148 Animal Learning & Behavior 1983,11 (1),148-150 On the nature 01 the primacy effect In memory procel8lng: A reply to Galfan ANTHONY A. WRIGHT and HECTOR C. SANTIAGO University of Texas, Houston, Texas and STEPHEN F. SANDS University of Texas, EI Paso, Texas The primacy effect, remembering the first items of a list better than the middle ones, "maintains its reputation as the Chinese puzzle of verbal learning" (Tulving & Madigan, 1970, p.4'4). To this day, there is little concrete evidence as to the cause, al- though many theories have been proposed to account for it: interference theory by way of a lack of pro- active interference (Foucault, 1928; Postman & Phillips, 1965), endpoint distinctiveness (Bower, 1971; Ebenholtz, 1972; Murdock, 1960), and rehearsal (Atkinson & Shiffrin, 1968; Waugh & Norman 1965), just to name a few. Recently, we (Sands &. Wright, 1980a) showed the first primacy effect for an animal, a rhesus monkey working in a serial-probe recognition (SPR) task. At about the same time, Roberts and Kraemer (1981), using a different species of monkey in a slightly different task, demonstrated a primacy effect too. Thus, we, along with Roberts and Kraemer, felt that the primacy effect in memory processing is characteristic (or at least possible) in primates generally. More recently, we have demon- strated it with pigeons (unpublished as yet), which indicates even wider generality within the animal kingdom. Gaffan (1983) has made a somewhat different inter- pretation. He suggested that our monkey primacy ef- fects resulted from a procedural artifact of requiring the monkeys to respond to initiate the list items. Implicit in his suggestion is that our primacy effects are not produced by the same memory processes that produce the primacy effect in human memory experi- ments. implied is that the findings of primacy effects m our monkeys can be dismissed as incon- sequential because such effects result from an unin- teresting procedural artifact. The authors gratefully acknowledge the helpful comments made by J. Watkins on an earlier draft of this manuscript. A. A. Wnght and H. C. Santiago's mailing address is: Univer- sity of Texas Health Science Center, Graduate School of Biomedi- Sciences, Houston, Texas 77025. S. F. Sand's mailing address IS: Department of Psychology, University of Texas at El Paso, El Paso, Texas 79968. Gaffan argued that the initiating response in our experiment, pushing down on the three-position lever, draws attention to the first list item makes it more distinctive, and consequently makes it easier to remember. Similar arguments were directed toward Roberts and Kraemer's study. He cites three experi- ments of his (Gaffan, 1977, 1979; Gaffan & Weiskrantz, 1980) in which he obtained serial-position functions. None revealed primacy effects, and all of them required the monkeys to initiate every item. According to his argument, all items in his experi- ment were equally attended to, equally distinctive, and equally well remembered. The causes of Gaffan's failure to find a primacy effect may have to do with his choice of a procedure rather than with any artifact in our procedures. Indeed, we would not expect all memory tests to reveal primacy effects with monkeys; they do not with human subjects. A failure to find a primacy effect does not prove that an ani- mal is not capable of producing a primacy effect; perhaps the proper procedure for revealing a pri- macy effect just has not yet been discovered. In two of the three experiments (Gaffan, 1979; Gaffan & Weiskrantz, 1980), there is no a priori reason to expect that his unusual and somewhat complicated procedure should reveal a primacy effect. In the third experiment (Gaffan, 1977), he employed a SPR pro- cedure. We have since replicated this experiment (Sands & Wright, 1980b, Experiment 3) and also found no primacy effect. It is contrary to his hy- pothesis that our experiment did not reveal a primacy effect because our monkey initiated the list items, a procedure which, according to Gaffan's response- dependent attentional hypothesis, should have pro- duced one. In both experiments, only three-item list lengths were used. Furthermore, the items in both experiments were drawn from six item pools, which tended to interfere with performance because the items were repeated many times. Our experiment demonstrated the comparatively poor performance with the high (proactive) interference when items were repeated and the much better performance when the items were not repeated. We routinely train our monkeys without interference by using a large item pool (3,000 or more items). Following adequate training, they perform almost as accurately as human subjects in the task. With our monkeys, we have demonstrated 83"'0 accuracy with 20-item list lengths, 86"'0 with to-item list lengths, and, more recently, even better than 90"'0 with to-item list lengths. We have never failed in training a monkey on the SPR task. We have successfully trained and tested five monkeys on this task. All five have shown robust and reliable primacy effects. There are advantages to using a memory task that Copyright 1983 Psychonomic Society, Inc.