The analysis of economic teaching strategies for a simple learning model

Abstract

This paper formulates in economic terms the decision problem for teaching a list of paired associate items. The analysis assumes that learning behavior for each item is described by the one-element model with parameters that can depend on both the student and the item. A dynamic programming formulation for the decision problem is developed that accounts for the cost of instruction as well as the cost of terminating the instruction in the unconditioned state. This formulation considers three alternatives at each decision: simply present the item with no elicited response, present the item and observe the students' response, and terminate instruction for the item. The paper proves that the optimum policy for a multi-item list can be calculated by considering each item individually and then aggregating the individual results into a total policy. Secondly, an algorithm for calculating the optimum policy for each item is derived. The algorithm is simple enough so that the optimum policy can be calculated by hand. Two examples illustrate the application of the algorithm.