Degree of learning and the measurement of forgetting

Abstract

This paper was directed toward problems involved in the measurement of forgetting uncontaminated by differences in degree of learning. More particularly, it was concerned with these measurements when some variable, such as a characteristic of the task, is being manipulated and when such a variable produces differences in rate of learning. If we are to assess properly the influences of these variables on retention, degree of learning must be equated, since degree of learning and retention are directly related. The two basic situations considered were those in which a constant number of learning trials was given and those in which learning was carried to a specified criterion of performance. The single-entry technique is appropriate only when a constant number of learning trials is used. When a criterion of performance is set for learning another procedure (multiple-entry projection) may be used. Although the mean predictions are fairly accurate by this method, predictions for individual Ss are not. In most studies of retention it seems most efficient to use a constant-trials procedure for learning. Finally, it was pointed out that some studies of short-term retention of single items have probably confounded effects of degree of learning on retention with the effects of variables producing differences in rate of learning the items.