Recall latency following pure- and mixed-strength lists: A direct test of the relative strength model of free recall.

Abstract

Strong items (e.g., (hose studied for a longer period of time) are not recalled faster than weak items in pure-strength lists. Although counterintuitive, this result is consistent with a relative strength model of free recall. In mixed-strength lists, by contrast, the relative strength model requires that strong items be recalled significantly faster than weak items. A considerable body of recent research on this issue suggests that, if anything, the opposite may be true. Four experiments reported here measured free-recall latency following pure- and mixed-strength lists. Recall latency for strong items was consistently shorter than that for weak items, but in mixed lists only. Moreover, as uniquely predicted by a relative strength model, in mixed lists, strong items were recalled more quickly than items from a pure-strength list of the same size, and weak items were recalled more slowly by a corresponding amount. The idea that stronger memories come to mind more quickly and easily than weaker ones has been around for a long time. Evidence consistent with this idea was reported by Marbe (see Bousfield, Cohen, & Silva, 1956) around the turn of the century and by Bousfield (Bousfield & Barclay, 1950; Bousfield et al, 1956; Bousfield, Whitmarsh, & Esterson, 1958) in the 1950s. As intuitively obvious as this result may seem, it is clearly incorrect under certain conditions (cf. MacLeod & Nelson, 1984). For example, Rohrer and Wixted (1994) showed that strengthening words on a list by allowing extra study time increased the number of items recalled but did not affect the average time to recall. They also showed that under the right conditions (namely, a long list of strong items vs. a short list of weak items), strong items were recalled more slowly than weak items. When are intuitions about the relationship between memory strength and recall latency correct and when are they incorrect? Wixted and Rohrer (1994) argued that relative strength models of free recall effectively guide thinking about this issue.