generalized of fact. To the contrary, the short-interstimulus interval curves from the humans show lower performance only at medial items, while Oscar's curves show a lower performance only at terminal items.
An unexpected result of this study was the failure of the adolescent humans and monkey to show any facilitation with additional stimulus off- time. This finding suggests the interesting possibility that rehearsal in visual memory is an ability that may be acquired at later stages of development not only for humans but also for rhesus monkeys. The human developmental literature has shown that active rehearsal strategies do not appear until the later elementary school years ( Appelet al., 1972; Hagen, Jongeward, & Kail, 1975). A similar conclusion would be consistent with the results of this experiment. Further support for the developmental hypothesis can be found in the striking similarities between the adolescent humans and Felix in absolute levels of accuracy, the effects of list length and interstimulus interval. Both showed lower levels of accuracy than their adult counterparts, both showed a progressive loss in accuracy with increasing list length, and both were unaffected by interstimulus interval.
If we accept the notion that Oscar is maintaining list items in a manner analogous to rehearsal, then this result has important implications for the nature of rehearsal in visual memory. Some investigators have shown that the ability to rehearse and maintain pictures is enhanced if the pictures can be described by a convenient verbal label ( Lutz & Scheirer, 1974; Tobachnik & Brotsky, 1976). Several other investigators, however, have shown that verbal descriptions are not necessarily an essential requirement to the rehearsal of pictures ( Phillips & Christie, 1977; Graefe & Watkins, 1980; Intraub, 1979; Tversky & Sherman, 1975). The results from Oscar provide evidence that rehearsal of pictorial materials does not depend exclusively on verbal encoding. Human subjects may employ verbal codes to mediate pictorial recognition. Indeed, one of our human subjects employed a verbal labeling strategy for the pictures in the later stages of testing and overtly rehearsed (verbalized) during the interstimulus interval and probe delay. The other adult human subject also reported the adoption of verbal labels as testing progressed. The advantage to a verbal encoding strategy, however, is not clear since performance of the human subjects remained relatively constant across all experimental sessions even though both reported a shift from visualization to verbalization strategies.
Serial-probe recognition performance of briefly (80 msec) presented Pictures was facilitated with a 1000 msec interstimulus interval relative to a 80 msec interstimulus interval in two human subjects and in a single highly-trained rhesus monkey (Oscar). These results demonstrated the benefits of additional time to process or rehearse successively presented pictures. Serial position curves for the adult humans showed an increase in the pre-recency portion of the curve with the 1000 msec interstimulus interval. Serial position curves for Oscar showed an increase in the recency portion of the curve with additional stimulus off time. In addition, serial position curves for Oscar, who in previous studies had shown