View article

In a previous issue, we (Wheeler & Roediger, PS 5 (4), 1992, pp. 240-245) reported experiments showing that for recently learned material, repeated testing revealed overall improvements in recall when short intervals occurred between the tests but forgetting occurred when the intervals between tests were lengthened to a week. This outcome solves the puzzle posed by the conflicting results of Ballard (1913) and Bartlett (1932), among many other investigators, who had reported differing effects of repeated testing.

Bahrick and Hall (this issue) argue that our conclusion is likely correct for recently learned material (episodic memory situations, in Tulving's, 1972, terms), but that for" a stable knowledge system acquired much earlier"(semantic memory), hypermnesia can be found with long intervals between tests. They argue that in semantic memory situations, little intertest forgetting occurs, so any recoveries of knowledge between tests will lead to overall (net) improvement. They cite data from Bahrick and Hall (1991) and Herrmann, Buschke, and Gall (1987) in support of their argument. Bahrick and Hall's observations (this issue) are interesting ones and should lead to further research on these matters. Previous research (Roediger, Payne, Gillespie, & Lean, 1982, Experiment 3) had shown hypermnesia for general knowledge material (repeated retrieval of category instances) with short intervals (several minutes) between tests, but the intervals used by Bahrick and Hall (1991) were a month. Squire, Haist, and Shimamura (1989) have also reported hypermnesia, with both amnesic patients and control subjects, in repeated cued-recall tests of public …