The length-difficulty law of serial learning: An empirical function and a theoretical derivation

Abstract

List length, item meaningfulness, and presentation rate were varied in a factorially designed serial learning experiment. In all conditions, the function relating length (r) with trials to criterion (t) was of the form t=a+b logr. Mathematical analysis revealed that such a function is implied by the classic negatively accelerated exponential mean learning curve. A theoretical expression for total errors to criterion as a linear function of list length was also derived and confirmed.