Telling a computer how to evaluate multidimensional situations

Abstract
If a person could tell a computer exactly how he would evaluate every alternative that might arise, the machine could decide between any two alternatives as the person himself would. A fairly realistic experiment was done to test the feasibility of “interpolation between corners” as a psychophysical method for telling a machine how to compute the worths of multidimensional alternatives. The results were satisfactory. A statistic called fractional disagreement is proposed as the proper measure of the machine's success in mimicking the man's decisions. The concept of conflict between the dimensions of two alternatives is explained. The size of the conflict and the size of the difference between the worths that the machine ascribed to the alternatives were found to help to predict the correctness of the machine's decision. Conflict and difference in computed worth should, therefore, be useful in defining regions where the machine should give the decision back to the man and tell him to make the choice.

This publication has 0 references indexed in Scilit: