Two rhesus monkeys were tested in 6-and Iü-item list memory tasks for performance changes as a function ofthe exposure duration ofthe list stimuli and the interstimulus interval (lSI) between successive list stimuli. Accuracy increased with longer item exposure duration and tended to decrease with longer ISI duration. Humans, by contrast, typically show increases in accuracy with ISI, a result taken as evidence of rehearsal. The decrease in accuracy for monkeys suggests that they were not using rehearsal processes in these list memory experiments. Further tests in which choice accuracy with predictahle ISls was compared with choice accuracy with unpredictahle ISls also yielded no evidence of rehearsal by the monkeys. This apparent absence of rehearsal mechanisms in monkeys, in situations also shown to support human rehearsal, is discussed as a potential difference in the visual working memory processes of the two species.Humans regularly engage in rehearsal. Indeed it is difficult to imagine a day going by without our rehearsing something, whether it be telephone numbers, addresses, shopping lists, or names of new acquaintances. Although rehearsal often involves language, this apparently is not a necessary requirement; rehearsal can occur with pictorial stimuli in humans (Watkins, Peynircioglu, & Brems, 1984). In this article, rehearsal will be considered as a subject-controlled, memory-based repetition of to-be-remembered material (cf. Johnson, 1980;Watkins & Peynircioglu, 1982). This relatively broad view of rehearsal includes both the overt and covert rehearsal of either visual or verbal materials. When considered thus, rehearsal does not depend on language, and this opens up the possibility that other animals may also have rehearsal-like processes. Procedures involving our human capabilities for language and speech have been valuable and frequently used in the study of rehearsal (see, e.g., Dark & Loftus, 1976;Rundus, 1971;Rundus & Atkinson, 1970; Modigliani & Hedges, 1987), but obviously they cannot be used in examining this question in other animals.During the past decade, the interstirnulus-interval (181) or blank-time procedure has been developed as a promising means of studying animal rehearsal. Rehearsal in humans is identified in this procedure through the increased accuracy that is observed when the ISls between the iterns of a to-be-remembered list or sequence are lengthened.
153For example, Intraub (1980) presented subjects with sequences of 16 pictures at different ISls and then tested item recognition. Subjects presented with pictures for l l O msec and no ISI performed at only 59 % accuracy, but as the ISI was lengthened to 385 msec, 620 msec, 1,390 msec, and finally 4,890 msec, performance increased to 71 %, 78 %, 91 %, and 92 %, respectively. 1 Performance in the 4,89O-msec ISI condition (viewing time, 110 msec) was as accurate as it was when each item was visible for 5 sec. This increase in accuracy with langer ISls will be referred to as the ISI effect in the present article.The ISI effect...