Explicit and Implicit Memory in the Elderly: Evidence for Double Dissociation Involving Medial Temporal- and Frontal-Lobe Functions

Transcription

1 Neuropsychology Copyright 1996 by the American Psychological Association, Inc. 1996, Vol. 10, No. 1, /96/$3.00 Explicit and Implicit Memory in the Elderly: Evidence for Double Dissociation Involving Medial Temporal- and Frontal-Lobe Functions Gordon Winocur, Morris Moscovitch, and Donald T. Stuss Rotman Research Institute of Baycrest Centre for Geriatric Care Groups of normal old people living in institutions or in the community and young adults were administered tests of implicit (IM) and explicit (EM) memory with word-stem (WSC) and word-fragment (WFC) completion paradigms. Neuropsychological tests sensitive to frontal and medial temporal lobe function were also administered. Age differences were observed on both tests of EM and on all neuropsychological tests. Priming effects on WSC were smaller in the institutionalized group than the other 2 groups. Comparisons of EM and IM test results with neuropsychological test scores revealed several effects, including significant correlations (a) between EM scores on both tests and performance on standard memory tests in both aged groups and (b) between IM scores of both aged groups on WSC and frontal-lobe test performance. The results provide evidence of a double dissociation with respect to involvement of brain regions in EM and IM. They also indicate that repetition priming in WSC and WFC involve different mechanisms and that frontal-lobe dysfunction is a factor in reduced priming on the WSC test. Memory loss is a frequent complaint of the elderly, but extensive research over several decades has shown that not all aspects of memory function decline at the same rate. Age differences favoring young adults are most consistently reported on explicit (or direct) tests of memory that use recall or recognition techniques to assess conscious recollection of contextually based information. This type of memory loss is often attributed to progressive deterioration of the hippocampus, a medial temporal lobe brain region that is critical to memory function and known to be particularly sensitive to the aging process (Albert & Stafford, 1986; Milner, 1966; Tomlinson, 1972). Implicit (or indirect) tests of memory do not refer to specific events. In these tests, memory is assessed by the effects of experience on some measure of performance that does not involve conscious awareness of any part of the prior experience. Tests of general knowledge, skill learning, and repetition priming in which exposure to a word biases subsequent Gordon Winocur, Rotman Research Institute of Baycrest Centre for Geriatric Care, Toronto, Ontario, Canada, Department of Psychology, Trent University, Peterborough, Ontario, Canada, and Departments of Psychology and Psychiatry, University of Toronto, Toronto, Ontario, Canada; Morris Moscovitch, Department of Psychology, Erindale College, University of Toronto and Rotman Research Institute of Baycrest Centre for Geriatric Care; Donald T. Stuss, Rotman Research Institute of Baycrest Centre for Geriatric Care and Departments of Psychology and Medicine (Neurology), University of Toronto, Toronto, Ontario, Canada. This research was supported by grants from the Medical Research Council of Canada and the Ontario Mental Health Foundation. We thank Lars Nyberg and Marko Jelicic for comments on an earlier version of this article. Portions of this research were presented at the Annual Meeting of the Memory Disorders Research Society, Boston, October Correspondence concerning this article should be addressed to Gordon Winocur, Rotman Research Institute of Baycrest Centre for Geriatric Care, 3560 Bathurst Street, North York, Ontario, Canada, M6A 2El. Electronic mail may be sent to winocur@psych.utoronto.ca. identification in a word completion test are common examples of implicit tests of memory. In general, cognitive processes that underlie normal performance on such tests are preserved in old age (see Howard, 1991; Light & LaVoie, 1993, for reviews), but there are notable exceptions (Chiarello & Hoyer, 1988; Davis et al., 1990; Howard, Shaw, & Heisey, 1986; Hultsch, Masson, & Small, 1991; Rose, Yesavage, Hill, & Bower, 1986). A primary purpose of the present study was to investigate one type of implicit memory (word-stem completion) on which age differences have been reported and to examine these differences from a neuropsychological perspective. Word completion as a measure of priming is usually studied in one of two ways. In word-fragment completion, the individual is presented initially with the complete word and then at test is given a sample of the word's letters (e.g., soldier; _ o i_ r); in word-stem completion, the study phase is the same, but at test, the first three letters of the target word are presented (e.g., soldier; s o l.... ). In both cases, at test, the individual is instructed to respond with the first word that comes to mind. Comparisons of old and young people on tests of wordfragment completion reliably show no differences (Light & Lavoie, 1993), but at least three investigations of word-stem completion found less priming in aged individuals (Chiarello & Hoyer, 1988; Davis et al., 1990; Hultsch et al., 1991). Hultsch et al. (1991) speculated that inconsistent findings of age effects in word completion priming studies are related to the extent to which implicit tests are contaminated by explicit memory. Critical differences in experimental procedure may bias individuals toward using conscious memory strategies that put the elderly at a disadvantage. For example, this would occur if an explicit test of memory were administered prior to the implicit priming test or if instructions for the priming test implied that memory would be tested. However, after careful examination, Hultsch et al. (1991) concluded that reduced priming by old people on word-stem completion tests represented a legitimate age effect independent of contaminating influences of explicit memory. 57

2 58 WINOCUR, MOSCOVITCH, AND STUSS There is a suggestion in the neuropsychological literature that word completion performance is not always normal in memory-impaired individuals (Squire, Shimamura, & Graf, 1987) and that stem completion may be mediated by frontal lobe regions of the brain. Mayes and Gooding (1989) tested a mixed group of brain-damaged amnesia patients on new associative priming in a word-stem completion task. One of the interesting findings was that priming effects in the amnesia patients, several of whom were Korsakoff patients with probable frontal lobe dysfunction, correlated significantly with performance on the FAS test of verbal fluency (Borkowski, Benton, & Shapiro, 1967). Similarly, a significant correlation between word-stem completion and performance on the Wisconsin Card Sorting Test (WCST), a test of frontal-lobe function, was observed in normal old people by Davis et al. (1990). On the other hand, it should be noted that Shimamura, Gershberg, Jurica, Mangels, and Knight (1992) observed normal word-stem priming in a group of 6 patients with frontal-lobe damage. On balance, overall pattern of results challenge the widely held view that word-stem completion is dependent primarily on perceptual identification mechanisms (Schacter, 1992; Tulving & Schacter, 1990) and suggest that nonperceptual mechanisms are also involved (Keane, Gabrieli, Fennema, Growdon, & Corkin, 1991). In contrast, word-fragment completion is probably a more purely perceptual priming test and as such relies more heavily on posterior brain mechanisms involved in word identification (Keane et al., 1991; Moscovitch, Vriezen, & Goshen-Gottstein, 1993; Schacter, 1992). The available evidence indicates that word-stem and word-fragment completion involve different cognitive operations that are mediated by different neural structures. Because the frontal lobes appears to be more vulnerable to aging effects than posterior regions (Coffey et al., 1992; Fuster, 1989), it can be hypothesized that performance on word-stem completion and word-fragment completion would be affected differentially in old age. In the present study, groups of young and old adults were compared on explicit and implicit tests with word stems and word fragments as partial cues. A primary purpose was to confirm selective age-related effects on word-stem completion and, if successful, to determine if reduced priming on this test is related to impaired frontal-lobe function, as assessed by standard neuropsychoiogical tests (e.g., Wisconsin Card Sorting, FAS). No such correlations were predicted for wordfragment completion. All participants were also administered explicit tests of cued recall, along with standard memory tests that involve conscious recollection (California Verbal Learning Test, Rey, and free recall). There was no a priori reason to expect that age would differentially affect performance on explicit tests that were cued by word stems or word fragments, but it was predicted that age differences in cued recall would correlate with performance on the standard memory tests. The elderly participants were divided into institutionalized and community-dwelling groups. Participants from both environments were selected according to identical criteria that controlled for effects of relevant variables such as age, health, IQ, and education. In previous work (Moscovitch & Winocur, 1983; Winocur & Moscovitch, 1983, 1990), we found that old people living in institutions consistently performed worse on some tests of cognitive function than carefully matched counterparts living in the community. These differences were most apparent on tests of frontal-lobe and medial temporal-lobe function, suggesting that these regions of the brain were particularly affected by the combined effects of aging and institutionalization. It was of interest to determine if differences between institutionalized and community-dwelling participants extended to the cognitive and memory functions under investigation in the present study. Participants Method A total of 50 old adults (age range = years) and 29 young adults (age range = years), living in the Peterborough, Ontario, Canada, area, served as participants in this experiment. Of the elderly sample, 26 lived in senior citizens' residences (institutionalized old; group INS), and 24 lived in their own homes (community old; group COM). The young adult group was recruited primarily through newspaper ads. All groups were comprised according to the same criteria. To qualify, participants had to be free of gross cognitive deficits or neurological conditions; anyone with a history of stroke or serious cardiovascular disease, chronic psychiatric problems, or sensory or perceptual disorders, was excluded from the study, as were people receiving medication that affected mood or thought processes. Medical information was obtained from various sources, including a health checklist, medical records, relatives, attending physicians, and other health care providers. In all cases, permission to obtain this information was obtained from the participants. As part of the screening process, all prospective participants were administered the Vocabulary subscale of the Wechsler Adult Intelligence Scale (Wechsler, 1981) (WAIS) and the CES Depression Scale (Radloff, 1977). Only those individuals scoring at least in the normal range on these tests were included in the study. Participants were also selected according to educational background and socioeconomic level. All those in the research had been to school for at least 8 years, and most had completed high school. For the most part, participants were drawn from middle-income levels with a wide range of professional and semiskilled occupations represented. Table 1 provides a summary of demographic and screening information for all groups. The three groups were similar in terms of the controlled variables. Test Materials and Procedure Two sets of 60 familiar nouns were created---one for the word-stem completion (WSC) tasks and the other for the word-fragment comple- Table 1 Characteristics of Participants in the Three Study Groups Institutionalized Community Young Characteristic M SD M SD M SD Age (years) Education (years) WAIS Vocabulary (absolute score) CES--Depression Scale (score) Note. CES = Center for Epidemiological Studies; WAIS = Wechsler Adult Intelligence Scale.

3 EXPLICIT AND IMPLICIT MEMORY 59 tion (WFC) tasks. All words in the WSC lists contained five letters. Each word began with a different three-letter stem that could be completed to form several words. All words in the WFC lists consisted of six letters. The stimulus words, 40 in each set, were printed in large black letters on 3 5 in. (8 x 13 cm) white index cards. All participants received explicit and implicit versions of the WSC and WFC tasks. Half of the participants received the WSC tasks followed approximately 6 weeks later by the WFC tasks; the other half were tested in the reverse order. Training and testing procedures were similar for both tasks. Initially, in both cases the 40 stimulus words were presented one at a time to each participant. As each word was presented, the participant was instructed to tell the examiner how many vowels there were in the word. A second presentation trial followed in which the words were shown in the same way, but this time participants were asked to read aloud each word. Testing followed a 20-min delay that was filled with nonverbal testing (e.g., Rey-Osterrieth and line orientation), and distracting conversation. For each test condition, the participant was given three response sheets, each consisting of 20 incomplete words. On two of the sheets, the letter cues referred to 20 stimulus words, and on the third to 20 control words that the participants had not seen before. The control words were included to provide baseline estimates of word completion. For WSC testing, the first three letters for each of the 60 words were presented followed by two blank lines (e.g., beach; Bea ). For WFC testing, three of the six letters were presented as cues, separated by blank lines (e.g., radish; ad h). The selection of letters as cues was arbitrary in each case. Test instructions were read aloud to each participant and were the same for WSC and WFC testing. Instructions for implicit recall, which preceded one list of stimulus words and the list of control words, were as follows: Here we have a list of parts of some common words. What I want you to do is fill out these words to make the first word that comes to mind. The instructions for explicit recall, which preceded one list of stimulus words, were as follows: Here we have a list containing parts of words that I showed you earlier on the cue cards. What I want you to do is fill these words correctly, so that they match the words you saw on the cue cards. The order of list presentation was randomly determined for each participant. The participants were instructed to respond as quickly as possible and advised that they had 3 min to complete each task. During the interval between WSC and WFC testing, participants were administered a series of neuropsychological tests. The tests were usually administered on two or three occasions, each separated by about a week. The following is a summary of each test, including the Rey-Osterrieth test, which was administered between the WSC study and test trials. Wisconsin Card Sorting Test. This is a standardized test of set formation and attentional shift in which participants sort cards containing multidimensional drawings into different categories. The procedure followed in this study has been described in detail by Milner (1964). The test yields several measures believed to be sensitive to frontal-lobe dysfunction (Milner, 1964); those reported here are total errors and perseverative errors. Word fluency. In this test, individuals are asked to generate orally words beginning with the letters, F, A, S, with 1 min allowed for each letter (Borkowski et al., 1967). The measure reported here is the total number of words produced in 3 min. The test is widely used to assess frontal lobe damage in clinical populations (Milner, 1964). Rey-Osterrieth Complex Figure Test. In this test, the individual initially copies a complex line drawing and 20 rain later reproduces it from memory. Standardized scoring procedures are used to assess performance at test. Deficits in recalling the drawing are associated with right temporal lobe dysfunction (Taylor, 1969). California Verbal Learning Test (CVL T). The CVLT is a widely used standardized test of short- and long-term free recall of a categorized list (Delis, Kramer, Kaplan, & Ober, 1987). In this test, four words from each of four categories are presented aurally in a mixed order, and the participant recalls as many of the 16 words as possible. The list was presented five times with recall tested following each presentation according to the procedure described by Delis et al. (1987). Delayed recall. A single list of 10 familiar words was constructed and printed in bold, black letters on white index cards. At study, the words were presented one at a time at a 2-s rate, and the participant was instructed to read each word aloud and be prepared recall them later. The list was presented once only and followed by a 60-s delay during which the participant was engaged in distracting conversation. A test of free recall followed in which the participant was instructed to provide as many words as possible in any order from the preceding list. This test was based on a similar test on which old people were impaired relative to young adults (Winocur & Moscovitch, 1990). Results The mean proportions of correct word completions in the explicit and implicit versions of the WFC and WSC tasks are presented in Table 2. In all cases, the scores were derived from the obtained value for participants minus their baseline levels. Baseline measures in the control condition of either task represented the proportion of completed words that matched the target words. The mean baseline rates for WFC and WSC across all groups was.17 and.19 respectively. There were no differences between groups on these measures nor did baseline scores correlate with neuropsychological test performance. Explicit Memory As expected, the proportions of correct responses in both explicit tests of memory were higher in the young group than the aged groups. This was confirmed by separate analyses of variance (ANOVAs), which revealed significant group effects in the WSC, F(2, 76) = 4.29, p <.001, and the WFC, F(2, 76) = 41.61, p <.001, versions of cued recall. Post hoc comparisons, conducted with the Tukey test, showed that episodic memory was not affected by type of cueing in any of the groups. Table 2 Mean Proportion of Words Correctly Recalled on Explicit Version of Word-Stem Completion (WSC) and Word-Fragment Completion (WFC) Tests WSC WFC Group M SD M SD Institutionalized Community-dwelling Young Note. Values represent each participant's obtained mean of each participant's recall score minus the baseline score.

4 60 WINOCUR, MOSCOVITCH, AND STUSS Table 3 Mean Proportion of Correct Word Completion on the Implicit Version of the Word-Stem Completion (WSC) and Word-Fragment Completion (WFC) Tests WSC WFC Group M SD M SD Institutionalized Community-dwelling Young Note. Values represent each participant's obtained priming score minus baseline score. Table 2 shows that the young group's proportion of correct scores generally was higher than those for the INS and COM groups. Post hoc comparisons showed that on the WSC test the three groups differed significantly from each other (ps <.01), indicating an overall age effect, as well as an adverse effect of institutionalization on cued recall performance. On the WFC test, the INS group differed significantly from the COM and young groups, but the latter groups did not differ from each other. Implicit Memory Table 3 shows that the proportion of word completion scores in implicit memory testing did not vary among groups on the WFC task. However, on the WSC task, the institutionalized group completed fewer stimulus words than the communitydwelling and young groups. These observations were confirmed by an ANOVA that revealed a significant difference between groups on the WSC test, F(2, 76) = 3.38,p <.05, but not on the WFC test, F(2, 76) < 1. Subsequent analysis with the Tukey test showed that the group effect on the WSC test resulted from significant differences between the institutionalized group and the other two groups on the WSC task (ps <.05). Neuropsychological Test Scores Mean scores for the three groups on all neuropsychological tests are presented in Table 4. The tests are organized as measures of frontal-lobe and medial temporal-lobe function. Frontal lobe. Strong group differences were observed on both tests of frontal-lobe function--the WCST, total errors, F(2, 76) = 19.29, p <.001; perseverative errors, F(2, 76) = 14.00,p <.001, and the FAS fluency task, F(2, 76) = 34.11, p <.001. The young group consistently performed better than the two aged groups on all measures although by the Tukey test differences between the young and community-dwelling groups on the WCST measures approached statistical significance (both ps <.10). Significant differences favoring the community-dwelling group were observed between the two aged groups on both WCST measures (bothps <.01) and the FAS test (p <.01). Overall, these results indicated an effect of normal aging and a combined effect of aging and institutionalization on frontal-lobe function. Medial temporal lobe. The memory tests yielded a similar pattern of results. Significant differences in group performance were observed on the CVLT, F(2, 76) = 43.28, p <.001, Rey-Osterrieth test, F(2, 76) = 13.76, p <.001, and the Delayed recall test, F(2, 76) = 41.62,p <.001. Tukey tests conducted on the CVLT scores showed that the three groups differed significantly from each other on the CVLT and Delayed recall tests (ps <.05). On both tests, the highest mean scores were obtained by the young group, followed by the community-dwelling group and the institutionalized group. A slightly different pattern was observed on the Rey- Osterrieth test. On this test, the institutionalized group was significantly worse than the young group (p <.01) and the community-dwelling group (p <.05). Although the young group appeared to perform better than the communitydwelling group, the difference did not reach statistical significance (p <.10). Correlations Between Test Scores Pearson correlation coefficients were computed for each group to assess relationships between performance on the explicit and implicit versions of the WSC and WFC tasks and the neuropsychological test scores. The coefficients are presented in the correlation matrices of Tables 5 and 6. Institutionalized group. Examination of the correlation matrices for the institutionalized group reveals a clear double dissociation involving the explicit (see Table 5) and implicit (see Table 6) memory scores on the WSC test and their neuropsychological correlates. This group's mean score on the WSC implicit test correlated significantly with its FAS test score (r =.55; p <.01) and its WCST error scores (total errors: r = -.47; p <.025; perseverative errors: r = -.45; p <.025). At the same time, performance on the explicit memory version of the WSC task correlated significantly with performance on the Delayed recall test (r =.47;p <.025) and CVLT (r =.54; p <.01). The explicit and implicit memory Table 4 Neuropsychological Test Scores for All Study Groups CVLT Rey Delayed recall FAS Total WCST error Perseverative Group M SD M SD M SD M SD M SD M SD Institutionalized Community-dwelling Young Note. CVLT = California Verbal Learning Test; Rey = Rey-Osterrieth Complex Figure Test; FAS = commonly used name for the Controlled Oral Word Association Test or Word Fluency Test; WCST = Wisconsin Card Sorting Test.

5 EXPLICIT AND IMPLICIT MEMORY 61 scores on the WSC task did not correlate significantly with each other. Performance by the institutionalized group on the WFCexplicit memory test was found to correlate significantly with the Delayed recall test (r =.45; p <.02), CVLT (r =.48; p <.025), and the WSC-explicit memory test score (r =.51; p <.025). The WFC-implicit memory score did not correlate significantly with performance on any other test, although the correlation between performance on this test and that of the WSC-implicit memory test approached statistical significance (r =.38;p <.10). Examination of the neuropsychological tests for the INS group revealed significant correlations between the Delayed recall test score and the CVLT (r =.83; p <.001) and FAS (r =.39; p <.05) tests, and between FAS and WCST error scores (total errors: r = -.44; p <.05; perseverative errors: r = -.48;p <.025). Community-dwelling group. A double dissociation, implicating different functions in explicit and implicit memory performance on the WSC task was also observed in the community-dwelling group. As in the institutionalized group, the WSC-implicit memory test score correlated significantly with the FAS (r =.54; p <.01) and WCST total errors (r =-.45; p <.05) and perseverative errors (r =-.57; p <.01) score (see Table 5). The community-dwelling group's score on the WSC-explicit memory test correlated significantly with the CVLT score (r---.44; p <.05), but not with the Delayed recall score (r =.28; p <.20; see Table 6). Interestingly unlike the institutionalized group, for the communitydwelling group, the correlation between the explicit and implicit versions of the WSC test did reach statistical significance (r =.44;p <.05). The correlation between the WFC-explicit memory score and the WSC-explicit memory score was statistically significant (r =.40; p <.05). The community-dwelling group's WFCexplicit memory score also correlated significantly with its scores on the Delayed recall test (r =.43; p <.05) and the Rey-Osterrieth test (r =.40). The correlation between the Table 5 Correlation Between Scores of Institutionalized and Community-Dwelling Elderly on Explicit Versions of Word-Stem Completion (WSC) and Word-Fragment Completion (WFC) Tests and Scores on Neuropsychological Tests Explicit test scores WCST errors Type of Delayed task FAS Total Perseverative CVLT Rey recall Institutionalized WSC ***.36.47** WFC **.12.45** Community-dwelling WSC * WFC *.43* Note. FAS = commonly used name of Controlled Oral Word Association Test or Word Fluency Test; WCST = Wisconsin Card Sorting Test; CVLT = California Verbal Learning Test; Rey = Rey-Osterrieth Complex Figure Test. *p <.05. **p <.025. ***p <.01. Table 6 Correlation Between Scores of Institutionalized and Community-Dwelling Elderly on Implicit Versions of Word-Stem Completion (WSC) and Word-Fragment Completion (WFC) Tests and Scores on Neuropsychological Tests Implicit test scores WCST errors Type of Delayed task FAS Total Perseverative CVLT Rey recall Institutionalized WSC.55*** -.47** -.45** WFC Communi~-dwelling WSC.54** -.45* -.57*** WFC * -.16 Note. FAS = commonly used name of Controlled Oral Word Association Test or Word Fluency Test; WCST = Wisconsin Card Sorting Test; CVLT = California Verbal Learning Test; Rey = Rey-Osterrieth Complex Figure Test. *p <.05. **p <.025. ***p <.01. WFC-explicit and the CVLT test scores approached significance (r =.36;p <.10). As in the institutionalized group, the Delayed recall test score correlated significantly with the CVLT score (r =.60; p <.01), although the significant correlation observed between the Delayed recall and FAS scores in the institutionalized group was not found in the community-dwelling group (r =.23;p <.20). In the community-dwelling group, the FAS score correlated significantly with the CVLT score (r =.44; p <.05) and with the perseverative error score on the WCST (r = -.42; p <.05). The correlation between the FAS and WCST test total score just failed to reach statistical significance (r = -.34;p <.10). Young group. There was very little relationship between any of the scores obtained by the young group. The only significant correlation was found between the WSC-implicit memory and WFC-implicit memory scores (r =.47; p <.01). The correlation between the two explicit memory scores was in the expected direction but failed to achieve statistical significance (r =.26;p <.20). Discussion A major finding of the present study was a functional double dissociation in old people between performance on implicit and explicit tests on the one hand and performance on tests sensitive to different brain regions on the other. The results confirmed that age differences occur in word-stem completion but not in word-fragment completion. In addition, we report the novel finding that in old people performance on stem completion, but not fragment-completion, was correlated with tests of frontal lobe function, but not with standard and laboratory tests of explicit memory. We also found the expected age difference on explicit versions of word-stem and word-fragment completion. However, performance on the explicit tests did not correlate with tests of frontal function but correlated instead with scores on standard memory tests that are sensitive to medial temporal lobe dysfunction.

6 62 WINOCUR, MOSCOVITCH, AND STUSS It is not surprising that performance by older adults on stem completion is not correlated with standard neuropsychological tests of memory sensitive to medial temporal lobe damage. This type of dissociation between implicit and explicit tests of memory has been reported many times in normal individuals and amnesia patients. The important finding is that stem completion was correlated with several measures of frontallobe function. That this correlation did not hold for wordfragment completion indicates that different cognitive processes underlie the two types of priming. This aspect of the results is consistent with the view that fragment completion is perceptually driven and mediated by the posterior neocortex, whereas stem completion has a substantial strategic component. Another explanation for the pattern of correlations observed in the present study is that the relatively poor performance of old people on the frontal tests reflects overall cognitive decline rather than selective impairment. Deficits in WCST and FAS have been reported in patients with posterior cortical lesions, and problems of initiation and word finding may be observed in patients with language disorders, regardless of lesion location (Anderson, Jones, Tranel, Tranel, & Damasio, 1990; Joanette & Goulet, 1986; Miceli, Caltagirone, Gainotti, Massulo, & Silveri, 1981). It could be argued that deficits in these functions, resulting from generalized cortical atrophy, account for performance of old people on our WSC-implicit and frontal tests. However, it is important to note that participants in both aged groups were carefully screened, with normal performance on WAIS Vocabulary an important criterion for selection. As shown in Table 2, there were no differences between any of the groups on this measure. Moreover, a global deficit interpretation cannot readily explain why WSC-implicit performance was selectively impaired and WFC-implicit performance was spared. Indeed, if old people were experiencing a severe and generalized cognitive decline, a double dissociation of the type observed here would be unlikely. Although there is no doubt that old people perform worse than young people on many tests of cognitive function, the present study shows that it is possible to distinguish between decline in specific functions and relate these impairments to corresponding decline in specific brain regions. To understand the nature of frontal lobe involvement in stem completion, it is necessary to compare the pattern of results on the explicit and implicit versions of stem completion. The low correlation between performance on stem-cued recall and tests of frontal function was predicted on the evidence that frontal lobe damage normally does not affect performance on explicit tests of memory that are cue-dependent and do not involve strategic processes. Frontal-lobe involvement might be expected on the explicit version of stem completion because of its conceptually driven component. The apparent inconsistency is compounded by evidence that the frontal lobes are implicated in the implicit test of stem completion, which traditionally has been viewed as relying more on automatic, perceptual processes (Blaxton, 1989; Jacoby, Lindsey, & Toth, 1992; Roediger, 1990). To resolve this apparent inconsistency, we propose that automatic and strategic processes can contribute to both conscious recollection and implicit memory. This contrasts with the traditional view that automatic processes are associated only with implicit memory and strategic processes with conscious recollection. Moscovitch (1989, 1992) has postulated that when the participant is aware of the proximal cues that are linked most directly to the memory trace, access to the memory trace is gained automatically. This automatic, cue-dependent retrieval process operates via the medial temporal lobes if the trace has been recently acquired. This interpretation accounts for the observed correlation between performance on the explicit versions of stem completion and fragment completion and other memory tests sensitive to medial temporal-, but not frontal-lobe damage. The frontal lobes are involved only if the cues are inadequate and strategic search processes are necessary to identify the effective proximal cue. With respect to implicit tests, the correlation between tests of frontal-lobe function and performance on stem completion makes the important point that despite being implicit strategic processes are involved in performing this task. In contrast, strategic processes do not contribute to performance on fragment completion. Stem and fragment completion are similar in that both include a perceptual component, and both tests require participants to produce the first word that comes to mind. However, they differ in the types of processes evoked by the separate cues. In word-stem completion, there is a strong generative component. The stem biases the participant to search the lexicon for a word beginning with the stem. In this regard, stem completion can be considered to be a variation of letter fluency, a test which is known to be sensitive to frontal-lobe dysfunction. Thus, word stems in addition to triggering perceptual identification processes, initiate generative search processes. Fragment completion does not evoke this strategy because unlike initial stems, letters drawn randomly from the word are poor cues for locating words in the lexicon. Instead, fragment completion is best viewed as a perceptual identification test in which the fragment evokes the perceptual record of the target. With respect to the idea that WSC in a priming task involves a generative search process that implicates the frontal lobes, the question arises as to whether a similar process occurs during baseline performance. Because baseline WSC performance did not correlate with FAS performance in any of the groups, that would not appear to be the case. In baseline testing, the word stem elicits a single prepotent response. Following priming, response alternatives are probably activated by the cue with the recently primed response vying with the prepotent baseline response for selection. The latter process is less automatic and more likely to involve the frontal lobes than the former. This account suggests that the frontal lobes may be involved in response-selection, as well as responsegenerative processes. The observed correlations between WSC and FAS, a generative task, on the one hand, and between WSC and WCST, a task with a significant selective component, on the other hand, is consistent with this view. Other investigators have also suggested that automatic and strategic processes contribute to performance on implicit and explicit tests of memory. The prevailing view is that no test of memory is process pure: Strategic processes intrude on performance on implicit tests of memory, and conversely automatic,

7 EXPLICIT AND IMPLICIT MEMORY 63 implicit processes influence performance on explicit tests. Acknowledging this position, we emphasize the importance of the additional point we are making. In our view, there is also an automatic component to conscious recollection that is independent of the automatic implicit component. A similar interpretation has been offered by other investigators who contrast voluntary with involuntary processes in conscious recollection (e.g., Richardson-Klavhen, Gardiner, & Java, 1994). In addition, we maintain that there is a strategic component involved in generating responses on some implicit tests that is independent of the strategic recollective process that may be operative on explicit tests. J. D. E. Gabrieli (October, 1994, personal communication) made a similar point in demonstrating a generative component to some implicit tests, like stem completion and category-exemplar generation, which is selectively impaired in Alzheimer's disease. The above framework accounts for the age-related deficits we obtained, as well as for most of the functional dissociations as revealed by different patterns of correlations among the various implicit and explicit memory tests. However, there arc some findings that still need to be explained. In the communitydwelling group, significant correlations were observed between performance on delayed recall of the Rey-Osterrieth figure and performance on explicit and implicit versions of fragment completions. In this group, significant correlations were found also between performance on explicit and implicit versions of stem completion. It is tempting to attribute the source of these correlations to perceptual components that the tests have in common. The difficulty is that this interpretation would predict other correlations that were not found. Specifically, performance on implicit and explicit tests of fragment completion did not correlate with each other, nor were similar positive correlations found in the institutionalized group. The finding that age-related deficits in WSC-priming in old people correlated with performance on frontal-lobe tests appears to conflict with reports of normal WSC performance in patients with Huntington's disease (HD) and in nondementia patients with Parkinson's disease (PD; e.g., Bondi & Kaszniak, 1991; Heindel, Salmon, Shults, Walicke, & Butters, 1989; Huberman, Moscovitch, & Freedman, 1994). Of course, such patients frequently show signs of frontal-lobe damage. However, none of these studies provided information on the relation between performance on WSC and performance on tests that are sensitive to frontal lobe function. This information is crucial, given the great variability in performance on frontal-lobe tests in these patients and in controls who themselves are usually old. It is important to bear in mind that in the present study performance on WSC was correlated with performance on frontal lobe tests in community-dwelling older adults, even though their scores on WSC were not significantly different from those obtained by young adults. Thus, the absence of a significant deficit in WSC in patients with HD or PD is not inconsistent with our hypothesis. It is expected that in these patients, as in our older adults, performance on WSC would correlate significantly with performance on frontal tests. The present results confirm previous findings that institutionalization exacerbates the process of cognitive decline associated with aging. As in our previous studies, we found substantial group differences on standard neuropsychological tests of frontal and medial temporal lobe function. Performance on experimental tests of memory was consistent with these observations, as well as with the framework we have proposed. The institutionalized group was most impaired on the explicit tests that were correlated with standard tests of medial temporal lobe function and on implicit stem completion, which was correlated with frontal function. Importantly, there were no statistical differences among the groups on implicit fragment completion that is presumed to be mediated by posterior neocortical structures that are relatively preserved with age. Although we took great care to match institutionalized and community-dwelling old people on psychological and healthrelated measures, it is clear that important differences remain between these groups. For example, our own research, has shown that the institutionalized elderly are more sensitive to psychosocial factors that can adversely affect performance on tests of medial temporal or frontal lobe function (Winocur & Moscovitch, 1987). It may also be the case that institutionalized old people who outwardly appear normal for their age may be at greater risk neurologically. Recently, Boone, Miller, and Lesser (1993) studied a group of apparently normal old people and found that 75% of participants over 70 years had detectable vascular lesions in white matter. A small group of these individuals had lesions of sufficient size to be associated with frontal cognitive deficits. It is possible that this subgroup is overrepresented in samples of normal institutionalized old people. Further research is needed to determine the full range of factors that differentiate healthy, high-functioning old people living in institutions from those living in the community. Our results have implications for issues related to aging and the organization of memory. A point to emphasize is the significance of the functional double dissociation that we obtained. To our knowledge, this is one of the first demonstrations in a single study of differential decline with age of separable memory functions mediated by different systems (see also Glisky, Polster, & Routhieaux, 1995). The finding of a double dissociation means that declines in performance with age cannot be attributed to a uniform effect across all functions, nor to a selective effect on those functions that are most vulnerable in old age. Thus, our finding of frontal lobe deficits in old people, for example, cannot be attributed to diffuse brain dysfunction, as is sometimes claimed (Goldberg & Bilder, 1987; Stuss & Gow, 1992). Double dissociation studies have a distinct advantage over single dissociation studies. By focusing only on a single function and ignoring others, the single dissociation approach can leave the impression that an observed deficit can be attributed exclusively to the function being studied. In the present study, the results show thht age differences in word stem completion can be attributed to frontal-lobe, but not medial temporal dysfunction, whereas the reverse is true of explicit tests of memory. References Albert, M. S., & Stafford, J. L. (1986). CT scan and neuropsychological relationships in aging and dementia. In G. Goldstein & R. E. Tasker (Eds.), Advances in clinical neuropsychology (Vol. 3, pp ). New York: Plenum.

8 64 WlNOCUR, MOSCOVITCH, AND STUSS Anderson, S. W., Jones, R. D., Tranel, A. P., Tranel, D., & Damasio, H. (1990). Is the Wisconsin Card Sorting Test an index of frontal lobe damage? (Abstract.) Journal of Clinical and Experimental Neuropsychology, 12, 80. Blaxton, T. A. (1989). Investigating dissociations among memory measures: Support for a transfer-appropriate processing framework. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, Bondi, M. W., & Kaszniak, A. W. (1991). Implicit and explicit memory in Alzheimer's disease and Parkinson's disease. Journal of Clinical and ExperimentalNeuropsychology, 13, Boone, K. B., Miller, B. L., & Lesser, I. M. (1993). Frontal lobe cognitive functions in aging: Methodologic considerations. Dementia, 4, Borkowski, J. G., Benton, A. L., & Spreen, O. (1967). Word fluency and brain damage. Neuropsychologia, 5, Chiarello, C., & Hoyer, W. J. (1988). Adult age differences in implicit and explicit memory: Time course and encoding effects. Psychology in Aging, 3, Coffey, C. E., Wilkinson, W. E., Parashos, I. A., Soady, S. A. R., Sullivan, R. J., Patterson, L. J., Figiel, J., Webb, G. S., Sprirzer, C. E., & Djang, W. T. (1992). Quantitative cerebral anatomy of the aging human brain: A cross-sectional study using magnetic resonance imaging. Neurology, 42, Davis, H. P., Cohen, A., Gandey, M., Colombo, P., Van Dusseldorp, G. V., Simolke, N., & Romano, J. (1990). Lexical priming deficits of the function of age. BehavioralNeuroscience, 104, Delis, D. C., Kramer, J. H., Kaplan, E., & Ober, V. A. (1987). The California Verbal Learning Test. San Antonio, TX: Psychological Corporation. Fuster, J. F. (1989). The prefrontal cortex. New York: Raven. Glisky, E. L., Polster, M. R., & Routhieaux, B. C. (1995). Double dissociation between item and source memory. Psychology and Aging, 9, Goldberg, E., & Bilder, R. M., Jr. (1987). The frontal lobes and hierarchical organization of cognitive control. In E. Perecman (Ed.), The frontal lobes revisited (pp ). New York: IRBN Press. Heindel, W. C., Salmon, D. P., Shults, C. W., Walicke, P. A., & Butters, N. (1989). Neuropsychological evidence for multiple implicit memory systems: A comparison of Alzheimer's, Huntington's, and Parkinson's patients. Journal of Neuroscience, 9, Howard, D. E. (1991). Implicit memory: An expanding picture of cognitive aging. In M. L. Howe & C. J. Brainerd (Eds.), Cognitive development in adulthood: Progress in cognitive development research (pp. 3-37). New York: Springer-Verlag. Howard, D. E., Shaw, R. J., & Heisey, J. G. (1986). Aging and the time course of semantic activation.journalofgerontology, 41, Huberman, M., Moscovitch, M., & Freedman, M. (1994). Comparison of patients with Alzheimer's and Parkinson's disease on different explicit and implicit tests of memory. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 7, Hultsch, D. F., Masson, M. E. J., & Small, B. J. (1991). Adult age differences in direct and'indirect tests of memory. Journal of Gerontology: Psychological Sciences, 46, Jacoby, L. L., Lindsey, D. S., & Toth, J. P. (1992). Unconscious influences revealed. A question of control. American Psychologist, 47, Joanette, Y., & Goulet, P. (1986). Criterion-specific reduction of verbal fluency in right brain-damaged right-handers. Neuropsychologia, 24, Keane, M. M., Gabrieli, J. D., Fennema, A. C., Growden, J. H., & Corkin, S. (1991). Evidence for a dissociation between perceptual and conceptual priming in Alzheimer's disease. Behavioral Neuroscience, 105, Light, L. L., & La Voie, D. (1993). Direct and indirect measures of memory in old age. In P. Graf & M. E. J. Masson (Eds.), Implicit memory (pp ). Hillsdale, NJ: Erlbaum. Mayes, A. R., & Gooding, P. (1989). Enhancement of word completion priming in amnesics by cueing with previously novel associates. Neuropsychologia, 27, Miceli, G., Caltagirone, C., Gainotti, G., Masullo, C., & Silveri, M. C. (1981). Neuropsychological correlates of localized cerebral lesions in non-aphasic brain-damaged patients. Journal of Clinical Neuropsychology, 3, Milner, B. (1964). Some effects of frontal lobectomy in man. In J. M. Warren & K. Akert (Eds.), The frontal granular cortex and behavior (pp ). New York: McGraw-Hill. Milner, B. (1966). Amnesia following operation on the temporal lobe. In C. W. M. Whitty & O. L. Zangwill (Eds.), Cognitive processes and the brain (pp ). London: Butterworth. Moscovitch, M. (1989). Confabulation and the frontal system: Strategic versus associative retrieval in neuropsychological theories of memory. In H. L. Roediger & F. I. M. Craik (Eds.), Varieties of memory in consciousness: Essays in honour of Endel Tulving (pp ). Hillsdale, N J: Erlbaum. Moscovitch, M. (1992). Memory and working-with-memory: A component process model based on modules and central systems. Journal of Cognitive Science, 4, Moscovitch, M., Vriezen, E., & Goshen-Gottstein, Y. (1993). Implicit tests of memory in patients with focal lesions or degenerative brain disorders. In H. Spinnler & F. Boiler (Eds.), Handbook ofneuropsychology (Vol. 8, pp ). New York: Elsevier. Moscovitch, M., & Winocur, G. (1983). Contextual cues and release from proactive inhibition in young and old people. Canadian Journal of Psychology, 37, Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, Richardson-Klavehn, A., Gardiner, J. M., & Java, R. J. (1994). Involuntary conscious memory and the method of opposition. Memory, 1, Roediger, H. L. (1990). Implicit memory: Retention without remembering.american Psychologist, 45, Rose, T. L., Yesavage, J. A., Hill, R. D., & Bower, G. H. (1986). Priming effects and recognition memory in Young and elderly adults. Experimental Aging Research, 12, Schacter, D. L. (1992). Understanding implicit memory: A cognitive neuroscience approach. American Psychologist, 47, Shimamura, A. P., Gershberg, F. B., Jurica, P. J., Mangels, J. A., & Knight, R. T. (1992). Intact implicit memory in patients with frontal lobe lesions. Neuropsychologia, 30, Squire, L. R., Shimamura, A. P., & Graf, P. (1987). Strength and duration of priming effects in normal subjects and amnesic patients. Neuropsychologia, 25, Stuss, D. T., & Gow, C. A. (1992). Frontal dysfunction after traumatic brain injury. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 5, Taylor, L. (1969). Localization of cerebral lesions by psychological testing. Clinical Neurosurgery, 16, Tomlinson, B. E. (1972). Morphological brain changes in nondemented old people. In H. M. Van Praag & A. F. Kalverbove (Eds.), Ageing of the central nervous system, biological and psychological aspects (pp ). Haarlem, The Netherlands: De Erven F. Bohn N.V. Tulving, E., & Schacter, D. L. (1990). Priming and human memory systems. Science, 247,

9 EXPLICIT AND IMPLICIT MEMORY 65 Wechsler, D. (1981). WAIS-R Manual." Wechsler Adult Intelligence Scale Revised. New York: Psychological Corporation. Winocur, G., & Moscovitch, M. (1983). Paired-associate learning in institutionalized and noninstitutionalized old people: An analysis of interference and context effects. Journal of Gerontology, 30, Winocur, G., & Moscovitch, M. (1987). An investigation of cognitive function in relation to psychosocial variables in institutionalized old people. Canadian Journal of Psychology, 41, Winocur, G., & Moscovitch, M. (1990). A comparison of cognitive function in community dwelling and institutionalized old people of normal intelligence. Canadian Journal of Psychology, 44, Received January 25, 1995 Revision received July 21, 1995 Accepted July 28, 1995