the mechanisms of the human mind are adapted to structure of the environment. I am currently pursuing a thesis called "The principle of rationality" which proposed that the mechanisms of the mind are optimized to the environment. A book is being completed (Anderson, in press) that develops this thesis successfully for relatively simple areas of human cognition such as basic memory and simple concept formation. The purpose of this award is to allow sufficient time to acquire the training required to effectively pursue this analysis for the more complex areas of problem solving and language. The first three years will be spent acquiring a sophisticated background in Bayesian decision theory. The last three years (overlapping in one year) will be spent retraining myself as to the current state of linguistics. In parallel with this, a rational theory of problem solving will be developed. This theory will combine an analysis of the probability of causal extrapolations involved in problem solving with the cost of various paths of problem solving. It will attempt to explain both the logic and control of problem solving as an optimization that responds to the combined consideration of probability and cost.