A presentation of an information processing theory of rote serial learning sufficient to predict (qualitatively and quantitatively) the shape of the serial error curve. In addition, other rote-learning phenomena are explained. The theory postulates a serial information processing mechanism that learns (on the average) one item from a serial list every k seconds, has a very small immediate memory span, and uses an anchor-point processing strategy for organizing its learning effort over time. Two ways described to make predictions from the postulates are by a computer programmed to process information and by a simple mathematical model.
