Journal of Experimental Psychology: Copyright 1985 by the American Psychological Association, Inc. Learning. Memory, and Cognition 0278-7393/85/$00.75 1985. Vol. II. No. 3. 501-518 Implicit and Explicit Memory for New Associations in Normal and Amnesic Subjects Peter Graf and Daniel L. Schacter University of Toronto, Toronto, Ontario, Canada Two experiments examined whether repetition priming effects on a word completion task are influenced by new associations between unrelated word pairs that were established during a single study trial. On the word completion task, subjects were presented with the initial three letters of the response words from the study list pairs and they completed these fragments with the first words that came to mind. The fragments were shown either with the paired words from the study list (same context) or with other words (different context). Both experiments showed a larger priming effect in the same-context condition than in the different-context condition, but only with a study task that required elaborative processing of the word pairs. This effect was observed with college students and amnesic patients, suggesting that word completion performance is mediated by implicit memory for new associations that is independent of explicit recollection. Remembering is commonly assumed to fragments of recently presented words and of involve the conscious or explicit retrieval of new words, subjects succeed more often with information about particular experiences. presented words than with new words. This This assumption derives largely from tradi- facilitation in completion performance is re- tional memory tests, such as free recall, cued ferred to as a repetition or direct priming recall, and recognition; a denning character- effect (cf. Cofer, 1967; Cramer, 1966). istic of these tests is the explicit instruction Direct priming effects have also been dem- to retrieve information about a particular onstrated with other tasks, such as word experience. However, this assumption is not identification, which requires subjects to consistent with the results of studies that have identify words that are presented very briefly, assessed performance with tests that make and lexical decision, which requires subjects no reference to particular experiences. On to decide whether or not a presented letter these tests memory is expressed by a facili- string constitutes a word. A variety of terms tation in performance that need not be ac- has been used to distinguish between the type companied by conscious recollection. For ex- of memory that is tapped by priming tests ample, on a word completion test subjects on the one hand, and by recall and recognition are presented word fragments (e.g., BRE tests on the other hand (e.g., Baddeley, 1982; or B E D for BREAD) and are instructed Bruner, 1969; Cohen & Squire, 1980; Graf, to complete them with the first words that Squire, & Mandler, 1984; Jacoby & Wither- come to mind. When required to complete spoon, 1982; Mandler, 1980; Moscovitch, 1982; O'Keefe & Nadel, 1978; Schacter, in press-a; Schacter & Moscovitch, 1984; Tulv- This research was supported by Grant U0299 to Peter Graf and Grant UO361 to Daniel L. Schacter from the ing, 1983; Warrington & Weiskrantz, 1982; Natural Sciences and Engineering Research Council of Wickelgren, 1979). For descriptive purposes, Canada and by a Special Research Program Grant from we use the terms implicit memory and explicit the Connaught Fund, University of Toronto. memory to distinguish between these forms We thank Reed Hunt, Larry Jacoby, Morris Moscov- of memory. Implicit memory is revealed when itch, Norm Slamecka, and Endel Tulving for their com- ments on earlier versions of this paper, Karen Raaflaub- performance on a task is facilitated in the Walsh for testing the amnesic patients, and Carol Mac- absence of conscious recollection; explicit donald for helping to prepare this manuscript. memory is revealed when performance on a Requests for reprints should be sent to Peter Graf, task requires conscious recollection of pre- Department of Psychology, University of Toronto, Toronto, Ontario, Canada M5S 1A1. vious experiences. The term priming effect is 501 502 PETER GRAF AND DANIEL L. SCHACTER used in the present article to refer to the neurological dysfunctions. Amnesic patients faeilitative effects of a learning episode on show a pattern of cognitive functioning that performance of a word completion or similar is typically characterized by poor memory task. for recent events along with relatively normal Three converging lines of evidence suggest intellectual functions and premorbid knowl- that implicit memory differs fundamentally edge and skills (e.g., Moscovitch, 1982; from explicit memory. First, performance on Schacter, 1983; Squire, 1982; Squire, Cohen, implicit and explicit memory tasks is affected & Nadel, 1984; Warrington & Weiskrantz, differentially by a variety of experimental 1982). One of the hallmarks of anterograde manipulations (e.g., Graf & Mandler, 1984, amnesia is the patients' inability to acquire 1985; Graf, Mandler, & Haden, 1982; Graf, and remember new associations; failure to Squire, & Mandler, 1984; Jacoby & Dallas, remember a list of unrelated word pairs on 1981; Roediger & Blaxton, 1983; Scarbor- an immediate recall test is diagnostic of ough, Gerard, & Cortese, 1979; Tulving, amnesia (e.g., Meyer & Yates, 1955; Rozin, Schacter, & Stark, 1982). Second, perfor- 1976; Scoville & Milner, 1957; Starr & Phil- mance on tests of implicit retention can be lips, 1970). Nevertheless, amnesic patients statistically independent of recognition per- show relatively normal priming or implicit formance (Eich, 1984; Jacoby, & Wither- retention, when the study materials are fa- spoon, 1982; Schacter, McLachlan, Moscov- miliar items that have a preexisting memory itch, & Tulving, 1984; Tulving et al., 1982). representation, such as words (e.g., Graf, Third, patients with organic amnesia, who Squire, & Mandler, 1984; Rozin, 1976; War- are severely impaired on explicit memory rington & Weiskrantz, 1970), highly related tests, are less impaired, and frequently nor- paired associates (e.g., Shimamura & Squire, mal, on various implicit tests (e.g., Brooks & 1984; Winocur & Weiskrantz, 1976), or lin- Baddeley, 1976; Cohen, 1984; Cohen & guistic idioms (e.g., Schacter, in press-a). The Squire, 1980; Graf, Shimamura, & Squire, finding of normal priming effects, in con- 1984; Graf, Squire, & Mandler, 1984; Jacoby junction with the observation that amnesic & Witherspoon, 1982; Moscovitch, 1982, patients seem unable to acquire new associ- 1984; Schacter, in press-a; Schacter, Harbluk, ations, is consistent with the view that implicit & McLachlan, 1984; Warrington & Weis- memory is mediated by activation of preex- krantz, 1968, 1970, 1974; Weiskrantz & War- isting representations. The reasoning here is rington, 1979). straightforward. A new association has no preexisting memory representation that can A widespread interpretation of the forego- be activated and thus mediate implicit ing findings is that implicit memory is based memory. on the activation of preexisting memory rep- resentations (e.g., Graf et al., 1982; Graf, It is important to observe, however, that Squire, & Mandler, 1984; Mandler, 1980; the negative evidence for the retention of new Mortensen, 1980; Morton, 1969, 1979;Rozin, associations in amnesia has been obtained 1976; Warrington & Weiskrantz, 1982; Wick- on explicit memory tasks, such as paired elgren, 1979). The study materials used in associate learning; these findings thus do not most previous experiments were familiar preclude the possibility that there is implicit items, such as individual words, that are memory for new associations. Several recent represented in long-term or semantic memory studies are relevant to this issue, but the prior to their appearance in a study list. It findings are inconsistent. On the one hand, has been argued that these preexisting rep- there is some evidence for the implicit reten- resentations are activated as a result of pre- tion of newly acquired associations in amnesic senting the study items, and that activation patients who exhibit profound deficits on occurs automatically and thus independently tests of explicit memory (e.g., Moscovitch, of processes that mediate explicit remem- 1984; Schacter, Harbluk, & McLachlan, 1984; bering. Weiskrantz & Warrington, 1979). To illus- An activation interpretation receives sup- trate, Moscovitch required patients to mem- port from several studies of amnesic patients. orize word pairs and later to read these Amnesia occurs as a consequence of various words, either in the same or in different RETENTION OF NEW ASSOCIATIONS 503 pairings, from a degraded display. He found tation, we expected an increase in completion that amnesic patients were faster at reading performance even in the different-context the words in the same pairs than in different condition. The critical question is, however, pairs, suggesting that they had acquired new whether completion performance in the same- associations between the paired words in the context condition exceeds performance in the study list. On the other hand, several studies different-context condition. To the extent that with college students provide conflicting evi- information about the pairing of words in dence concerning the implicit retention of the study list contributes to priming, test new associations (Carroll & Kirsner, 1982; fragments should be completed with words Franks, Plybon, & Auble, 1982;