Implicit Memory

views updated


Psychological investigations of human memory have traditionally been concerned with conscious recollection or explicit memory for specific facts and episodes. Since the 1980s, however, there has been growing interest among the scientific community in a nonconscious form of memory, referred to as implicit memory (Roediger and McDermott, 1993; Schacter, 1987; Toth, 2000). Implicit memory does not require any explicit recollection of specific episodes. Recent experimental investigations have revealed dramatic differences between implicit and explicit memory, and these differences have had a major impact on psychological theories of the processes and systems involved in human memory.

To understand the nature and significance of implicit memory, it is necessary first to consider the types of experimental paradigms that are used to assess both explicit and implicit memory. In a traditional explicit memory paradigm, there are three main phases: a study episode in which people are exposed to a set of target materials, such as a list of words or a set of pictures; a retention interval, which typically lasts for several minutes or hours, during which people perform tasks unrelated to the study phase; and a memory test in which people are asked to think back to the study phase and either produce the target materials (recall) or discriminate the targets from items that were not presented during the study phase (recognition).

The typical implicit memory experiment includes a study phase and a retention interval that are similar to those used in studies of explicit memory. The critical difference between implicit and explicit memory experiments is observed during the test phase. Instead of being asked to try to remember previously studied items, people are simply instructed to perform a perceptual or cognitive task, such as identifying a word from a brief exposure or rating how much they like a face or an object; no reference is made to the prior study episode. However, some of the test items represent previously exposed target items—that is, words, faces, or objects that have been presented during the study episode; other test items represent novel or nonstudied words, faces, or objects that have not been presented during the study phase. Implicit memory for a previously studied item is inferred when it can be shown that subjects' task performance is influenced by prior exposure to studied items.

The hallmark of implicit memory is a facilitation of performance that is attributable to information acquired during a specific episode on a test that does not require conscious recollection of the episode. This facilitation of performance is often referred to as direct or repetition priming. Examples of tests used to assess priming include stem or fragment completion tasks, in which people are asked to complete word stems or fragments (e.g., tab) with the first word that comes to mind (e.g., table), and priming is inferred from an enhanced tendency to complete the stems with previously studied words relative to non-studied words; and perceptual identification tasks, in which people try to identify a word or object from a brief (e.g., fifty milliseconds) perceptual exposure, and priming is indicated by more accurate identification of recently studied items than of new, nonstudied items.

Although it has been studied quite extensively, priming is not the only type of implicit memory. For instance, tasks in which people learn motor or cognitive skills may involve implicit memory, in the sense that skill acquisition does not require explicit recollection of a specific previous episode but does depend on information acquired during such episodes. Similarly, tasks in which people are required to make various kinds of cognitive judgments, such as judging whether a name is famous or how much they like a face, can be influenced by implicit memory, in the sense that the cognitive judgment may be altered by information acquired during a study episode, even in the absence of conscious recollection of the episode.

Dissociations Between Implicit and Explicit Memory

One of the most fascinating aspects of implicit memory is that it can be separated or dissociated from explicit memory—that is, there are various experimental manipulations that affect implicit and explicit memory differently, and there are populations of people for whom explicit memory is impaired while implicit memory is spared.

Population Dissociations

Perhaps the most dramatic dissociation between implicit and explicit memory has been provided by studies of brain-damaged patients with organic amnesia. Amnesic patients are characterized by a severe impairment of explicit memory for recent events, although intelligence, perception, and language are relatively normal. Lesions to either medial temporal or diencephalic brain regions typically produce this memory deficit. In contrast, a number of studies have demonstrated that amnesic patients show intact implicit memory. For example, researchers have found repeatedly that amnesic patients show just as much priming as do normal subjects on stem completion and similar tasks, despite their inability to remember explicitly the target items or the study episode in which the items were encountered (Shimamura, 1993). In addition, it has been demonstrated that amnesic patients often show normal or near-normal learning of motor and perceptual skills, and that their cognitive judgments can be biased by information acquired during specific episodes that the amnesic patient cannot remember explicitly (Mayes, 2000).

Researchers have reported several other population dissociations. For instance, healthy older adults generally show worse explicit memory than younger adults. However, older adults often produce the same amount of priming on implicit tests (Zacks, Hasher, and Li, 2000). Similarly, patients with depression or schizophrenia have deficits in explicit remembering (compared to healthy control subjects) but produce intact priming.

Pharmacological Dissociations

The administration of certain drugs (e.g., benzodiazepines, a class of drugs used in anesthesia and to treat anxiety) produce similar dissociations between implicit and explicit memory. Scientists have long known that these drugs produce a temporary form of amnesia that wears off as the drug is metabolized. However, recent research indicates that these effects are restricted to explicit memory. In a typical study, one group of subjects is administered a dose of a benzodiazepine and another group is given a placebo, after which both groups are presented with study materials. Some time later, memory is tested. Compared to the placebo group, the group given the drug exhibits poor explicit memory for the studied materials. In contrast, the two group produce equivalent performance on implicit tests, such as stem and fragment completion. Thus, drug-induced amnesia produces the same type of dissociation as that produced by organic amnesia: It affects conscious recollection but appears to have little or no effect on nonconscious influences of memory (Curran, 2000).

Functional Dissociations

Experimental manipulations also produce dissociations between implicit and explicit memory, providing insight into functional differences between these forms of memory. For example, researchers have observed one important dissociation in experiments that have compared the effects of semantic and nonsemantic study tasks on implicit and explicit memory. Research has well established that performance on explicit recall and recognition tests is higher following semantic, rather than nonsemantic, study of an item; researchers have called this the levels of processing effect. For example, when subjects perform a semantic encoding task during the study phase of an experiment (e.g., rate the pleasantness of a word or provide a definition of the word), subsequent probability of explicitly remembering the word is much higher than if subjects perform a nonsemantic encoding task during the study phase (e.g., count the number of vowels and consonants in the word). In contrast, however, research has shown that the magnitude of priming on a stem completion task or perceptual identification task is less affected, or even unaffected, by the same manipulation: Priming effects do not differ significantly following semantic and nonsemantic encoding (Roediger and McDermott, 1993).

Conversely, the similarity between the physical (or perceptual) features of the stimuli as presented at the time of study and test has a strong effect on many implicit memory tests but little or no effect on many explicit tests. An example is the effect of study modality. If some of words on a study list are presented visually and some are presented aurally, later explicit memory for the words is typically unaffected. However, when memory is tested with the implicit tests of stem and fragment completion, study modality has a large impact; visually presented words leading to more priming than the aurally presented words. Presenting study items as pictures versus words produces similar matching effects on implicit memory, in which words produce more priming on word-based implicit tests (e.g., stem and fragment completion) and pictures produce more priming on implicit tests which use pictorial cues (such as picture fragment completion).

The foregoing is only a partial list of dissociations between implicit and explicit memory. In addition, similarities between implicit and explicit memory have also been observed; this is to be expected, because both are forms of memory and hence presumably share some common characteristics. However, the differences between the two are most revealing theoretically and have led to a variety of proposals concerning the nature of implicit memory.

Theoretical Accounts of Implicit Memory

Some explanations of implicit memory focus heavily on the sorts of dissociations described above and emphasize differences between implicit and explicit memory. Other explanations seek to elucidate common principles that may underlie both implicit and explicit memory phenomena.

Activation View

An early theoretical view held that implicit memory is attributable to the temporary activation of preexisting units or nodes in memory: Exposure to a word or object automatically activates a memory representation of it, and this activation subsides rapidly. Although able to accommodate some experimental results, this general idea has difficulty accounting for other important results. First, priming can be surprisingly long-lived under certain conditions lasting weeks and months. Second, priming has been observed following exposure to unfamiliar materials, such as nonsense words or novel patterns and objects, that do not have any preexisting representation in memory as a unit. Third, priming often shows perceptual specificity (e.g., effects of study modality), which seems incompatible with the idea that priming is attributable to abstract, amodal representations of words and concepts.

Multiple Memory Systems

Neuropsychological and neuroscientific analyses have proposed multiple memory systems for implicit and explicit memory. Because amnesia is typically associated with damage to the hippocampus and medial temporal lobes, scientists believe that these parts of the brain are necessary for explicit memory. Initially, theories differentiated between two memory systems (e.g., declarative versus procedural, episodic versus semantic) to account for explicit and implicit memory, respectively. However, as evidence for dissociable forms of implicit memory mounted, the number of proposed systems has increased. A 2000 version of this theory proposes four: episodic memory, semantic memory, the perceptual representation system (PRS), and procedural memory (Schacter, Wagner, and Butter, 2000). Episodic memory stores information about episodes from one's personal past, enabling the experience of recollection. Semantic memory stores general knowledge about the world, including facts, conceptual information, and vocabulary. The PRS is a perceptual memory systems that processes information about the form and structure of words and objects independently of their semantic content. Finally, procedural memory represents knowledge of cognitive and motor skills. Explicit memory is assumed to be a product of the episodic memory system, whereas various forms of implicit memory are produced by the other systems.

Population and pharmacological dissociations argue for the existence of multiple memory systems in the brain. For example, amnesia produces deficits on explicit memory but typically not on verbal implicit memory tests (such as stem and fragment completion) or on tests of skill learning, indicating that amnesia damages the episodic memory system but not the other systems. Even stronger support for the multiple-memory-systems view comes from reports of double dissociations, dissociations in which two different patient groups (with damage to different parts of the brain) exhibit complementary dissociative patterns on implicit and explicit memory tests. For example, amnesic patients (with damage to medial-temporal regions of the brain) have disrupted memory on explicit tests but not on implicit tests, as noted. The opposite dissociation occurs in patients with occipital-lobe lesions, who exhibit preserved explicit memory coupled with deficits in implicit memory for visual-perceptual information (Gabrieli et al., 1995). This provides strong support for the view that brain systems mediating performance on these two types of tests differ. Similar dissociative patterns support distinctions among the other imputed memory systems (Schacter et al., 2000).

Transfer-Appropriate Processing

Although population and pharmacological dissociations provide strong support for distinct memory systems, functional dissociations have often been interpreted in terms of the divergent processing requirements of different memory tests within a single memory system. The general idea is that dissociations between implicit and explicit memory are special cases of the general principles of transfer-appropriate processing and encoding specificity, which hold that memory performance is best when encoding and retrieval processes match. Specifically, it is held that most standard explicit memory tests require a good deal of conceptual processing: semantically based, subject-initiated attempts to recollect the study episode. By contrast, performance on such implicit memory tests as word completion and perceptual identification is held to be largely dependent on perceptual, or data-driven, processing—processing that is determined largely by the physical characteristics of test cues. It thus follows that explicit memory—but not priming—should benefit from semantic study processing (which researchers believe support conceptually driven processing) more than from nonsemantic study processing, whereas priming should be strongly dependent on matching of surface features between study and test (Jacoby, 1983; Roediger and McDermott, 1993). However, it is possible to devise implicit tests that entail conceptual processing, and the transfer-appropriate processing view has led to predictions and corresponding demonstrations of dissociations between implicit tests, by contrasting implicit tasks that draw primarily on perceptual processing with implicit tasks that draw primarily on conceptual processing.

The transfer-appropriate processing view is parsimonious and has enjoyed great success in accounting for functional dissociations between explicit and implicit tests, and among implicit tests of different types (i.e., perceptual and conceptual). Thus, the distinction between perceptual and conceptual processing is likely an important aspect of any explanation of implicit memory phenomena. However, it does not provide a complete account because it does not naturally account for population and pharmacological dissociations. In particular, it has been generally found that both organic and drug-induced amnesia disrupt performance on explicit tests (whether conceptual or perceptual) but produce normal levels of priming on conceptual as well as perceptual implicit tests. Likewise, older and younger adults generally produce equivalent levels of conceptual and perceptual priming, even though older adults perform worse on tests of explicit memory. These dissociations are difficult to explain in terms of perceptual versus conceptual processing.

Component-Processes Approach and Evidence from Neuroimaging

The complementary successes of multiple memory systems view and the transfer-appropriate processing approach has produced the view that emphasizes multiple forms of priming and attempts to articulate the component processes that mediate performance on various memory tasks (Foster and Jelicic, 1999). Neuroimaging techniques, such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI), play a critical role in supplementing traditional approaches. These techniques attempt to determine neural regions involved in explicit and implicit memory tests. One result is the substantial degree of overlap in the active brain regions during explicit and implicit retrieval tasks. Despite this overlap, there are important differences in the patterns of brain activation. In particular, explicit retrieval relies heavily on anterior frontal (especially right prefrontal) lobe as well as the medial-temporal regions, including the hippocampus. Scientists believe the activity in frontal lobe reflects an explicit retrieval mode in which the individual intentionally tries to retrieve information about past events and is generally oriented toward the past. They believe the medial temporal activity contributes to the recollective experience itself when a memory is successfully retrieved (Nyberg and Cabeza, 2000). Another finding is that priming on implicit tests is often associated with decreased activity in certain regions of the brain, which may reflect a reduction in processing demands when a stimulus is processed a second time (Schacter and Badgaiyan, 2001). The brain areas involved depend on whether the implicit test is perceptual or conceptual. Perceptual tests, such as stem or fragment completion, produce decreased processing in visual cortex (in the posterior occipital lobe). In contrast, conceptual implicit tests produce decreased activity in the inferior frontal lobe and the mid-temporal lobe. In general, these results indicate that the same neural substrates required for the initial (perceptual or conceptual) processing are reengaged at the time of test and exhibit the effects of the initial processing by their subsequent reduced activity.



Curran, H. V. (2000). Psychopharmacological approaches to human memory. In M. S. Gazzaniga, ed., The new cognitive neurosciences, 2nd edition. Cambridge, MA: MIT Press.

Foster, J. K., and Jelicic, M., eds. (1999). Memory: Systems, process, or function? New York: Oxford University Press.

Gabrieli, J. D. E. et al. (1995). Double dissociation between memory systems underlying explicit and implicit memory in the human brain. Psychological Science 6, 76-82.

Jacoby, L. L. (1983). Remembering the data: Analyzing interactive processes in reading. Journal of Verbal Learning and Verbal Behavior 22, 485-508.

Mayes, A. R. (2000). Selective memory disorders. In E. Tulving and F. I. M. Craik, eds., The Oxford handbook of memory. New York: Oxford University Press.

Nyberg, L., and Cabeza, R. (2000). Brain imaging of memory. In E. Tulving and F. I. M. Craik, eds., The Oxford handbook of memory. New York: Oxford University Press.

Roediger, H. L., and McDermott, K. B. (1993). Implicit memory in normal human subjects. In F. Boller and J. Grafman, eds., Handbook of neuropsychology. Amsterdam: Elsevier.

Schacter, D. L. (1987). Implicit memory: History and current status. Journal of Experimental Psychology: Learning, Memory, and Cognition 13, 501-518.

Schacter, D. L., and Badgaiyan, R. D. (2001). Neuroimaging of priming: New perspectives on implicit and explicit memory. Current Directions in Psychological Science 10, 1-4.

Schacter, D. L., Wagner, A. D., and Buckner, R. L. (2000). Memory systems of 1999. In E. Tulving and F. I. M. Craik, eds., The Oxford handbook of memory. New York: Oxford University Press.

Shimamura, A. P. (1993). Neuropsychological analyses of implicit memory: History, methodology, and theoretical interpretations. In P. Graf and M. E. J. Masson, eds., Implicit memory: New directions in cognition, development, and neuropsychology. Hillsdale, NJ: Erlbaum.

Toth, J. P. (2000). Nonconscious forms of human memory. In E. Tulving and F. I. M. Craik, eds., The Oxford handbook of memory. New York: Oxford University Press.

Zacks, R. T., Hasher, L., and Li, K. Z. H. (2000). Human memory. In F. I. M. Craik and T. A. Salthouse, eds., The handbook of aging and cognition, 2nd edition. Mahwah, NJ: Erlbaum.

Daniel L.Schacter

Revised byNeil W.Mulligan