Priming and Multiple Memory Systems: Perceptual Mechanisms of Implicit Memory The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters Citation Schacter, Daniel L. 1992. Priming and multiple memory systems: Perceptual mechanisms of implicit memory. Journal of Cognitive Neuroscience 4(3): 244-256. Published Version doi:10.1162/jocn.1992.4.3.244 Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:3627272 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA Priming and Multiple Memory Systems: Perceptual Mechanisms of Implicit Memory Daniel L. Schacter Department of Psychology Harvard University Abstract Research examining the relation between explicit and im- of cortically based subsystems that operate at a presemantic plicit forms of memory has generated a great deal of evidence level and support nonconscious expressions of memory. Three concerning the issue of multiple memory systems. This article PRS subsystems are examined: visual word form, structural focuses on an extensively studied implicit memory phenome- description, and auditory word form. Pertinent cognitive, neu- non, known as direct or repetition priming, and examines the ropsychological, and neurobiological evidence is reviewed, al- hypothesis that priming effects on various tasks reflect the ternative classificatory schemes are discussed, and important operation of a perceptual representation system (PRSta class conceptual and terminological issues are considered. INTRODUCTION memory systems. However, interest in the relation be- tween explicit and implicit forms of memory has been During the past 25 years, questions concerning the na- sparked by demonstrations of striking dissociations be- ture and number of memory systems have been at the tween the two that do indeed suggest that different un- forefront of cognitive, neuropsychological, and neuro- derlying systems are involved in explicit and implicit biological research (for historical overview, see Polster, memory, respectively. Thus, for example, it has been Nadel, & Schacter, 1991). In the study of human memory, known for many years that amnesic patients exhibit ro- a key line of evidence for multiple memory systems has bust and sometimes normal learning of various percep- been provided by investigations concerned with the de- tual, motor, and cognitive skills despite impaired or scriptive distinction between explicit and implicit forms absent explicit memory for having acquired them (e.g., of memory (Graf & Schacter, 1985; Schacter, 1987). Ex- Cohen & Squire, 1980; Milner, Corkin, & Teuber, 1968). plicit memory refers to intentional or conscious recol- Amnesic patients can also exhibit classical conditioning lection of prior experiences, as assessed in the laboratory effects despite poor explicit memory (Daum, Channon, by traditional tests of recall or recognition; implicit mem- & Canavar, 1989; Weiskrantz & Warrington, 1979), and ory, by contrast, refers to changes in performance or acquire knowledge needed to perform complex com- behavior that are produced by prior experiences on tests puter-related tasks despite the absence of any recollec- that do not require any intentional or conscious recol- tion for having previously performed the tasks (Glisky, lection of those experiences. The distinction between Schacter, & Tulving, 1986; Glisky & Schacter, 1987, 1988, explicit and implicit memory is similar to distinctions 1989). between memoy with awareness vs. memoy without Perhaps the most intensively studied form of implicit awareness (Jacoby & Witherspoon, 1982), declarative memory has come to be known as repetition or direct memory vs. nondeclarative memoy (Squire, 1992), and priming: the facilitated identification of perceptual ob- direct memory vs. indirect rnemoy (Johnson & Hasher, jects from reduced cues as a consequence of a specific 1987). However, these distinctions are used less fre- prior exposure to an object (e.g., Tulving & Schacter, quently in the literature than is the explicitlimplicit dis- 1990). Priming can be thought of as a form of implicit tinction, and there are various reasons to prefer the memory in the sense that it can occur independently of explicithmplicit contrast over alternative terms (Roedi- any conscious or explicit recollection of a previous en- ger, 1990). counter with a stimulus. Thus, amnesic patients can show The explicitlimplicit distinction is a descnjwive one that entirely normal priming as a consequence of a recent contrasts two different ways in which memory for pre- encounter with a word or object, despite impaired or vious experience can be expressed; it does not refer to, even absent explicit memory for the word or object; and or necessarily imply the existence of, distinct underlying studies of nonamnesic, normal subjects have shown that 0 192Masachusetts Institute of Technology Journal of Cognitive Neuroscience Volume 4, Number 3 various experimental manipulations affect priming and some of these patients could be attributed to the break- explicit memory in different and even opposite ways (for down of a system that represents information about the reviews, see Richardson-Klavehn & Bjork, 1988; Roediger, visual and orthographic form of words. Evidence that 1990; Schacter, 1987; Schacter, Chiu, & Ochsner, 1993; such a system operates at a presemantic level is provided Shimamura, 1986). These and other observations indicate by studies that have focused on brain-damaged patients that the kind of information about a recently encountered who maintain relatively intact abilities to read words yet word or object that supports priming is quite different exhibit little or no understanding of them (cf. Sartori, from the kind of information that supports explicit rec- Masterson, &Job, 1987; Schwartz, Saffran, & Marin, 1980). ollection for an encounter with the word or object. More- Importantly, such patients can read words with irregular over, priming has also been dissociated from skill spellings, thereby indicating that they can gain access to learning: studies of dementia indicate that patients with the representations in the word form system (see Schac- Alzheimer’s disease show impaired priming and intact ter, 1990, for further elaboration). Data from neuro- motor skill learning, whereas patients with Huntington’s imaging studies using positron emission tomography disease show the opposite pattern (e.g., Butters, Heindel, (PET) suggest that the visual form system is based in & Salmon, 1990). A number of investigators have argued regions of extrastriate occipital cortex and is neuroana- further that priming is the expression of a neurocognitive tomically distinct from brain regions subserving semantic system that differs functionally and neuroanatomically processing (e.g., Petersen et al., 1989). from the neurocognitive system that supports explicit Several lines of evidence have led to the proposal that remembering and skill learning, respectively (cf., Cohen, the visual word form system subserves priming effects 1984; Schacter, 1985, 1990; Butters et al., 1990; Squire, on so-called data driven or perceptually based implicit 1987, 1992; Tulving, 1985; Tulving & Schacter, 1990). memory tasks, such as stem or fragment completion, This article examines in some detail one such pro- where subjects provide the first word that comes to mind posal, namely, that priming reflects, to a very large extent, in response to three-letter stems or graphemic fragments, the operations of a perceptual representation system and perceptual or word identijication, where subjects (PRS) that can function independently of the episodic or attempt to identi@ briefly presented words. One such declarative memory system that supports explicit mem- line of evidence is that amnesic patients show normal ory (Schacter, 1990, 1992; Tulving & Schacter, 1990). PRS priming of familiar words and word pairs on completion, refers to a class of domain-specific subsystems, based in identification, and similar tasks (cf. Cermak et al., 1985; cortical regions, that process and represent information Graf, Squire, & Mandler, 1984; Moscovitch, 1982; Schac- about the form and structure, but not the meaning and ter, 1985; Shimamura & Squire, 1984; Tulving, Hayman, other associative properties, of words and objects. This & Macdonald, 1991; Warrington & Weiskrantz, 1974). article will focus on delineating and evaluating charac- These results are consistent with the proposal that visual teristics of, and evidence for, three PRS subsystems: visual word priming is mediated by a perceptual system based word form, structural description, and auditory word in posterior cortical regions, because the critical sites of form. Although they probably do not constitute an ex- brain damage in amnesic patients typically involve the haustive list of PRS subsystems, various kinds of evidence limbic system and medial temporal lobe structures; pos- about them is available, including data from priming terior cortex is spared in the amnesic syndrome (e.g., studies. Each of the subsystems differs from the others Rozin, 1976; Scoville & Milner, 1957; Squire, 1992; Weis- in several ways, but all share common features: they krantz, 1985). operate at apresemantic level, that is, at a level of pro- Further evidence from the study of amnesia that bears cessing