JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 7,617-626 (1968) The Measure of Interference in Primary Memory1 NANCY C. WAUCH Harvard Medical School, Boston, Massachusetts 02114 AND DONALD A. NORMAN University of California, San Diego, California 92037 The Ss attempted to remember a minimally rehearsed serial association after attending to a sequence in which items were either ordered at random or repeated according to certain rules. The results indicate that, although a new and unpredictable item may displace an earlier one from primary memory, a recently presented and redundant one does not. Whether a minimally rehearsed pair of words or digits can still be remembered at the end of an unfamiliar sequence seems to depend not so much on how long ago they were presented as on how far back in the sequence they occurred. Imagine that an S has just listened to a list of randomly ordered verbal items, attending to them one at a time. Item i + 1 from the end of the list is now presented again, and the S’s task is to recall the one that followed it. The probability R(i) that he can do so has been found to decrease rapidly with i: it is close to unity when i = 1 and approaches chance when i> 8 (Waugh and Norman, 1965). This function, moreover, is independent of the rate at which the list was presented and therefore of the total time elapsing between the presentation and attempted recall of a given item (Norman, 1966). Although increas- ing the time available for the rehearsal of the individual items in a list may increase the 1 This work was supported by Research Grants No. MH-08119-03, MH-08083-03, and NB-07454-01 from the National Institutes of Health, United States Public Health Service, to Harvard Medical School, to Harvard University, Center for Cognitive Studies, and to the University of California, San Diego, respectively; and by Research Grant No. SD-187 from the United States Department of Defense, Advanced Research Projects Agency, to Harvard University, Center for Cognitive Studies. likelihood that any one of them will be later recognized or freely recalled (Waugh, 1967) it seems to have little effect on serial retention when items are not rehearsed in groups. These facts have suggested that short-term memory is based on two independent storage systems, retention in the one depending on an item’s recency and in the other, on the extent to which it has been rehearsed. In the dis- cussion that follows, it will in fact be assumed that every item attended to is by definition placed in primary memory (PM), a store of limited capacity from which it can immediately be retrieved. Rehearsing a single item in PM increases the likelihood that it is transferred into a more stable store; if two successive items in a list are to be retained in their proper order over any relatively long interval, however, they must be rehearsed as a pair during the time in which they are both in PM. The rapid forgetting of an unrehearsed association can accordingly be said to reflect the retroactive interference exerted on it by immediately ensuing verbal events. The purpose of the present study was to determine the appropriate measure of this interference. Six possibilities were evaluated in Exps. I, II, and III. Two of them will now be described. Hypothesis A. Any item just presented can displace an older one from PM, so that the 617