Cognitive Brain Research 17 (2003) 484–494 www.elsevier.com / locate / cogbrainres

Research report

Category-specific medial temporal lobe activation and the consolidation of semantic memory: evidence from fMRI John Kounios a , *, Phyllis Koenig b , Guila Glosser b , Chris DeVita b , Kari Dennis b , Peachie Moore b , Murray Grossman b a

b

Department of Psychology, Drexel University MS626, 245 North 15 th Street, Philadelphia, PA 19102 -1192, USA Department of Neurology-2 Gibson, Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 18194 -4283, USA Accepted 29 April 2003

Abstract Semantic memory consolidation was studied by comparing medial temporal lobe (MTL) fMRI activation to ANIMAL, IMPLEMENT and ABSTRACT nouns in healthy seniors to that of young adults. Relative to healthy seniors, young adults were predicted to show greater MTL activation for IMPLEMENTS, but not ANIMALS, because the ANIMALS category consists of highly intercorrelated and overlapping features that should require less MTL-mediated binding than IMPLEMENTS over a shorter period of time during concept consolidation. ABSTRACT meanings are context-dependent and do not consist of fixed feature sets. Thus it was predicted that ABSTRACT words would not involve age-related feature binding mediated by the MTL. These predictions were confirmed by the results. Our observations are consistent with the hypothesis that the structure of a category influences the consolidation of knowledge in semantic memory.  2003 Elsevier B.V. All rights reserved. Theme: Neural basis of behavior Topic: Neural plasticity Keywords: Semantic memory; Memory consolidation; Concepts; Categories; Category specific effects; Feature binding; Medial temporal lobe; Hippocampus; Neuroimaging

1. Introduction The temporally-graded nature of retrograde amnesia implicates the hippocampal formation in the initial acquisition of knowledge [73,82]. As this knowledge accrues in memory, it becomes consolidated over time, thereby establishing an enduring neocortical representation [48,57]. While various models of consolidation have debated the details of this process [64,65], there is considerable evidence that knowledge becomes consolidated in the cerebral cortex through a process mediated by perihippocampal structures in the medial temporal lobe (MTL) such as the subiculum, dentate gyrus, parahippocampal

*Corresponding author. Tel.: 11-215-895-1895; fax: 11-215-7628625. E-mail address: [email protected] (J. Kounios). 0926-6410 / 03 / $ – see front matter  2003 Elsevier B.V. All rights reserved. doi:10.1016 / S0926-6410(03)00164-2

cortex, perirhinal cortex, and entorhinal cortex [23]. Recent functional neuroimaging studies provide additional support for the claim that MTL structures participate in the formation of long-term memories [37,54,76]. Most of the debate about the mechanisms of consolidation has been concerned with episodic memory (i.e. memory for past experiences). In contrast, the possible consolidation of semantic memory (i.e. memory for word meanings, concepts, and knowledge about the world) has received less attention. The primary source of evidence for semantic memory consolidation is found in the patient literature. For instance, MTL lesions can result in temporally-graded retrograde amnesia for (nonautobiographical) facts, depending on the extent of the lesions [71]. This retrograde amnesia presumably results from the interruption of an ongoing consolidation process mediated by the MTL. Furthermore, hippocampal damage can impair rapid acquisition of semantic information more than it impairs

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

slow learning of semantic information over multiple presentations, suggesting that the hippocampus is particularly important to the initial acquisition of information, though eventual neocortical consolidation can apparently proceed even without hippocampal participation, albeit in a degraded fashion [42]. Additional evidence comes from Semantic Dementia patients, where imaging studies and pathologic analyses implicate damaged or dysfunctional MTL structures [16,38,63]. Patients with Semantic Dementia exhibit a semantic memory deficit in which they show more difficulty naming objects whose labels have a later age of acquisition [52]. Although other work emphasizes the contribution of autobiographical memory to apparent graded effects in semantic memory [81,89], these observations are nevertheless consistent with the notion that semantic knowledge acquired at an earlier age has had more time to be consolidated in memory before disease interrupted the consolidation process. Studies of the neuroanatomy of semantic memory have yielded evidence that different categories of knowledge are consolidated in different brain areas. Although there is some inconsistency across studies [13], patients with disease affecting left ventral temporal-occipital cortex often appear to be relatively compromised in their comprehension of natural kinds such as ANIMALS (where capital letters refer to a concept) [20,26,27,39,68,84], while disease compromising left lateral temporal-occipital and left inferior and lateral frontal cortices often appears to diminish comprehension of manufactured artifacts such as IMPLEMENTS [11,20,21,33,40,61,78,80,87]. Support for this category-specific distinction has also come from functional neuroimaging work with healthy adults. Although debate persists concerning the reliability and interpretation of such activation patterns [46], these studies generally demonstrate recruitment of ventral temporaloccipital cortex for natural kinds [12,17,20,56,67] and activation of left lateral temporal-occipital [12,20,56,59] and left inferior-lateral frontal regions [32,34,56,67] for manufactured artifacts. The mechanisms by which semantic information apparently becomes distributed in these particular brain regions remain a matter of active investigation. Based on known primary connections in the areas yielding activation, the ‘sensory-motor’ hypothesis attributes category-specific differences to the presumed perceptual and motor-related features contributing preferentially to each category of knowledge [2,56]. For example, knowledge of visual– perceptual features is hypothesized to be an important characteristic of natural kinds such as ANIMALS, because this visual–perceptual information is important for distinguishing between different types of ANIMALS; knowledge of ANIMALS is consequently associated with the visual–perceptual processing stream in ventral temporaloccipital cortex. Likewise, visual–motion and motor–action features are said to be important attributes of manufactured artifacts; knowledge of IMPLEMENTS is thus

485

related to lateral temporal-occipital [6,50,66,70] and inferior-lateral frontal regions [45,74,75] associated with visual–motion and motor–action, respectively. Recent investigations have cast some doubt on the sensory-motor account as a complete explanation of category-specific effects [13,58,85]. For example, patients with a deficit for natural kinds do not necessarily have a disproportionate impairment for processing information about visual–perceptual features compared to motor-action features [51,53,62,79]. Functional neuroimaging studies also raise questions about the sensory-motor account. For example, healthy young adults show activation in left posterolateral temporal cortex and left prefrontal cortex for both IMPLEMENTS and for ABSTRACT nouns having few sensory-motor features [5,35]. Other studies relate activation in left prefrontal cortex to the difficulty or complexity of semantic access rather than to the representation of motor features [30,35,86]. Such results suggest that the sensory–motor theory of concept representation is, at best, incomplete, and that other principles are likely involved. An alternate account of category-specific effects turns on the observation of important structural differences among categories of knowledge. It appears that naturalkinds categories such as ANIMALS consist of shared and overlapping features that are more highly intercorrelated than is the case for manufactured artifacts such as IMPLEMENTS [22,31,58,85]. For instance, if an ANIMAL has feathers, then it probably has a beak also; however, if an IMPLEMENT has a handle, this does not necessarily imply that it has a button as well. Such structural differences among categories may also contribute to the patterns of brain activation associated with specific categories of knowledge. In particular, the overlapping, intercorrelated visual features of ANIMALS may enable the neural representation of ANIMAL knowledge relatively early in the visual processing stream [35]. This hypothetical mechanism may be related to genetically determined adaptive pressures such as the recognition of potential predators or food [14]. In the following study, we test the hypothesis that differences among categories of knowledge are reflected, in part, by the mechanisms through which knowledge of these categories becomes consolidated in the brain. Toward this end, we measured patterns of brain activation in healthy seniors in response to ANIMAL, IMPLEMENT, and ABSTRACT words and compared these activation patterns with those obtained from young adults, with special attention to the MTL. Our rationale in comparing category-specific patterns of brain activation from healthy seniors to that of young adults is that consolidation, which can take place over decades [37,73], should increase with time. This increasing consolidation should result in less MTL involvement with age. However, based on considerations discussed above, there are reasons to believe that the consolidation process and the associated MTL recruitment

486

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

should also depend on the particular category of knowledge. First, words of different semantic categories typically vary in age of acquisition, a factor known to influence semantic information processing [10]. For instance, concrete words such as ANIMALS and IMPLEMENTS are acquired relatively early, while ABSTRACT nouns appear to be acquired later [28,55]. All things being equal, such concrete words should therefore consolidate at an earlier age. However, we hypothesize that a second likely influence on semantic memory consolidation may result from structural differences between the ANIMAL and IMPLEMENT categories, and that this factor is probably as potent an influence on consolidation as age of acquisition. Such concrete categories have usually been described as collections of linked features [72]. As mentioned above, previous work has shown the ANIMAL category to consist of overlapping and intercorrelated features to a greater extent than the IMPLEMENT category, implying that the links between ANIMAL features in semantic memory are generally stronger than the links between IMPLEMENT features [58,85]. One prominent view of the role of the hippocampal formation is that this structure provides temporary links between features represented in various neocortical areas, and that consolidation occurs as these initially diffuse neocortical feature representations are ‘weaned’ off of MTL-mediated feature binding as direct links develop among their constituent features [4,64]. This view could be extended to predict that categories naturally consisting of highly intercorrelated (i.e. neocortically interlinked) features should consolidate more quickly and easily than categories consisting of relatively uncorrelated features, because less MTL mediation of feature links should be necessary for high-intercorrelation categories. From this perspective, ANIMALS should generally be consolidated relatively early and easily and therefore exhibit little or no MTL activation in either young adults or seniors. By comparison, the acquisition and neural representation of IMPLEMENTS may involve relatively greater MTL activation in children or young adults than in healthy seniors. This is because it should take additional time for knowledge of IMPLEMENTS to become fully consolidated in young adults, while an additional halfcentury of consolidation is likely to reduce or virtually eliminate MTL activation for this category in seniors. Consequently, there should be differential age-related MTL activation only for IMPLEMENTS. To summarize, we predicted that ANIMALS and IMPLEMENTS would elicit little or no significant MTL activation in older subjects, because both of these categories of knowledge should be consolidated in these subjects. In contrast, younger subjects should show evidence of MTL activation relative to older subjects specifically for IMPLEMENTS because this category has a structure that may make consolidation slower or harder. ABSTRACT nouns were also included in this experi-

ment for exploratory reasons. Little is known about the neural substrates of abstract nouns that could suggest a specific prediction about their consolidation over time [5,35,41,47,49]. Abstract categories are, however, thought to be qualitatively different from concrete ones in that each abstract category consists of a system of semantic relations into which different sets of features may fit, rather than consisting of a single, relatively fixed set of linked sensory-motor features [72]. Knowledge associated with ABSTRACT nouns appears to be modulated by fluid networks of propositions that are highly context dependent [29,44,69]. According to this view, ABSTRACT nouns may exhibit less binding-or consolidation-related MTL activity at any age because of the dynamic, contextdependent nature of these words and their minimal sensory-motor content.

2. Materials and methods

2.1. Participants Participants included sixteen healthy right-handed seniors (seven females and nine males) with a mean age of 73.9 years and a mean education of 13.8 years. These seniors received a thorough medical screening to rule out conditions that could affect cognitive functioning. The comparison group of young adults consisted of sixteen right-handed native English speakers who were students at the University of Pennsylvania (analyses of data from the comparison group have been discussed elsewhere [35]). This sample, tested at the same time as the seniors, consisted of nine females and seven males, with a mean age of 23.4 years and a mean education of 16.0 years. None of the subjects were taking any sedating or centrallyacting medication at the time of testing. Structural images obtained at the time of the fMRI study for the purpose of normalization were reviewed to rule out structural brain insult. Subjects participated in an informed consent procedure approved by the IRB of the University of Pennsylvania.

2.2. Stimulus materials We presented blocks of printed words to subjects that included ANIMAL, IMPLEMENT, and ABSTRACT nouns. There were 60 words of each noun type, matched for mean frequency [IMPLEMENTS513.1, ANIMALS5 12.3, ABSTRACT514.7, F(2,177)50.18, ns] and mean letter length [IMPLEMENTS56.1, ANIMALS55.9, ABSTRACT56.3, F(2,177)50.44, ns] [25]. A cohort of 42 native English speaking undergraduates assessed the words for familiarity. All word meanings were known to all students. All but seven (3.8%) of the words were judged to be familiar by over 93% of these students. Specifically, one animal was judged familiar by only 86%

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

of the undergraduates, three abstract nouns were judged familiar by only 88%, two abstract nouns were judged familiar by only 79%, and one abstract noun was judged familiar by only 76% of the students. The words were either unambiguously nouns, or if not, a word’s frequency of occurrence as a noun was at least five times that of its verb homonym, according to form class-sensitive frequency measures [25]. To obtain age of acquisition data for our stimuli, we probed twenty young adults [eight males, twelve females; mean (6S.D.) age521.1 (64.9) years; mean (6S.D.) education514.1 (63.3) years] using a judgment technique validated by actual age of acquisition observations of children [60]. Mean (6S.D.) age of acquisition for the words in these categories is: ANIMALS557.62 (620.3) months; IMPLEMENTS5 67.71 (621.8) months; and ABSTRACT5115.32 (620.6) months (all pairwise comparisons differing significantly at the P,0.01 level, according to t tests).

2.3. Procedure To minimize the potential confounds of covarying categories of knowledge and processes that access this knowledge, we employed a relatively neutral task that encouraged uniform processing across categories. Subjects were thus asked to make ‘pleasantness’ decisions for each item [35]. This task has been used for over 30 years to probe or activate ‘deep’ or ‘semantic’ knowledge associated with words while avoiding a request for specific information [43,88]. Each stimulus word was displayed for 3 s followed by a 1-s interstimulus interval. Words were presented sequentially, blocked by type, with each 10-word block lasting 40 s. Blocks were presented in a fixed random order with no overt indication of where blocks began or ended. Pleasantness judgments were also made on two baseline blocks of pronounceable pseudoword stimuli and four blocks of filler words (verbs) which were interspersed among the noun blocks (and which were equated to the nouns for word length and, in the case of the verbs, for frequency). (Results from the verb blocks will be discussed in a separate report.) Subjects were not informed that different categories of words were being administered. Three runs containing nonrepeated stimuli were presented, each run including two blocks of each word category. Subjects indicated their judgment for each word by right- or lefthand button press for, respectively, ‘pleasant’ or ‘not pleasant’ decisions. Response type and latency were recorded by the computer presenting the stimuli. Before each run, subjects were acclimated to the MRI environment by viewing the words ‘Get Ready’ on screen for 20 s. Brief rest periods were included between runs. Our stimulus presentation system, compatible with high magnetic fields, backprojected the printed words onto a screen at the magnet bore. The subject viewed the screen through a system of mirrors. A portable computer (Macin-

487

tosh 1400C or G3) outside the magnet room used PSYSCOPE presentation software [18] to present stimuli and record response accuracies and latencies. Subjects were familiarized with the task prior to entering the magnet bore, and the task was practiced by each subject.

2.4. Image acquisition and statistical analysis Imaging was executed at 1.5 T on a GE Echospeed scanner capable of ultrafast imaging. We used a standard clinical quadrature radiofrequency head coil. Firm foam padding was used to restrict head motion. Each imaging protocol began with a 10–15 min acquisition of 5-mm thick adjacent slices for determining regional anatomy, including sagittal localizer images (TR5500 ms, TE510 ms, 1923256 matrix), T2-weighted axial images (FSE, TR52000 ms, TEeff585 ms), and T1-weighted axial images of slices used for fMRI anatomic localization (TR5600 ms, TE514 ms, 1923256 matrix). Gradient echo echoplanar images were acquired for detection of alterations of blood oxygenation accompanying increased mental activity. All images were acquired with fat saturation, a rectangular FOV of 20315 cm, flip angle of 908, 5 mm slice thickness, an effective TE of 50 ms, and a 64340 matrix, resulting in a voxel size of 3.7533.7535 mm. The echoplanar acquisitions consisted of 18 contiguous slices covering the entire brain every 2 s. To manage susceptibility artifact, a separate acquisition lasting 1–2 min was needed for phase maps to correct for distortion in echoplanar images [3]. We also inspected the raw data of individual subjects. Raw data were stored by the MRI computer on DAT tape and then processed off-line. Initial data processing was carried out with Interactive Data Language (Research Systems) on a Sun Ultra 60 workstation. Raw image data were reconstructed using a 2D FFT with a distortion correction to reduce artifact due to magnetic field inhomogeneities. Individual subject data were then prepared for pseudosubject analysis and analyzed statistically using statistical parametric mapping (SPM 99) developed by the Wellcome Department of Cognitive Neurology [24]. This system, operating on a MATLAB platform, combines raw difference images from individual subjects into a statistical t-score map. Briefly, the images in each subject’s time series were registered to the initial image in the series. The images were then aligned to a standard coordinate system [83] using the Montreal Neurological Institute template. The data were spatially smoothed with a 12-mm Gaussian kernel to account for interindividual differences in gyral anatomy and small variations in the location of activation across subjects, and low-pass filtering was implemented by controlling autocorrelation with a first-order autoregressive method. Global means were normalized by proportional scaling. The data were analyzed parametrically using t-test comparisons converted to z scores for each compared voxel. We report differences between conditions that are

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

488

Table 1 Proportion of ‘pleasant’ responses and response latencies (in milliseconds) from young adults and healthy seniors for implement, animal, and abstract nouns Category

Implements Animals Abstract

Young adults

more quickly by seniors than by young adults [t(29)52.03; P,0.05]. Young adults judged IMPLEMENTS to be less pleasant than ANIMALS [t(15)52.90; P,0.01]. By comparison, healthy seniors judged IMPLEMENTS to be more pleasant than ANIMALS [t(14)53.73; P,0.005] and ABSTRACT nouns [t(14)52.28; P,0.05]. IMPLEMENTS [t(29)5 4.46; P,0.001] and ABSTRACT nouns [t(29)53.00; P, 0.01] were judged more pleasant by seniors than young adults. These observations may suggest age-related effects on pleasantness judgments.

Healthy seniors

Judgment

Latency

Judgment

Latency

0.60 (0.21) 0.71 (0.17) 0.69 (0.14)

1324 (239) 1210 (211) 1303 (205)

0.90 (0.15) 0.80 (0.17) 0.83 (0.12)

1159 (211) 1235 (206) 1282 (175)

Data are expressed as mean (6S.D.); due to equipment malfunction, the behavioral data for one of the senior participants was lost. Consequently, the means represent 16 young adults and 15 healthy seniors.

3.2. Imaging data statistically significant at least at the P,0.05 level following correction for multiple comparisons for both the height and the extent of activation, unless otherwise stated.

Table 2 summarizes the anatomical locations and extent of the peak activations of each cluster that exceeded our statistical threshold for pleasantness judgments for each category of knowledge in healthy seniors. Images illustrating activation patterns for each category of knowledge in comparison to a pronounceable pseudoword baseline are provided in Fig. 1. As can be seen, significant activations associated with the category of IMPLEMENTS are found in left posterolateral temporal-parietal cortex, left inferior frontal cortex, and the caudate. This pattern of categoryspecific activation was evident regardless of the baseline condition (pronounceable pseudowords, or other word categories). This replicates the anatomic distribution of activation for IMPLEMENTS using the identical stimuli in young adults [35]. By comparison, activations associated with the ANIMALS-minus-pseudowords contrast recruited left ventral temporal-occipital cortex in the area of the lingual and fusiform gyri. This again replicates the

3. Results

3.1. Behavioral data Mean pleasantness ratings and mean response latencies associated with each category of words are shown in Table 1. The results yielded no evidence of age-related slowing. Young adults judged ANIMALS more quickly than IMPLEMENTS [t(15)53.35; P,0.005], and ABSTRACT nouns [t(15)52.76; P,0.05]. Healthy seniors judged IMPLEMENTS more quickly than ANIMALS [t(14)5 2.48; P,0.05] and ABSTRACT nouns [t(14)52.85; P, 0.01]. Of particular note, IMPLEMENTS were judged

Table 2 Locus and extent of peak activations during pleasantness judgments of word categories in healthy seniors Contrast

Activation locus (Brodmann area)

Coordinates x

y

Activation extent ([ voxels)

Z value

Corrected P value

z

IMPLEMENTS2 PSEUDOWORDS

Left posterolateral temporal-parietal (40, 39, 22) Left inferior frontal (44, 6) Right caudate

244 232 8

244 12 212

48 8 20

1891 a 1891 a 1891 a

6.06 4.65 4.46

0.001 0.006 0.02

IMPLEMENTS2 ANIMALS

Left posterolateral temporal (39, 22) Left inferior frontal (44, 6) Right caudate

256 232 8

240 12 216

20 12 28

535 1364 a 1364 a

4.79 4.76 5.06

0.003 0.004 0.001

IMPLEMENTS2 ABSTRACT

Left posterolateral temporal (22) Left inferior frontal (44, 6) Right caudate

252 232 8

228 12 220

16 12 24

1695 a 1695 a 1695 a

4.07 5.05 4.58

0.055 0.001 0.008

ANIMALS2

Left ventral temporal-occipital (19) PSEUDOWORDS

216

264

24

105

4.73

0.004

ANIMALS2 IMPLEMENTS

Left ventral temporal-occipital (19)

212

256

28

138

4.16

0.04

ANIMALS2 ABSTRACT

Right inferior frontal (47)

32

40

28

273

4.90

0.002

ABSTRACT2 IMPLEMENTS

Right inferior frontal (45, 44)

48

20

12

100

4.23

0.03

a

For the specified contrast, these peaks are part of the same large cluster.

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

489

Fig. 1. Activations for noun categories in comparison to a pronounceable pseudoword baseline in healthy seniors. (A) IMPLEMENTS; (B) ANIMALS.

anatomic distribution of activation seen in young adults for this category of knowledge using the identical stimuli [35]. For the ABSTRACT-minus-pseudowords contrast in healthy seniors, we observed activation that did not meet our statistical threshold in left lateral temporal-parietal cortex (x5248, y5236, z544; z score53.29, P,0.001, uncorrected) and in left occipital cortex (x528, y5268, z516; z score53.54, P,0.001, uncorrected). Nevertheless, this also replicates the anatomic distribution of activation seen with the identical stimuli in young adults [35]. Right inferior frontal activation was seen for ANIMALS and ABSTRACT nouns when other word categories were used as a baseline. Table 3 and Fig. 2 summarize direct comparisons of these normalized contrasts with the corresponding normalized contrasts in young adults. For the IMPLEMENTSminus-pseudowords contrast, young adults recruited the MTL region significantly more than healthy seniors. Greater activation that was marginally significant was

observed in left MTL in young adults compared to healthy seniors for both the IMPLEMENTS-minus-ABSTRACT contrast [x5224, y50, z5212; z score53.91, P,0.0001, uncorrected] and the IMPLEMENTS-minus-ANIMALS contrast [x5224, y50, z5212; z score53.49, P,0.001, uncorrected]. Healthy seniors, by comparison, activated the left cingulate and caudate significantly more than young adults for the IMPLEMENTS-minus-pseudowords contrast and the IMPLEMENTS-minus-ANIMALS contrast. There were no age-related differences for the contrasts involving ANIMALS and ABSTRACT nouns.

4. Discussion The present study investigated the possibility of consolidation in semantic memory paralleling current conceptions of episodic memory consolidation. Our strategy was to assess MTL activation in healthy older adults relative to

490

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

Table 3 Locus and extent of peak activations in brain regions during pleasantness judgments of word categories comparing healthy seniors and young adults Contrast

Activation locus (Brodmann area)

Coordinates x

Young adults minus healthy seniors IMPLEMENTS2 Left medial temporal-hippocampus (28) pseudowords IMPLEMENTS2 ABSTRACT

Left medial temporal-hippocampus (28)

Healthy seniors minus young adults IMPLEMENTS2 Left cingulate-caudate (24, 23) pseudowords IMPLEMENTS2 ABSTRACT

Left cingulate-caudate (24, 23)

Activation extent ([ voxels)

y

Z value

Corrected P value

z

224

24

28

94

4.20

0.03

224

0

212

54

3.91

0.07

220

212

32

344

4.08

0.04

220

28

28

385

5.00

0.001

young adults during retrieval of semantic information from different categories, with the presence of MTL activation being taken to indicate that the information being retrieved is not fully consolidated in the neocortex. Greater MTL activation was predicted for object categories in younger subjects than for older ones, because more consolidation has presumably taken place in older subjects. Moreover, based on a straightforward extension of standard consolidation theory, this effect was predicted to be categoryspecific: We predicted greater MTL activation in young adults specifically for IMPLEMENTS but not for ANIMALS or ABSTRACT nouns. This was hypothesized to have occurred because consolidation of the category of IMPLEMENTS does not benefit from the ANIMALS category’s high feature intercorrelations or overlapping of features across exemplars. The main findings can be summarized as follows. First, the present results from healthy seniors replicate previous neuroimaging findings of category-specific effects from young adults reviewed in the Introduction. Specifically, significant activations associated with the category of IMPLEMENTS in healthy seniors included left posterolateral temporal cortex and left inferior frontal cortex. By comparison, the category of ANIMALS in healthy seniors recruited left ventral temporal-occipital cortex. Second, the major prediction of greater category-specific MTL activation in young adults compared to healthy seniors was borne out. Thus, young adults had significantly greater MTL activation than healthy seniors for IMPLEMENTS but not for ANIMALS. This finding was not dependent on a particular baseline, as it was evident relative to both the pseudoword and ABSTRACT baselines. Furthermore, these activation patterns may have an analog in the behavioral results. Mean reaction time for IMPLEMENTS was faster for seniors than for young adults, while the mean reaction times for ANIMALS and ABSTRACT words did not differ significantly for seniors and young adults. This suggests that the apparently incomplete consolidation of IMPLEMENTS in young adults

resulted in a cost in speed of information processing relative to the seniors. In sum, these findings are consistent with the notions of differential consolidation of semantic categories depending on category structure and evolutionary significance. Several additional points should be noted about these results. First, the view we have developed here implies that the contrast of IMPLEMENTS minus ANIMALS should reveal MTL activation in young adults. We may not have observed this in our previous work [35] because ANIMALS may elicit some small, but statistically nonsignificant, MTL activation in young adults which obscured MTL activation for IMPLEMENTS relative to ANIMALS. Because the semantic consolidation theory tested here predicts such an effect, additional work with larger numbers of young subjects is needed to examine this point more rigorously. Second, the present results cannot be attributed to seniors generally exhibiting less brain activation during such tasks. Not only does the category-specific nature of the predicted effect argue against this interpretation, but normalization removed global activation differences between the seniors and young adults. Furthermore, the fact that seniors exhibited greater relative activation than younger subjects in the caudate and left cingulate for the IMPLEMENTS minus pseudowords and IMPLEMENTS minus ABSTRACT contrasts weighs against any explanation based on general lack of activation for seniors. The caudate plays a central role in several frontal–striatal– thalamic loops that may be implicated in age-related activation associated with IMPLEMENTS [1]. It is unclear at present whether this finding reflects actual changes in processing mechanisms or storage sites of specific types of semantic information during healthy aging, or whether it reflects compensatory activity in support of diminished semantic information processing with age. For example, recruitment of the caudate has been associated with both the representation of motor knowledge [36] and with age-related upregulation of working memory in support of

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

491

Fig. 2. Direct contrast of young adults and healthy seniors for IMPLEMENTS. (A) IMPLEMENTS-minus-pseudowords in young adults minus healthy seniors (z516); (B) IMPLEMENTS-minus-pseudowords in healthy seniors minus young adults.

language processing [19,77]. However, the fact that seniors responded to IMPLEMENTS more quickly than young adults weighs against the compensation explanation and in favor of the notion that the processing mechanisms or storage sites of IMPLEMENT information may change with age. Activation of the cingulate has been associated with executive resources such as selective attention and control over resource deployment during decision-making [7,8,15], and healthy seniors may depend more on activation of brain regions supporting these resources than young adults, especially for IMPLEMENTS, because the relatively diffuse feature structure of this category may, even with greater neocortical consolidation in seniors, require a larger share of limited processing resources.

Third, the present results are unlikely to be due to an inability by the healthy seniors to activate the MTL in response to any semantic category. Assuming that the MTL activation to IMPLEMENTS in the young adults plays some functional role (i.e. the standard assumption of neuroimaging and psychophysiological research), the fact that the seniors, who yielded little MTL activation, exhibited no deficit in behavioral performance (in fact, they responded more quickly to IMPLEMENTS than did the young adults) suggests that their relative lack of MTL activation was not due to age-related deterioration, but rather to a quantitative change corresponding to consolidation. This notion is supported by the fact that the seniors’ overall patterns of brain activation in response to the three

492

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

categories of words were generally quite similar to those of the young adults. Fourth, though our results include null findings such as the absence of significant MTL activation for the seniors, neither the present study nor the semantic memory consolidation model tested here depend on the logic of null hypothesis testing. The key finding that the young adults exhibited greater MTL activation to IMPLEMENTS than did the seniors is predicted by the semantic consolidation model; the fact that the seniors yielded a nonsignificant level of MTL activation does not negate this point. This model derives further support from the finding of faster reaction times to IMPLEMENTS in seniors than in young adults. More generally, the lack of significant MTL activation in the seniors, while a null finding, is not problematic for future tests of the theory, as the theory only predicts less MTL activation in seniors than in young adults and does not necessarily predict the absence of MTL activation (though such an absence would not be inconsistent with the theory). Nevertheless, further research testing for graded MTL effects across several age groups would provide a useful and more stringent test of the theory. Fifth, as is the case for virtually all semantic memory research, a caveat must be given concerning the stimuli used in the present study. Although the words selected from the ANIMALS and IMPLEMENTS categories were carefully equated along several dimensions, there is always the possibility that some unknown, uncontrolled stimulus factor was driving the results rather than these findings being caused by the different category structures revealed by prior research. Future studies using different stimuli will help to generalize these findings beyond the present stimulus set, thereby reducing the likelihood that ancillary factors were influencing the results. And sixth, although we cannot yet assert that the present results are generalizable beyond the pleasantness task we adopted in this study to other tasks, we believe that future efforts with other tasks will replicate our findings (for a discussion of issues related to the pleasantness task, see [35]). There is no a priori reason to believe that IMPLEMENTS are more likely to be associated with an affective connotation than animals, and while we are not aware of any evidence directly addressing the relative capacity of IMPLEMENTS and ANIMALS to adopt an affective connotation, our intuition is that ANIMALS are more likely to be associated with an affective connotation than IMPLEMENTS. Although we cannot rule out the possibility that increased MTL activation in young adults is related to category-specific difficulty in labeling IMPLEMENTS with a pleasantness value, we are not aware of any evidence consistent with this kind of effect. Finally, ABSTRACT words did not exhibit significant MTL activity, even though such words have a later age of acquisition than ANIMAL or IMPLEMENT words. Cognitive psychology and neuroscience provide little foundation on which to base relevant speculation about ABSTRACT

words beyond some theoretical and empirical evidence that ABSTRACT categories are qualitatively different from CONCRETE categories such as ANIMALS and IMPLEMENTS [9]. Assuming that the MTL mediates linkage of concept features across various neocortical areas during concept acquisition, the absence of detectable MTL activity associated with ABSTRACT words in either younger or older subjects is consistent with the observation that ABSTRACT categories differ from CONCRETE categories in some fundamental way, such as a lack of sensory–motor content and the flexible, context-dependent nature of semantic representations associated with ABSTRACT nouns [29,44,69,72]. In conclusion, the present study demonstrates categoryspecific, age-related MTL activation in semantic memory consistent with current notions of the role of the MTL in feature binding and consolidation of episodic memory. These results also suggest that knowledge structure is an important influence on consolidation. Finally, these results not only provide an important point of similarity between semantic and episodic memory, thereby suggesting common mechanisms and structure, but they highlight the fact that semantic memory is dynamic in nature, rather than being a static repository of concepts and facts.

Acknowledgements This work was supported in part by the US Public Health Service grants AG15116 and AG17586 to M.G., and DC04818 and MH57501 to J.K.

References [1] G.E. Alexander, M.D. Crutcher, M.R. DeLong, Basal ganglia– thalamocortical circuits: parallel substrates for motor, oculomotor, prefrontal and limbic functions, Prog. Brain Res. 85 (1980) 119– 146. [2] D.A. Allport, Distributed memory, modular subsystems and dysphasia, in: S.K. Newman, R. Epstein (Eds.), Current Perspectives in Dysphasia, Churchill Livingstone, Edinburgh, 1985, pp. 32–61. [3] D. Alsop, Correction of ghost artifacts and distortion in echoplanar MR with an iterative reconstruction technique, Radiology 194 (1995) 338. [4] P. Alvarez, L.R. Squire, Memory consolidation and the medial temporal lobe: A simple network model, Proc. Natl. Acad. Sci. USA 91 (1994) 7041–7045. [5] M. Beauregard, H. Chertkow, D. Bub, S. Murtha, R. Dixon, A. Evans, The neural substrate for concrete, abstract, and emotional word lexica: a positron emission tomography study, J. Cog. Neurosci. 9 (1997) 441–461. [6] E. Bonda, M. Petrides, D. Ostry, A. Evans, Specific involvement of human parietal systems and the amygdala in the perception of biological motion, J. Neurosci. 16 (1996) 3737–3744. [7] M.M. Botvinick, T.S. Braver, D.M. Barch, C.S. Carter, J.D. Cohen, Conflict monitoring and cognitive control, Psychol. Rev. 108 (2001) 624–652. [8] M.M. Botvinick, L.E. Nystrom, K. Fissell, C.S. Carter, J.D. Cohen,

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23] [24] [25] [26]

[27]

[28]

[29] [30]

Conflict monitoring versus selection-for-action in anterior cingulate cortex, Nature 402 (1999) 179–181. S.D. Breedin, E.M. Saffran, H.B. Coslett, Reversal of a concreteness effect in a patient with semantic dementia, Cognit. Neuropsychol. 11 (1995) 617–660. M. Brysbaert, I. van Wijnendaele, S. De Deyne, Age of acquisition effects in semantic processing tasks, Acta Psycholog. 104 (2000) 215–226. S.F. Cappa, M. Frugoni, P. Pasquali, D. Perani, F. Zorat, Categoryspecific naming impairment for artefacts: a new case, Neurocase 4 (1998) 391–397. S.F. Cappa, D. Perani, T. Schnur, M. Tettamanti, F. Fazio, The effects of semantic category and knowledge type on lexical-semantic access: a PET study, Neuroimage 8 (1998) 350–359. A. Caramazza, The organization of conceptual knowledge in the brain, in: M.S. Gazzaniga (Ed.), The New Cognitive Neurosciences, MIT Press, Cambridge, MA, 2000, pp. 1037–1046. A. Caramazza, J.R. Shelton, Domain-specific knowledge systems in the brain: the animate–inanimate distinction, J. Cog. Neurosci. 10 (1998) 1–34. C.S. Carter, A.W. MacDonald, M.M. Botvinick, L.L. Ross, V.A. Stenger, D. Noll, J.D. Cohen, Parsing executive processes: strategic vs. evaluative functions of the anterior cingulate cortex, Proc. Natl. Acad. Sci. USA 97 (2000) 1944–1948. D. Chan, N.C. Fox, R.I. Scahill, Patterns of temporal lobe atrophy in semantic dementia and Alzheimer’s disease, Ann. Neurol. 49 (2001) 433–440. L.L. Chao, J. Haxby, A. Martin, Attribute-based neural substrates in temporal cortex for perceiving and knowing about objects, Nature Neurosci. 2 (1999) 913–919. J.D. Cohen, B. MacWhinney, M.R. Flatt, J. Provost, PsyScope: a new graphic interactive environment for designing psychology experiments, Behav. Res. Methods Instrum. Comput. 25 (1993) 101–113. A. Cooke, C. DeVita, C. Gethers, D. Alsop, J.C. Gee, J. Detre, P. Koenig, G. Glosser, M.B. Stern, H.I. Hurtig, M. Grossman, Information processing speed during functional neuroimaging of sentence comprehension, J. Cog. Neurosci. 12 (2000) S43. H. Damasio, T.J. Grabowski, D. Tranel, R.D. Hichwa, A.R. Damasio, A neural basis for lexical retrieval, Nature 380 (1996) 499–505. E. De Renzi, F. Lucchelli, Are semantic systems separately represented in the brain? The case of living category impairment, Cortex 30 (1994) 3025. J.T. Devlin, L.M. Gonnerman, E.S. Andersen, M. Seidenberg, Category-specific semantic deficits in focal and widespread brain damage: a computational account, J. Cognit. Neurosci. 10 (1998) 77–94. H. Eichenbaum, The long and winding road to memory consolidation, Nature Neurosci. 4 (2001) 1057–1058. R.S.J. Frackowiak, K.J. Friston, C.D. Frith, R.J. Dolan, J.C. Mazziotta, Human Brain Function, Academic Press, San Diego, 1997. W.N. Francis, H. Kucera, The Frequency Analysis of English Usage, Houghton-Mifflin, Boston, 1982. G. Gainotti, M.C. Silveri, Cognitive and anatomic locus of lesion in a patient with a category-specific semantic impairment for living beings, Cognit. Neuropsychol. 13 (1996) 357–389. P. Garrard, K. Patterson, P.C. Watson, J.R. Hodges, Category specific semantic loss in dementia of Alzheimer’s type: functional–anatomic correlations from cross-sectional analyses, Brain 121 (1998) 633– 646. R. Gelman, First principles organize attention to and learning about relevant data: number and the animate–inanimate distinction as examples, Cognit. Sci. 14 (1990) 79–106. D. Gentner, Some interesting differences between verbs and nouns, Cognit. Brain Theory 4 (1981) 161–178. C. Gerlach, I. Law, A. Gade, O.B. Paulson, The role of action

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39] [40]

[41]

[42]

[43]

[44] [45]

[46]

[47]

[48] [49]

[50] [51]

[52]

[53]

493

knowledge in the comprehension of artifacts—a PET study, NeuroImage 15 (2002) 143–152. L.M. Gonnerman, E.S. Andersen, J.T. Devlin, D. Kempler, M.S. Seidenberg, Double dissociation of semantic categories in Alzheimer’s disease, Brain Lang. 57 (1997) 254–279. T.J. Grabowski, H. Damasio, A.R. Damasio, Premotor and prefrontal correlates of category-related lexical retrieval, Neuroimage 7 (1998) 232–243. N.R. Graff-Radford, A.R. Damasio, B.T. Hyman, Progressive aphasia in a patient with Pick’s disease, Neurology 40 (1990) 423–429. S.T. Grafton, L. Padiga, M.A. Arbib, G. Rizzolatti, Premotor cortex activation during observation and naming of familiar tools, Neuroimage 6 (1997) 231–236. M. Grossman, P. Koenig, C. DeVita, G. Glosser, D. Alsop, J. Detre, J.C. Gee, The neural basis for category-specific knowledge: an fMRI study, Neuroimage 15 (2002) 936–948. M. Grossman, P. Koenig, C. DeVita, G. Glosser, D. Alsop, J. Detre, J.C. Gee, The neural representation of verb meaning: an fMRI study, Human Brain Mapp. 15 (2002) 224–234. F. Haist, J.B. Gore, H. Mao, Consolidation of human memory over decades revealed by functional magnetic resonance imaging, Nature Neurosci. 4 (2001) 1139–1145. J.A. Harasty, G.M. Halliday, C. Code, W.S. Brooks, Quantification of cortical atrophy in a case of progressive fluent aphasia, Brain 119 (1996) 181–190. J. Hart, R.S. Berndt, A. Caramazza, Category specific naming deficit following cerebral infarction, Nature 316 (1985) 439–440. A.E. Hillis, A. Caramazza, Category-specific naming and comprehension impairment: a double dissociation, Brain 114 (1991) 2081–2094. P.J. Holcomb, J. Kounios, J.E. Anderson, W.C. West, Dual coding, context availability, and concreteness effects in sentence comprehension: An electrophysiological investigation, J. Exp. Psychol.: Learn. Mem. Cognit. 25 (1999) 721–742. J.S. Holdstock, A.R. Mayes, C.L. Isaac, Q. Gong, N. Roberts, Differential involvement of the hippocampus and temporal lobe cortices in rapid and slow learning of new semantic information, Neuropsychologia 40 (2002) 748–768. T.S. Hyde, J.J. Jenkins, Differential effects of incidental tasks on the organization of recall of a list of highly associated words, J. Exp. Psychol. 82 (1969) 472–481. R. Jackendoff, Semantic Structures, MIT Press, Cambridge, MA, 1990. M. Jeannerod, M.A. Arbib, G. Rizzolatti, H. Sakata, Grasping objects: the cortical mechanisms of visuomotor transformation, Trends Neurosci. 18 (1995) 314–320. J.E. Joseph, Functional neuroimaging studies of category specificity in object recognition: A critical review and meta-analysis, Cognit., Affect. Behav. Neurosci. 1 (2001) 119–136. K.A. Kiehl, P.A. Liddle, A.M. Smith, A. Mendrek, B.B. Forster, R.D. Hare, Neural pathways involved in the processing of concrete and abstract words, Human Brain Mapping 7 (1999) 225–233. B.J. Knowlton, M.S. Fanselow, The hippocampus, consolidation, and online memory, Curr. Opin. Neurobiol. 8 (1998) 293–296. J. Kounios, P.J. Holcomb, Concreteness effects in semantic processing: ERP evidence supporting dual-coding theory, J. Exp. Psychol.: Learn. Mem. Cognit. 20 (1994) 804–823. Z. Kourtzi, N. Kanwisher, Activation in human MT / MST by static images with implied motion, J. Cognit. Neurosci. 12 (2000) 48–55. M. Laiacona, E. Capitani, R. Barbarotto, Semantic category dissociations: a longitudinal study of two cases, Cortex 33 (1997) 441–461. M.A. Lambon Ralph, K.S. Graham, A.W. Ellis, J.R. Hodges, Naming in semantic dementia—what matters?, Neuropsychologia 36 (1998) 775–784. M.A. Lambon Ralph, D. Howard, G. Nightingale, A.W. Ellis, Are

494

[54]

[55] [56]

[57] [58]

[59]

[60]

[61] [62]

[63]

[64]

[65]

[66]

[67]

[68]

[69] [70]

[71] [72]

J. Kounios et al. / Cognitive Brain Research 17 (2003) 484–494 living and nonliving category-specific deficits causally linked to impaired perceptual or associative knowledge? Evidence from a category-specific double dissociation, Neurocase 4 (1998) 311–338. E.A. Maguire, C.J. Mummery, C. Buchel, Patterns of hippocampal– cortical interaction dissociate temporal lobe memory systems, Hippocampus 10 (2000) 475–482. J.M. Mandler, How to build a baby: II. Conceptual primitives, Psychol. Rev. 99 (1992) 587–604. A. Martin, C.L. Wiggs, L.G. Ungerleider, J.V. Haxby, Neural correlates of category-specific knowledge, Nature 379 (1996) 649– 652. J.L. McGaugh, Memory—a century of consolidation, Science 287 (2000) 248–251. K. McRae, V.R. de Sa, M. Seidenberg, On the nature and scope of featural representations for word meaning, J. Exp. Psychol.: Gen. 126 (1997) 99–130. C.J. Moore, C.J. Price, A functional neuroimaging study of the variables that generate category-specific object processing differences, Brain 122 (1999) 943–962. C.M. Morrison, T.D. Chappell, A.W. Ellis, Age of acquisition norms for a large set of object names and their relation to adult estimates and other variables, Quart. J. Exp. Psychol. 50A (1997) 528–559. H.E. Moss, L.K. Tyler, A progressive category-specific deficit for nonliving things, Neuropsychologia 38 (2000) 60–82. H.E. Moss, L.K. Tyler, M. Durrant-Peatfield, E.M. Bunn, ‘Two eyes of a see-through’: impaired and intact semantic knowledge in a case of selective deficit for living things, Neurocase 4 (1998) 291–310. C.J. Mummery, K. Patterson, C.J. Price, J.R. Hodges, A voxel-based morphometry study of semantic dementia: relationship between temporal lobe atrophy and semantic memory, Ann. Neurol. 47 (2000) 36–45. J.M.J. Murre, K.S. Graham, J.R. Hodges, Semantic dementia: relevance to connectionist models of long-term memory, Brain 124 (2001) 647–675. L. Nadel, A. Samsonovich, L. Ryan, M. Moscovitch, Multiple trace theory of human memory: computational, neuroimaging, and neuropsychological results, Hippocampus 10 (2001) 352–368. D.R. Patzwahl, T. Elbert, J.M. Zanker, E.O. Altenmuller, The cortical representation of object motion in man is interindividually variable, NeuroReport 7 (1996) 469–472. D. Perani, S.F. Cappa, V. Bettinardi, S. Bressi, M. Gorno-Tempini, M. Matarrese, F. Fazio, Different neural systems for the recognition of animals and manmade tools, NeuroReport 6 (1995) 1637–1641. V. Pietrini, P. Nertempi, A. Vaglia, M.G. Revello, V. Pinna, F. Ferro-Milone, Recovery from Herpes simplex encephalitis: selective impairment of specific semantic categories with neuroradiological correlation, J. Neurol. Neurosurg. Psychiatry 51 (1988) 1284–1293. S. Pinker, Learnability and Cognition: The Acquisition of Argument Structure, MIT Press, Cambridge, MA, 1989. A. Puce, T. Allison, S. Bentin, J.C. Gore, G. McCarthy, Temporal cortex activation in humans viewing eye and mouth movement, J. Neurosci. 18 (1998) 2188–2199. J.M. Reed, L.R. Squire, Retrograde amnesia for facts and events: findings from four new cases, J. Neurosci. 18 (1998) 3943–3954. B. Rehder, B.H. Ross, Abstract coherent categories, J. Exp. Psychol.: Learn. Mem. Cognit. 27 (2001) 1261–1265.

[73] N.L. Rempel-Clower, S. Zola-Morgan, L.R. Squire, D.G. Amaral, Three cases of enduring memory impairment after bilateral damage limited to the hippocampal formation, J. Neurosci. 16 (1996) 5233– 5255. [74] G. Rizzolatti, L. Fadiga, M. Matelli, V. Bettinardi, E. Paulesu, D. Perani, F. Fazio, Localization of grasp representations in humans by PET: 1. Observation versus execution, Exp. Brain Res. 111 (1996) 246–252. [75] G. Rizzolatti, L. Fadiga, M. Mattelli, V. Bettinardi, E. Paulesu, D. Perani, Localization of cortical areas responsive to the observation of hand grasping movements in humans: a PET study, Exp. Brain Res. 112 (1996) 103–111. [76] L. Ryan, L. Nadel, T. Keil, K. Putnam, D. Schayer, T. Troward, M. Moscovitch, Hippocampal activation during retrieval of remote memories, Neuroimage 11 (2000) S396. [77] B. Rypma, V. Prabhakaran, J.E. Desmond, G.H. Glover, J.D.E. Gabrieli, Load-dependent roles of frontal brain regions in the maintenance of working memory, Neuroimage 9 (1999) 216–226. [78] C. Sacchett, G.W. Humphreys, Calling a squirrel a squirrel but a canoe a wigwam: A category-specific deficit for artefactual objects and body parts, Cognit. Neuropsychol. 4 (1992) 131–185. [79] D. Samson, A. Pillon, V. De Wilde, Impaired knowledge of visual and nonvisual attributes in a patient with a semantic impairment for living entities: a case of a true category-specific deficit, Neurocase 4 (1998) 273–290. [80] M.C. Silveri, G. Gainotti, Interaction between vision and language in category-specific semantic impairment, Cognit. Neuropsychol. 5 (1988) 677–709. [81] J.S. Snowden, H.L. Griffiths, D. Neary, Semantic dementia: autobiographical contribution to preservation of meaning, Cognit. Neuropsychol. 11 (1994) 265–288. [82] L.R. Squire, F. Haist, A.P. Shimamura, The neurology of memory: Quantitative assessment of retrograde amnesia in two groups of amnesic patients, J. Neurosci. 9 (1989) 828–839. [83] J. Talairach, P. Tournoux, Co-planar Stereotaxic Atlas of the Human Brain, Thieme, New York, 1988. [84] D. Tranel, H. Damasio, A.R. Damasio, A neural basis for the retrieval of conceptual knowledge, Neuropsychologia 35 (1997) 1319–1327. [85] L.K. Tyler, H.E. Moss, M. Durrant-Peatfield, P. Levy, Conceptual structure and the structure of concepts: A distributed account of category-specific deficits, Brain Lang. 75 (2000) 195–231. [86] A.D. Wagner, E.J. Pare-Blagoev, J. Clark, R.A. Poldrack, Recovering meaning: left prefrontal cortex guides controlled semantic retrieval, Neuron 31 (2001) 329–338. [87] E.K. Warrington, T. Shallice, Category-specific semantic impairments, Brain 107 (1984) 829–854. [88] E.K. Warrington, L. Weiskrantz, A new method of testing long-term retention with special reference to amnesic patients, Nature 217 (1968) 972–974. [89] R. Westmacott, L. Leach, M. Freedman, M. Moscovitch, Different patterns of autobiographical memory loss in semantic dementia and medial temporal lobe amnesia: a challenge to consolidation theory, Neurocase 7 (2001) 37–55.