No abstract

In three experiments, analyses of individual-subject data show that temporal point of interruption for a practice recitation of a serial list affected neither the ultimate amount of time needed to master that list nor the amount of time needed for remastery 24 h later. Learning and relearning were by the unpaced whole-presentation method, with scheduled test interruptions at different stages of original learning.Tradition has it that for purposes of research, the learning procedure for verbal materials is a succession of study-test cycles. Some such cycles are batch defined in that the learner studies a list of material and then is tested on it, studies again, is tested again, and so on. Other such cycles are item defmed in that on seeing or hearing one item, such as a stimulus in paired associate learning or the nth item in serial learning, the learner is tested by having to anticipate a certain other item, such as the corresponding response or the next item, followed immediately by a study phase. wherein the learner is presented with the appropriate other item for confirmational or instructional purposes. Recurrences of these item cycles are generally separated by intervening cycles on other items.Tradition aside, one might wonder at the practice of imposing so many test trials in such a regular and unrelenting manner. To so proceed is not simulative of memorization in nonlaboratory situations, nor is it necessary to keeping track of how well the laboratory learner is doing. More importantly, there arises the question of how test operations contribute to, or detract from, the learning process. The study and test phases of a cycle must certainly interact in complex ways, to the end that generalizations about learning beyond the several standard laboratory paradigms are of unknown value.The literature on test effects in the area of verbal learning supports several conclusions. With respect to paired associate learning, it is clear that a good schedule of study and test events should include tests early in the sequence (LaPorte & Voss, 1974), but that multiple testing in the form of successive retesting is a waste of time (Bregman & Wiener, 1970;Izawa, 1967 Izawa, , 1970. A likely interpretation of this is that the results of a test are the basis for selective study of the more difficult pairs, with successive retests offering no further