Journal of Experimental Psychology: Learning, Memory, and Cognition 1994, Vol. 20, No. 3, 608-620 Copyright 1994 by the American Psychological Association, Inc. 0278-7393/94/$3.00 Theoretical Implications of the Mnemonic Benefits of Perceptual Interference Elliot Hirshman, Dawn Trembath, and Neil Mulligan How should interfering with the perception of items during study affect memory for those items? Recent research by Nairne (1988) and Hirshman and Mulligan (1991) has demonstrated that backward pattern masking during study enhances later memory. This article examines whether traditional explanations of encoding benefits, including rehearsal, visual distinctiveness, and encoding effort, can account for this result. No evidence was found for any of these hypotheses. An explanation that focuses on the compensatory processing of higher level perceptual representa- tions is proposed. This explanation provides a plausible explanation of the results of 7 experiments. The article concludes with a discussion of the implications of the explanation for perceptual priming and other manipulations of perceptual interference. How does interfering with the perception of items during study affect memory for those items? Nairne (1988) and Hirshman and Mulligan (1991) recently investigated this question using visual masking. These investigators examined memory performance on items presented in two study condi- tions. In the control or intact condition items were presented for 2.5 s. In the experimental or interference condition items were presented briefly (i.e., 100 ms) and then followed with a backward pattern mask (i.e., a row of Xs) presented for 2.4 s. The mask obscures the visual features of the study item, interfering with its perception. Surprisingly, these studies have demonstrated that interfer- ing with perception using visual masking can actually improve memory performance. Nairne (1988) reported an advantage for the interference condition in recognition memory. Hirshman and Mulligan (1991) replicated Nairne's recognition memory results and demonstrated that this effect also occurs in free recall. 1 These results seem surprising from several perspectives. First, many classical (Atkinson & Shiffrin, 1968; Waugh & Norman, 1965) and contemporary (Baddeley, 1981; Raaijmak- ers & Shiffrin, 1981; Sperling, 1986) models predict that interfering with perception during study should impair memory performance because encoding is assumed to occur in a limited-capacity channel in these models. (See Anderson & Craik, 1974; Baddeley & Hitch, 1974; Johnston, Wagstaff, & Griffith, 1972; Loftus, Hanna, & Lester, 1988; and Murdock, Elliott Hirshman, Dawn Trembath, and Neil Mulligan, Department of Psychology, University of North Carolina at Chapel Hill. This research was supported by a University Research Council grant from the University of North Carolina to Elliott Hirshman and by Research Grant F49620-92-J-0119 from the U.S. Air Force Office of Scientific Research to Joan Gay Snodgrass. We thank Henry Roediger III, Peter Graf, and an anonymous re- viewer for helpful comments on an earlier version of this article. We greatly appreciate Mary Elizabeth Broadfoot's assistance in testing subjects. Correspondence concerning this article should be addressed to Elliott Hirshman, Department of Psychology, CB #3270, 206 Davie Hall, University of North Carolina, Chapel Hill, North Carolina 27599- 3270. Electronic mail may be sent to elliot@memlab.psych.unc.edu. 1965, for evidence supporting this assumption.) In this limited- capacity channel, any stimulus (e.g., a visual mask) that interferes with perception will use encoding resources that would have been devoted to the target item, thus impairing memory for the target item. Second, the results violate intuitions about the relationship between presentation duration and memory performance. Although we would normally expect that increasing presenta- tion duration would lead to better memory performance (Roberts, 1972), the preceding results demonstrate a negative relation between presentation duration and memory perfor- mance. Memory performance on items presented for 100 ms is better than memory performance on items presented for 2.5 s. Although this mnemonic benefit of visual masking is theoreti- cally puzzling, several other manipulations that alter the visual characteristics of study items also enhance memory, suggesting that the phenomenon may be general. For example, the deletion of letters from to-be-identified study items produces a memory enhancement known as the generation effect (Slamecka & Graf, 1978). Similarly, inverting to-be-identified study items also enhances recognition memory for these items (Kolers, 1973,1975). The similarity among these experimental manipu- lations is intriguing, and an important long-range goal is to characterize the similarities and differences among the pro- cesses contributing to these three effects. (See Graf, 1982, for the hypothesis that similar mechanisms produce the memorial effects of generation and text inversion.) We begin here by presenting and testing alternative explana- tions of the mnemonic benefit of visual masking. The recurrent 1 It is important to note that Nairne (1988) did not find a mnemonic benefit of perceptual interference in his studies of free recall. Hirshman and Mulligan (1991) argued that this null effect occurred because of the large number of identification errors that occurred on interference items during study in Nairne's (1988) experiments. Hirshman and Mulligan (1991) supported this claim by demonstrating (in Experiment 5) that increasing these errors produced a replication of Nairne's (1988) findings. These arguments emphasize that it is critical to minimize the errors in the interference condition at study to produce an appropriate measure of the mnemonic effect of perceptual interference. 608 This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.