On nominal and functional serial position curves: Implications for short-term memory models?

When the presentation rate is varied in single-trial free-recall experiments, early and middle items of a list are recalled better with a slower rate, with no differences in recall for the last few items. This finding has been interpreted as evidence in support of 2-store theories of memory: Presentation rate affects the long- but not the short-term store. However, if serial position curves are plotted not as a function of the trial on which each item of a list was presented but of the trial on which it was last rehearsed, rate effects are reversed: The superiority of the slow presentation rate manifests itself primarily at the end of the list. A 2-store Markov model that accounts for this reversal is described, showing that conventional 2-store models of memory are in no way contradicted by these results. (10 ref)