1. The Contribution of Perceptual Mechanisms to the Spacing Effect Jason Arndt & Julie Dumas Middlebury College Abstract Recent explanations of the spacing effect posit a central role for semantic priming mechanisms (e.g., Challis, 1993). Russo and his colleagues (Russo, et al., 1998; 2002) have further provided evidence consistent with the contribution of perceptual priming mechanisms to the spacing effect for items not conducive to semantic analysis (e.g., nonwords), but no such contribution for items conducive to semantic analysis (e.g., words). We further test this empirical regularity by presenting repetitions of both spaced and massed words in different fonts. The results of two experiments demonstrate that the spacing effect for words was eliminated when the perceptual format changed between item repetitions. Theoretical Background  Priming theories of the spacing effect posit that the spacing effect will be observed in both intentional and incidental memory tasks  Encoding variability theories propose that maximizing feature difference across repetitions should enhance memory  Decision-based models propose that false recognition reduction occurs because participants become more conservative  Due to the information encoded at study, or the characteristics of the memory test (e.g., Schacter, Israel, & Racine, 1999; Miller & Wolford, 1999)  Memory-based models propose that false recognition reduction occurs because studied items become more discriminable from unstudied items (e.g., Wickens & Hirshman, 2000; Wixted & Stretch, 2000) Within- vs. Between-Subjects Designs  Decision-based and memory-based theories of false recognition can be evaluated by comparing the effects of an independent variable across within- and between-subjects designs (Schacter, et al., 1999)  Participants are free to adopt different decision strategies across conditions in a between-subjects (or lists) design  Participants maintain a single decision strategy across conditions in within-subjects (or lists) designs (Stretch & Wixted, 1998) Decision-Based Models  Decision-based models predict that false recognition differences across conditions can exist when an independent variable is manipulated between subjects (or lists), because participants are free to adopt different decision strategies in different conditions  However, when an independent variable is manipulated within subjects (and lists), decision-based models predict that false recognition differences across conditions will be equated, given the assumption that participants adopt a single decision strategy for all test items (Stretch & Wixted, 1998)  Therefore, if false recognition differs across conditions in a between-subjects design, but those differences are not evident when the same conditions are manipulated within subjects, decision-based models are supported Memory-Based Models  Memory-based models propose that false recognition is caused by the properties of encoded memory representations  On the presumption that memory-based processes do not differ across between- and within-subjects designs, memory-based models expect that false recognition differences should replicate across the two designs Experiments  Materials: 48 sets of twelve items (referred to as themes henceforth) from Nelson, McEvoy, and Schreiber (1998)  All items in a given theme produced a single item (the lure item) in free association with some nonzero probability  Theme items: rose, stem, blossom, lily, vase, dandelion, orchid, tulip, petals, daisy, garden, bloom  Lure item: flower  Each study list composed of several themes presented in blocked format  Usefulness of visual information at study was manipulated by presenting study items in unusual-looking fonts (see Figure 1; Reder, Donavos, & Erickson, 2002).  Correlated condition: all fonts in a theme presented in the same font (each font presented twelve times)  Uncorrelated condition: font repeated twelve times in a study list, but font randomly assigned to words in a theme  Unique condition: a different font used to present each word in a theme  Test lists composed of studied items, lures associated with studied items, and a comparable set of unstudied theme items, as well as lure items related to unstudied themes.  Studied items presented in same font utilized at study  Lure items presented in a font utilized to present at least one associated theme item  Unstudied items presented in a font shown at study  Lure items associated with unstudied themes presented in a font utilized to present associated theme items  Participants asked to provide judgments of recognition memory phenomenology (i.e., remember-know responses)  Design: Font condition (correlated vs. uncorrelated vs. unique) varied between subjects in Experiment 1 and within subjects in Experiment 2. Predictions  Decision-based model  Font condition should affect false recognition differently across conditions in Experiment 1 (between subjects)  Font condition should not affect false recognition differently across conditions in Experiment 2 (within subjects)  Memory-based model  Effects of font condition should replicate across between- and within-subjects designs Dependent Measures Accurate Recognition  old-new d’  Computed on hits to studied items vs. false alarms to unstudied theme items False Recognition  lure d’  Computed on false alarms to lure items related to studied themes vs. false alarms to lure items related to unstudied theme items  High values indicate high levels of false recognition  old vs. lure d’  Computed on hits to studied items vs. false alarms to lure items related to studied themes  Low values indicate high levels of false recognition Results  Font condition did not strongly affect accurate recognition  No effect of font condition in Experiment 1  Unique condition showed superior performance relative to uncorrelated condition in Experiment 2  Font condition strongly affected false recognition  Highest levels of false recognition found in correlated condition, lowest levels of font condition found in unique condition  Font condition affected false recognition similarly in Experiments 1 and 2  Remember responses generally reflected patterns of d’ presented below Conclusions  Replication of false recognition patterns across between- and within-subjects design is inconsistent with decision-based accounts of false recognition reduction (e.g., Miller & Wolford, 1999; Schacter, et al., 1999)  Replication of false recognition patterns across between- and within-subjects design is consistent with memory-based models of false recognition reduction (e.g., Wickens & Hirshman, 2000; Wixted & Stretch, 2000)  Results of this study demonstrate that the study of visual information reduces semantic false recognition, replicating and extending prior results (e.g., Smith & Hunt, 1998; Schacter, et al., 1999) References Miller, M.B., & Wolford, G.L. (1999). Theoretical Commentary: the role of criterion shift in false memory. Psychological Review, 106, 398-405. Nelson, D.L., McEvoy, C.L., & Schreiber, T.A. (1998). The University of South Florida word association, rhyme, and word fragment norms. http://w3.usf.edu/FreeAssociation/ Reder, L.M., Donavos, D., & Erickson, M.A. (2002). Perceptual match effects in direct tests of memory: the role of contextual fan. Memory & Cognition, 30, 312-323. Schacter, D.L., Israel, L., & Racine, C. (1999). Suppressing false recognition in younger and older adults: The distinctiveness heuristic. Journal of Memory and Language, 40, 1-24. Smith, R., & Hunt, R.R. (1998). Presentation modality affects false recognition. Psychonomic Bulletin & Review, 5, 710-715. Stretch, V., & Wixted, J.T. (1998). On the difference between strength-based and frequency- based mirror effects in recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 681-690. Wickens, T.D., & Hirshman, E. (2000). False memories and statistical decision theory: Comment on Miller and Wolford (1999) and Roediger and McDermott (1999). Psychological Review, 107, 377-383. Wixted, J.T., & Stretch, V. (2000). The case against a criterion-shift account of false memory. Psychological Review, 107, 368-376. Acknowledgements This work was supported by grants 5-T32-MH19983 and 2- R01-MH52808 from the National Institute of Mental Health. Please contact Jason Arndt (jarndt@middlebury.edu) for a pre-print of this article (in press, Journal of Memory and Language) Figure 1: Examples of Fonts ROSE STEM BLOSSOM LILY VASE DANDELION ORCHID TULIP PETALS DAISY GARDEN BLOOM Old vs. Lure d’ Experiment 1 Experiment 2 Old - New d’ and Lure d’ Experiment 1Experiment 2