Relative category-specific preservation in semantic dementia ...

Brain & Language 124 (2013) 257–267

Contents lists available at SciVerse ScienceDirect

Brain & Language journal homepage: www.elsevier.com/locate/b&l

Relative category-specific preservation in semantic dementia? Evidence from 35 cases Catherine Merck a,⇑,1, Pierre-Yves Jonin a,1, Hélène Vichard a, Sandrine Le Moal Boursiquot a, Virginie Leblay a, Serge Belliard a,b a b

CHU Pontchaillou, Service de neurologie, CMRR, Rennes, France Inserm, Unité U1077, Caen, France

a r t i c l e

i n f o

Article history: Accepted 10 January 2013 Available online 11 February 2013 Keywords: Semantic dementia Alzheimer’s disease Category-specificity Fruit and vegetables Fusiform gyrus

a b s t r a c t Category-specific deficits have rarely been reported in semantic dementia (SD). To our knowledge, only four previous studies have documented category-specific deficits, and these have focused on the living versus non-living things contrast rather than on more fine-grained semantic categories. This study aimed to determine whether a category-specific effect could be highlighted by a semantic sorting task administered to 35 SD patients once at baseline and again after 2 years and to 10 Alzheimer’s disease patients (AD). We found a relative preservation of fruit and vegetables only in SD. This relative preservation of fruit and vegetables could be considered with regard to the importance of color knowledge in their discrimination. Indeed, color knowledge retrieval is known to depend on the left posterior fusiform gyrus which is relatively spared in SD. Finally, according to predictions of semantic memory models, our findings best fitted the Devlin and Gonnerman’s computational account. Ó 2013 Elsevier Inc. All rights reserved.

1. Introduction Semantic memory is currently defined as a system where general knowledge (or ‘‘conceptual knowledge’’) about words, living and nonliving entities, people, public events and places, is stored in the form of symbolic representations (Tulving, 1972). The notion of semantic memory first appeared in the 1960s (Quillian, 1966), in influential cognitive psychology works on artificial intelligence, with an emphasis on language abilities. Evidence from neuropsychological studies with brain-injured patients subsequently raised questions as to whether semantic memory consists of a unitary system, or whether multiple systems are required. Should it be regarded as an amodal system accessible via every input modality (Caramazza, Hillis, Rapp, & Romani, 1990; Fodor, 1983) or, on the contrary, as a multimodal system with separate verbal and visual semantic stores (Beauvois, 1982; McCarthy & Warrington, 1988)? Although the jury is still out on this last point (Gainotti, 2011, 2012), semantic storage deterioration is usually characterized by crossmodal disorders, that is, it manifests itself across different formats of stimulus presentation and response modalities (Warrington & Shallice, 1979). Considerable progress has been made in our ⇑ Corresponding author. Address: Service de neurologie, CMRR du CHU Pontchaillou, Rennes, 2 rue Henri Le Guilloux, 35 033 Rennes Cedex, France. Fax: +33 (0)2 99 28 41 32. E-mail address: [email protected] (C. Merck). 1 These authors contributed equally to this work. 0093-934X/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.bandl.2013.01.003

understanding of semantic memory, through observations of patients presenting with category-specific semantic deficits (see case review by Capitani, Laiacona, Mahon, and Caramazza (2003)). These case studies have inspired new cognitive neuropsychological accounts of semantic system organization, which can be divided into two sets. The first set considers that semantic memory is composed of multiple subsystems that are either partially or totally independent at the functional and anatomical levels. For example, according to sensory/functional theory (SFT) and its variants (Warrington & McCarthy, 1987; Warrington & Shallice, 1984), semantic knowledge about concepts is topographically organized according to the properties that are mostly distinctive for a given category or more relevant during their acquisition. Similarly, according to the domainspecific knowledge (DSK) systems hypothesis (Caramazza & Mahon, 2003; Caramazza & Shelton, 1998), the semantic system is divided into topographically organized domains of knowledge: animals, fruit/vegetables, conspecifics and, possibly, tools. The second set of accounts regards this semantic system as a unitary one, without any explicit functional or anatomical organization. Here, all the features of the different domains of knowledge are brought together within the same distributed network. The internal structure of knowledge is governed by the frequency of co-occurrence between features and the distinctiveness of those features for a given entity. In these models (computational account; Devlin, Gonnerman, Andersen, & Seidenberg, 1998; Gonnerman, Andersen, Devlin, Kempler, & Seidenberg, 1997; conceptual structure account; Tyler & Moss, 2001), concepts are therefore represented by shared or

258

C. Merck et al. / Brain & Language 124 (2013) 257–267

distinctive features. This theoretical framework was essentially derived from studies of patients with Alzheimer’s disease or semantic dementia, and was particularly inspired by the time course of these patients’ gradual loss of conceptual knowledge. Critically, those two sets of models make very different predictions. Multiple subsystems accounts assume that one subsystem breakdown will result in a category-specific deficit. By contrast, in unitary accounts, internal structure of knowledge relies upon a differential probabilistic vulnerability between features. Any category-specific deficit will therefore result from impairment of features assumed to be the most vulnerable to pathology. Semantic dementia (SD) is frequently regarded as a model of progressive semantic breakdown. It is a type of lobar degeneration characterized by the gradual loss of conceptual knowledge (Moreaud et al., 2008; Neary et al., 1998; Snowden, Goulding, & Neary, 1989; Warrington, 1975), resulting in limited vocabulary in speech (anomia), poor comprehension and deficits in the identification not just of objects and persons, in both the visual and verbal input modalities, but also of smells, tastes and sounds (Bozeat, Lambon Ralph, Patterson, Garrard, & Hodges, 2000; Luzzi et al., 2007; Snowden, Thompson, & Neary, 2004). This selective impairment in semantic memory occurs without any generalized intellectual impairment, deficit in day-to-day memory or visual perceptual abilities. Moreover, language remains fluent, free from syntactic errors, well-structured and without any phonological deficits. This syndrome arises out of temporal lobe atrophy, often bilateral but predominantly on the left side (Hodges, Patterson, Oxbury, & Funnell, 1992). Atrophy is particularly pronounced in the inferolateral areas of the anterior temporal lobes, now known to be a core region supporting semantic cognition (Binney, Embleton, Jefferies, Parker, & Ralph, 2010; Lambon Ralph, Pobric, & Jefferies, 2009; Patterson, Nestor, & Rogers, 2007; Pobric, Jefferies, & Ralph, 2010; Rogers et al., 2006; Visser, Embleton, Jefferies, Parker, & Ralph, 2010; Visser, Jefferies, & Lambon Ralph, 2010). Category-specific deficits have rarely been reported in SD cases. However, tasks involving naming to description, description-topicture matching and verbal definitions yielded a disproportionate breakdown for sensory/perceptual features compared with functional/associative features (Lambon Ralph, Patterson, Garrard, & Hodges, 2003). Nevertheless, four SD cases with a category-specific semantic deficit have been documented in the literature. In only one of these cases was there evidence of better preservation for living categories (Patient IW; Lambon Ralph, Howard, Nightingale, & Ellis, 1998), with relatively poor performance on perceptual attributes. By contrast, the other three SD cases (MF; Barbarotto, Capitani, Spinnler, & Trivelli, 1995; LI; Zannino et al., 2006, and KH; Lambon Ralph et al., 2003) exhibited the reverse pattern of deficits, with poorer performance for living things. While perceptual features were more impaired than functional ones for KH and LI, no significant difference was found between the two for MF. Temporal lobe atrophy was bilateral, but more pronounced on the right side for MF and KH, whereas LI’s temporal lobe atrophy was more

pronounced on the left side, with atypical diffuse atrophy extending to the parietal regions. By contrast, IW’s temporal lobe atrophy was left-sided only, with an intact right temporal lobe. It is important to emphasize that both Barbarotto et al. (1995) and Lambon Ralph et al. (2003) described a domain-specific deficit, focusing their analysis on living versus nonliving entities rather than a discrete category-specific one (i.e., between animals versus vegetables knowledge for example). We could highlight another critical point from those previous studies. Given that SD patients typically present with expressive and receptive language complaints (Belliard et al., 2007), it may be unwise to assess semantic knowledge with tasks that rely on verbal expressive behavior. Despite this, of the five tasks making up Lambon Ralph et al.’s (2003) semantic assessment, three necessitated verbal output. Similarly, their assessment of semantic attribute knowledge (i.e., functional or sensory) required participants to understand short sentences describing an item, then either to name or point to the item in question. Therefore, in our opinion, the possibility of a category effect on the SD patients’ performance cannot be ruled out. We are not aware of any previous group study that has specifically investigated the effect of category on semantic knowledge in SD, despite the fact that contemporary cognitive accounts of semantic knowledge organization are largely based on categoryspecific effects. The aim of this study was thus to determine whether a category-specific semantic effect could be highlighted by means of a semantic sorting task administered to a large cohort of SD patients.

2. Methods 2.1. Participants 2.1.1. Semantic dementia Between 1991 and 2007, 55 patients who fulfilled the diagnostic criteria for SD (Neary et al., 1998) were followed up at the memory clinic of Rennes University Hospital (Belliard, Merck, Jonin, Lemoal, & Vercelletto, 2011). All of them presented with the typical clinical features of SD: a history of complaints about worsening comprehension deficits, anomia, and difficulty identifying objects and/or persons, reflecting a predominant and distressing loss of conceptual knowledge, contrasting with the relative preservation of day-to-day memory and perceptual abilities. Speech was still fluent, without any phonological or syntactic errors. Of these 55 patients, 35 were administered a complete neuropsychological battery within 3 months of diagnosis (see Table 1), together with a 64-item semantic sorting task. The results of the background neuropsychological assessment are summarized in Table 2. Many of them (thirteen patients) performed below normal levels on the Mini Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975), essentially due to comprehension

Table 1 Demographic and clinical features. Semantic dementia (n = 35)

Alzheimer’s disease (n = 10)

Controls (n = 12)

Mean (SD)

Range

Mean (SD)

Range

Mean (SD)

Range

Age Sex (M:F) Education (in years) Illness duration (in months) Side of atrophy (Left:bilateral:right)

62.7 (6.4) 25:10 9.5 (3) 36 (19.6) 15:17:3

48–73

74 (7.9) 3:7 10.7 (3.7) 30.7 (13.6)

59–87

62.8 (4.5) 5:7 11 (4.7)

55–70

Severity of atrophy Left side Right side

2.1 (0.65) 1.36 (1.07)

0.5–2 0–1

7–18 12–96

7–20 12–60

7–17

259

C. Merck et al. / Brain & Language 124 (2013) 257–267 Table 2 Background neuropsychology. Tests

Maximum score

Controls

Semantic dementia

Alzheimer’s disease

Mean (SD) or cut-off (at 5%)

Mean (SD)

Range

Mean (SD)

Range

21.6 (4.3)

16– 28

72 (5.8)

63– 79

MMSE

30

29.7 (0.5)

25.2 (3.4)

19–30

Intellectual efficiency–Raven’s Colored Progressive Matrices

36

15

28.3 (6.2)

15–36

50 10 10 36

16 7 4 5

33.3 (10.2) 9.1 (1.8) 7.3 (2.7) 11.6 (7.3)

15–50 2–10 3–10 0–20.5

9 8

4 3

5.6 (1.1) 4.3 (1.1)

4–9 2–7

Attentional and executive functions Digit Symbol (WAIS-R, standard score) Trail Making Test, Part A (s) Trail Making Test, Part B (s)

19 – –

5 77 180

3–15 25–427 60–349

MCST-categories completed MCST-perseverative errors Stroop Interference

– – –

3 9

8 (3.3) 77.1 (67.8) 175.3 (114.7) 4.4 (1.9) 3.6 (4.3) 7.72 (10.22)

36 18 18 10

29 17 17 9

29.9 (5.05) 14 (3.7) 17.7 (0.8) 9.7 (0.5)

13–35 2–18 14–18 8–10

80

69

35 (19.1)

10–66

54

38

44.2 (4.6)

35–53

10 36 36

8 30 29

9.7 (0.6) 34.5 (3.2) 33.7 (2.9)

8–10 19–36 22–36

Nonverbal episodic memory ‘‘La Ruche’’, a visuospatial learning task Immediate free recall – sum of the 5 trials Immediate forced – choice recognition Delayed free recall Delayed recall of Rey Osterrieth complex figure –form A Working memory Digit span forward (WAIS-R) Digit span backward (WAIS-R)

Language Oral comprehension, token test (short form) Irregular word reading test Regular word reading test Isolated word repetition test (Boston Diagnostic Aphasia Examination) Naming task (DO 80) Visual perceptual abilities Benton Facial Recognition Test Protocole d’Evaluation des Gnosies Visuelles (PEGV) Identical figure matching task Embedded figure task Copy of Rey-Osterrieth complex figure

deficits (errors on ‘‘language items’’ and on ‘‘orientation items’’ as season, state, city and hospital, due to their lexical-semantic impairment). Despite their spared day-to-day memory, ten of them presented with a deficit on nonverbal memory tasks: ‘‘La Ruche’’, a visuospatial learning task (Violon & Wijns, 1984) and the delayed recall of the Rey-Osterrieth Complex Figure Test – form A (ROCF; Osterrieth, 1944). Naming abilities, as measured by an oral picture-naming task (‘‘Test de dénomination orale d’images’’, Do 80; Deloche & Hannequin, 1997) were severely compromised, mainly because of the underlying semantic disruption, and irregular word reading was impaired as well. Only seven patients scored below the normal range on language tasks that did not critically depend on semantic object knowledge, namely the Token Test (De Renzi & Faglioni, 1978), regular word reading and isolated word repetition (Boston Diagnostic Aphasia Examination; Goodglass & Kaplan, 1972). Apart from five cases, our SD patients performed normally on all four visuospatial tasks, namely the copy condition of the Rey-Osterrieth complex figure, (Osterrieth, 1944), the ‘‘Identical figure matching’’ and ‘‘Embedded figures’’ subtests of the ‘‘Protocole d’Evaluation des Gnosies Visuelles’’ (PEGV; Agniel, Joanette, Doyon, & Duchein, 1992), and the Benton Facial Recognition Test (Benton, Sivan, Hamsher, Varney, & Spreen, 1994). Finally, 23 patients were impaired on at least one of the six attentional and executive tasks, namely the Digit Span and Digit Symbol subtests of the WAIS-R (Wechsler, 1981), the Trail Making Test (Reitan, 1958), Raven’s Colored Progressive Matrices (Raven, Raven, & Court, 1998),

18

0–6 0–16 ( 9)– (+41)

the Modified Card Sorting Test (MCST; Nelson, 1976), and the Stroop Test (1935). The neurological examinations were normal. Neuroimaging (CT and/or MRI scans) consistently revealed atrophy, predominantly in the temporal lobes, bilateral in 17/35 patients, more pronounced on the left side for 15/35 and on the right side for 3/35. The temporal lobe atrophy on the CT and/or MRI scans was independently scored by a neurologist and a radiologist on a Likert-type scale, ranging from ‘‘0 = no atrophy’’ to ‘‘3 = severe atrophy’’, for each hemisphere (see Table 1). These patients underwent an annual clinical follow-up. Two years after diagnosis, 21/35 SD participants were administered the same neuropsychological battery, including the experimental sorting task, as part of this follow-up.

2.1.2. Alzheimer’s disease To check whether any category-specific effect we could observe in SD reflected a genuine feature of this disease, 10 patients in the mild to moderate stages of Alzheimer’s disease (AD) were also included in this study (see Table 1). All fulfilled the National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association (NINCDS–ADRDA) criteria (McKhann et al., 1984). The patients’ complaints and cognitive profiles were dominated by episodic memory disorders.

260


We recruited the AD participants in order to contrast their performance on the sorting task with that of the SD group, and thus to estimate the specificity of the SD pattern of performance. The patients in the AD and SD groups were matched for years of education (U = 125.5; Z = 1.315; p = 0.216), but not for sex (v2(1) = 5.679; p = 0.0172) or age (U = 39.5; Z = 3.705; p < 0.001). At a cognitive level, the MMSE global score was significantly lower for AD participants than for SD patients (U = 61, Z = 2.241; p = 0.025). By contrast, the AD participants scored higher on the DO 80 naming task (U = 5, Z = 4.557; p < 0.001), suggesting that their lexical-semantic abilities were less impaired (see Table 2). 2.1.3. Controls Twelve healthy control participants (5 men and 7 women) free from any history of neurological or psychiatric disorders and drug or alcohol abuse (see Table 1) were recruited. All were matched with the SD patients for age (U = 174.5, Z = 0.014; p = 0.989), sex (v2(1) = 3.428; p = 0.0641) and years of education (U = 95, Z = 0.899; p = 0.368). 2.2. Stimuli and procedure The semantic sorting task consisted of 64 stimuli divided in four categories (animals, fruit and vegetables, tools and kitchenware), selected from the 480 words and pictures of the ‘‘Imagier du Père Castor’’ playing cards (Flammarion) (see Table in supplementary Material). Items were matched between categories for lexical frequency (Lexique database; New, Pallier, Brysbaert, & Ferrand, 2004): oral lexical frequency, one-way ANOVA: F(3, 60) = 1.855, p = 0.147; written word lexical frequency: F(3, 60) = 1.696, p = 0.177. Items were also matched between categories for number of letters: F(3, 60) = 1.480, p = 0.229; number of phonemes: F(3, 60) = 1.815, p = 0.154; and number of syllables: v2(9) = 10.28, p = 0.328. The procedure of testing was as follows: the same 64 items were presented first as words, then as pictures, in a predetermined order. The task consisted in sorting the stimuli at both superordinate and subordinate levels. At the superordinate level, participants were instructed to sort the stimuli into the four superordinate semantic categories labelled as follows: ‘‘animals’’, ‘‘food’’ (for fruit and vegetables), ‘‘tools’’ and ‘‘kitchenware’’. At the subordinate level, participants had to sort words into two functional attributes, then into two perceptual features (see Table in supplementary Material for functional and perceptual attributes labels). Still at the subordinate level, subjects were asked to sort pictures into two functional attributes only, since perceptual features are obvious from visual input (pictures) modality. Thus, participants firstly had to sort 64 words into four superordinate categories, then into two functional subordinate attributes and finally into two perceptual subordinate features. Secondly, they had to sort the same 64 items presented as pictures into four superordinate categories and into two functional subordinate attributes. Whenever participants did not understand the meaning of the verbal labels used in the task (e.g., for the ‘‘Functional’’ sorting: ‘‘Is the whale a wild or a domestic animal?’’), the examiner explained the labels until the participants understood them. 2.3. Statistical analysis Intergroup and intragroup analyses were performed using SPSS Statistics 12.0, focusing on category, type of feature, and temporal lobe atrophy variables. For all ANOVAs, homogeneity of covariance was tested with Mauchly’s test of sphericity, and the Greenhouse– Geisser correction applied where appropriate (i.e., whenever Mauchly’s test was significant at the 0.05 threshold). The statistical

level of significance was set at 0.05. Moreover, whenever the significance level was close to the 5% level, analyses were replicated using nonparametric methods. Lastly, the Bonferroni correction was applied for multiple comparisons (corrected alpha level for 6 comparisons, p = 0.008). 3. Results 3.1. Controls Control participants performed at a close-to-ceiling level in almost all conditions. Notwithstanding, a better range of scores was observed for the animals category, with poorer performances for the kitchenware and tools categories. Fruit and vegetables yielded intermediate levels of performance. A within-participants ANOVA yielded an effect of category, but only for the ‘‘total’’ variable (i.e. the sum of all sorting scores) (Total: F(3, 33) = 3.145; p = 0.038; Superordinate-words: F(3, 33) = 1.658; p = 0.220; Superordinate-pictures: n/a ; Functional-words: F(3, 33) = 0.240; p = 0.868 ; Functional-pictures: F(3, 33) = 3.022; p = 0.082 ; Perceptual-words: F(3, 33) = 3.319; p = 0.058). Planned comparisons failed to reveal any significant difference between categories, after the Bonferroni correction (Animals versus fruit and vegetables: t(11) = 0; p = 1; Animals versus tools: t(11) = 2.031; p = 0.067; Animals versus kitchenware: t(11) = 2.493; p = 0.03 ; Fruit and vegetables versus tools: t(11) = 1.864; p = 0.089; Fruit and vegetables versus kitchenware: t(11) = 1.780; p = 0.103; Tools versus kitchenware: t(11) = 0.233; p = 0.820). These data suggested the absence of any effect of category on sorting performance in control participants (see Table 3 and Fig. 1). We then performed a repeated measure ANOVA: [kinds of subordinate sorting level X semantic categories]. No significant interaction was reached for control participants (F(6, 66) = 2.222; p = 0.122), suggesting the absence of any category effect whatever subordinate sorting conditions considered. Moreover, a non-parametric analysis (Friedman ANOVA) failed to reveal any significant difference between the three subordinate sorting levels for the fruit and vegetables category (v2(2) = 1.333; N = 12; p = 0.513). This result therefore seemed to rule out any potential bias of the task concerning heterogeneous complexity of the three different subordinate sorting conditions for the fruit and vegetables category. 3.2. Semantic dementia 3.2.1. Semantic dementia participants: baseline 3.2.1.1. Performances on the semantic sorting task. The participants with SD performed more poorly on all conditions and categories than controls (t-tests, p < 0.01 for all comparisons). The SD patients performed better with visual (pictures) than verbal (words) presentation, t(34) = 7.302; p < 0.0001. The SD patients also performed better at the superordinate level of sorting than at the subordinate level, t(34) = 9.265; p < 0.0001. Although the analysis of the SD participants’ performance at the subordinate level revealed greater impairment for perceptual features than for functional ones, t(34) = 3.629; p = 0.001, this significant difference could be attributed to a single category (tools: t(34) = 3.925; p < 0.001), as comparisons between features remained nonsignificant for the other categories. Within the SD group, repeated-measures ANOVAs yielded an effect of category only for the ‘‘total’’ variable, the ‘‘functionalpictures’’ and ‘‘perceptual-words’’ conditions (Total: F(3,102) = 6.419; p = 0.001; Superordinate-words: F(3, 102) = 2.394; p = 0.073; Superordinate-pictures: F(3, 102) = 0.229; p = 0.733; Functional-words: F(3, 102) = 1.753; p = 0.161 ; Functional-pictures: F(3, 102) = 3.794; p = 0.013; Perceptual-words: F(3, 102) = 9.045; p < 0.0001). Planned

NS

NS NS

NS

87.5–100 25–100 43.75–100 63–100 68.75–100 0–94 25–100 69–94

Range

97 87 85.42 84.4 93.8 69.3 64.29 85.6

(SD) Mean

(2.43) (12.04) (18.17) (13.44) (8.57) (17.71) (22.06) (10.72)

93.75–100 56–100 37.5–100 56–100 68.75–100 25–100 0–93.75 63–94 Perceptual words

Mean

99 85.2 77.38 86.3 91.7 67 55.06 78.1 93.75–100 38–100 50–100 50–100 87.5–100 38–100 25–100 38–100

Range (SD)

(2.43) (13.24) (12.6) (15.08) (4.18) (16.24) (21.4) (20.62) 99 90.9 88.69 91.3 97.4 80.5 67.86 81.3

Mean

93.75–100 38–100 43.75–100 69–94 93.75–100 25–100 25–93.75 50–100

Range (SD)

(2.43) (18.05) (14.66) (8.84) (3.22) (19.28) (19.78) (17.29)

Mean

99 80 75 81.3 97.4 73 60.71 80.6 Controls SD Baseline SD Follow up AD Controls SD Baseline SD Follow up AD Functional pictures

(SD)

Range

(4.18) (17.04) (14.97) (11.88) (8.84) (19.20) (19.78) (7.82)

P

Category effect

Kitchenware Tools Fruit and vegetables Animals

Semantic categories

Table 3 Performances on semantic sorting task for controls, SD patients (at baseline and follow up) and for AD patients. Performances were expressed in mean correct percent (SD: Standard Deviation) and range scores. Level of significance of the category-specific effect: NS (Nonsignificant effect; p > 0.05); (significant effect; p < 0.05); (significant effect; p < 0.001).


261

comparisons between categories revealed the selective preservation of fruit and vegetables compared with the other three categories in the ‘‘perceptual-words’’ condition, after the Bonferroni correction (Fruit and vegetables versus animals: t(34) = 2.859; p = 0.007; Fruit and vegetables versus tools: t(34) = 5.182; p < 0.0001; fruit and vegetables versus kitchenware: t(34) = 3.839; p = 0.001; animals versus tools: t(34) = 2.578; p = 0.014; Animals versus kitchenware: t(34) = 1.182; p = 0.245; Tools versus kitchenware: t(34) = 0.780; p = 0.441). In the ‘‘functional-pictures’’ condition, fruit and vegetables were better preserved than animals only, after the Bonferroni correction (Fruit and vegetables versus animals: t(34) = 3.191; p = 0.003; Fruit and vegetables versus tools: t(34) = 2.173; p = 0.037 ; Fruit and vegetables versus kitchenware: t(34) = 1.371; p = 0.179 ; Animals versus tools: t(34) = 1.370; p = 0.180; Animals versus kitchenware: t(34) = 2.174; p = 0.037; Tools versus kitchenware: t(34) = 0.524; p = 0.603). (see Table 3 and Fig. 1). We therefore focused our subsequent analyses on these two sorting conditions. 3.2.1.2. Effects of confounding variables. We then sought to verify whether this relative preservation of fruit and vegetables could be modulated by the items’ lexical frequency, as the latter’s impact on word identification is well documented (Lambon Ralph, Graham, Ellis, & Hodges, 1998). A split-plot ANOVA failed to demonstrate any significant interaction between category and lexical frequency, in either the ‘‘functional-pictures’’ or ‘‘perceptualwords’’ conditions, F(3, 102) = 0.047; p = 0.986; F(3, 102) = 0.732; p = 0.535. Furthermore, we controlled for the potential effects of the following demographic and disease-related variables: (1) age at onset; (2) disease duration; (3) severity of temporal lobe atrophy; (4) lexical-semantic impairment (approximated by contrasting naming scores: three subgroups of patients were defined according to their overall naming score); (5) sex, since Gainotti (2010) has suggested that sex may play a role in category-specific deficits following brain lesions. We therefore performed split-plot ANOVAs with these different variables together with the category factor, to look for any significant interactions. The apparent preservation of fruit and vegetables, compared with the other categories, depended neither on age at onset (‘‘functional-pictures’’: F(3, 99) = 1.607; p = 0.193; ‘‘perceptual-words’’: F(3, 99) = 1.372; p = 0.256), nor on disease duration (‘‘functional-pictures’’: F(3, 99) = 0.760; p = 0.519; ‘‘perceptual-words’’: F(3, 99) = 0.326; p = 0.806), on severity of temporal lobe atrophy (‘‘functional-pictures’’: F(3, 90) = 0.429; p = 0.732; ‘‘perceptual-words’’, F(3, 90) = 1.801; p = 0.153), on lexical-semantic impairment (‘‘functional-pictures’’: F(6, 84) = 0.634; p = 0.703; ‘‘perceptual-words’’: F(6, 84) = 1.349; p = 0.245), nor on sex (‘‘functional-pictures’’: F(3, 99) = 0.886; p = 0.451; ‘‘perceptual-words’’: F(3, 99) = 1.851; p = 0.143). Taken together, these results indicate that the preservation of the fruit and vegetables category in SD participants persisted after the confounding variables analyses for both functional and perceptual features. In the functional condition, patients exhibited better preservation of fruit and vegetables than of animals, whereas in the perceptual condition, fruit and vegetables were better preserved than all the other categories. It should be noted that no such effect was observed in controls, although the close-to-ceiling effect prevented us from generalizing (see Laws, Gale, Leeson, & Crawford, 2005). 3.2.1.3. Intra-individuals analysis. We therefore decided to investigate the robustness of this preservation effect at the individual (i.e., single patient) level. Would this effect be present in only a small number of patients, thus accounting for a possible group effect in the absence of a genuine category-specific effect?

262


Fig. 1. Performances on the two critical subordinate sorting conditions (‘‘functional-pictures’’ and ‘‘perceptual-words’’) for controls, SD and AD patients. Mean performance were expressed as percent correct (%) for the four semantic categories (animals, fruit and vegetables, tools and kitchenware). Stars pointed out a significant category-specific effect (p < 0.05). Horizontal drawbar indicated the significant planned comparisons (paired t-tests; p < 0.05).

For each patient, and whenever the post hoc analysis yielded a significant difference between two categories, we computed the difference in sorting scores (e.g., for functional features [fruit and vegetables score – animals score]). This allowed us to divide the SD patients into two subgroups: (1) ‘‘fruit and vegetables +’’, when the difference in performance between the fruit and vegetables and target categories (i.e., animal for functional properties; animals, tools and kitchenware for perceptual properties) was positive; (2) ‘‘fruit and vegetables ’’, when the difference in performance between the fruit and vegetables and target categories was either negative or equal to zero. For functional and perceptual features, almost two thirds of the patients displayed better sorting performance for fruit and vegetables. For perceptual features, a very high proportion of participants yielded a positive difference between fruits and vegetables and tools/kitchenware, whereas the same was observed in only a half of the participants when fruits and vegetables were compared with animals. (‘‘Perceptual-words’’: 49% of participants with better performance for fruit and vegetables than for animals; ‘‘Perceptual-words’’: 74% for fruit and vegetables versus tools; ‘‘Perceptual-words’’: 74% for fruit and vegetables versus kitchenware; ‘‘Functional-pictures’’: 63% for fruit and vegetables versus animals). The question then arose as to whether these two subgroups of SD patients (with or without fruit and vegetables preservation effect) differed according to clinical or neuropsychological variables. For functional features, we failed to find any significant difference between patients presenting with the dominant ‘‘fruit and vegetables-preserved effect’’ and other patients, regarding age at onset (t(33) = 0.176; p = 0.862), disease duration (t(33) = 1.337; p = 0.191), educational level (t(32) = 1.449; p = 0.165), lexicalsemantic impairment (t(29) = 1.634; p = 0.113), severity of temporal lobe atrophy (t(30) = 1.188; p = 0.244), or sex (Fisher’s exact test, p = 0.709). The same results were found for perceptual features, all ps > 0.05 (age at onset: t(33) = 0.624; p = 0.537; disease duration: t(33) = 0.678; p = 0.507; educational level: t(32) = 0.522; p = 0.605; lexical-semantic impairment: t(29) = 0.007; p = 0.994, severity of temporal lobe atrophy: t(30) = 0.394; p = 0.696), and only a trend toward significance for sex (Fisher’s exact test, p = 0.059). Furthermore, no significant difference was found between participants presenting left-predominant atrophy and those with bilateral atrophy (‘‘Functional-pictures’’: Fisher’s exact test, p = 0.712. ‘‘Perceptual-words’’: Fisher’s exact test, p = 0.166). Given the small number of participants presenting with right-predominant atrophy (n = 3), we were not able to statistically control for the effect of this side of atrophy. In short, data from the SD participants at diagnosis (baseline) provided some evidence for a robust category-specific effect in

the ‘‘functional-pictures’’ and ‘‘perceptual-words’’ conditions of our semantic sorting task. This effect corresponded to the relative preservation of fruit and vegetables, especially when compared with animals for functional features and with the other three categories for perceptual features. 3.2.2. Semantic dementia participants: follow-up 3.2.2.1. Performances on the semantic sorting task at follow-up. The semantic sorting task was administrated to 21 patients at 24months follow-up. We therefore replicated the same analysis, in order to determine whether the category specific preservation observed at baseline remains at follow-up. Comparisons between performances at baseline and at followup revealed a significant decrease at all levels of analysis (paired t-tests, all ps < 0.001), consistent with the degenerative nature of the disease. Repeated-measures ANOVAs again revealed a significant effect of category (i.e., fruit and vegetables preservation) for both ‘‘functional-pictures’’ (F(3, 60) = 6.671; p = 0.002) and ‘‘perceptualwords’’ (F(3, 60) = 3.542; p = 0.02). Planned comparisons (paired t-tests) highlighted significant differences for the following pairs: fruit and vegetables versus animals (t = 4.806; p < 0.001) and fruit and vegetables versus tools (t = 4.017; p = 0.001) for ‘‘functional pictures’’, and fruit and vegetables versus tools (t = 3.784; p = 0.001) for ‘‘perceptual-words’’. The [size of category effect time of assessment] interaction was not significant, irrespective of feature type (‘‘functional-pictures’’: F(3, 60) = 0.870; p = 0.417. ‘‘perceptual-words’’: F(3, 60) = 0.742; p = 0.531). Thus, the magnitude of this fruit and vegetables preservation effect remained comparable between baseline and follow-up at 24 months (see Table 3 and Fig. 2). 3.2.2.2. Effects of confounding variables. This consistent categoryspecific effect after 2 years was still not modulated by the degree of lexical-semantic impairment (here, two subgroups of patients were defined according to their overall naming score). The [category naming score] interaction was not significant for either ‘‘functional-pictures’’ (split-plot ANOVA, with category as withinparticipants factor and naming subgroups as between-participants factor, F(3, 48) = 1.842; p = 0.170) or ‘‘perceptual-words’’, F(3, 48) = 0.023; p = 0.995. 3.2.2.3. Intra-individuals analysis. We then calculated the percentage of patients who continued to display a positive difference in performance between fruit and vegetables and the target categories 24 months later, and the percentage of patients who presented


263

Fig. 2. Evolution of SD patients’ performances between baseline and follow up for the ‘‘functional-pictures’’ and ‘‘perceptual-words’’ sorting conditions. Mean performance expressed as percent correct (%).

a positive difference at follow-up but had not done so at baseline (Fig. 3). For ‘‘functional-pictures’’, 92% (11/12) of SD participants maintained a positive difference. Of those participants who did not present a positive difference at baseline, 67% (6/9) presented one at follow-up. For ‘‘perceptual-words’’, 75% (12/16) of SD participants maintained a positive difference in performance between fruit and vegetables versus tools. It is noteworthy that 80% (4/5) of the participants who did not present this effect at baseline did display relative preservation of fruit and vegetables at follow-up. Briefly, the results of this intraindividual analysis at follow-up emphasized the robustness of our finding of a category-specific preservation. SD participants presenting with this effect at baseline massively presented with the same effect after 2 years (more than 80%). At the same time, depending on the type of feature considered, a large proportion of participants presented with this effect at follow-up even though they did not at baseline. 3.3. Alzheimer’s disease The AD participants’ performances did not suggest any ceiling effect, as the percentage of correct sorting generally ranged between 57.25% and 97.75%. Nonetheless, unlike the pattern of performance found in the SD group, the AD patients did not exhibit any category-specific effects (e.g. ‘‘functional-pictures’’: F(3, 27) = 2.388; p = 0.091; ‘‘Perceptual-words’’: F(3, 27) = 0.546; p = 0.655) (see Fig. 1).

It is worth noting that the two patients groups did not differ regarding their performance on the four semantic categories at ‘‘functional-pictures’’ and ‘‘perceptual-words’’ levels, except for kitchenware in the ‘‘perceptual-words’’ condition (nonparametric Mann Whitney test, U = 76.5, Z = 2.717, p = 0.006). Beyond an apparent difference between AD and SD participants’ lexicalsemantic abilities as could be inferred based on ‘‘DO 80’’ performances, both groups virtually presented with similar performances at the three subordinate sorting conditions (see Table 3). 4. Discussion The aim of our study was to explore whether any effect of category could influence semantic knowledge in SD. Indeed, our results suggest that fruit and vegetables category may be selectively resistant to the massive semantic disruption that occurs in SD. Thereafter we will comment methodological issues regarding our main finding of a category-specific preservation in SD. We then will look at how this finding can be related to neuroanatomical factors and finally discuss the relevance of the principal models of semantic system organization. 4.1. Relative preservation of fruit and vegetables in semantic dementia We administered a semantic sorting task to 35 patients with SD, once at baseline (i.e., within 3 months of diagnosis), and again after 2 years. Above and beyond the expected global impairment for all

Fig. 3. Changes in ‘‘Fruit and vegetables effect’’ distribution (expressed as %) among SD participants between baseline and follow-up, on ‘‘functional-pictures’’ condition (comparison between fruit and vegetables versus animals) and on ‘‘perceptual-words’’ one (comparison between fruit and vegetables versus tools). Fruit and vegetables + = proportion of patients with a selective preservation of fruit and vegetables. Fruit and vegetables = proportion of patients with a selective deficit for fruit and vegetables or a difference between performances on fruit and vegetables and target categories equal to zero.

264


conditions and categories, we found a main effect of category on performance. This effect consisted of the relative preservation of fruit and vegetables compared with the animals, tools and kitchenware categories. Moreover, we failed to demonstrate any category effect in the participants with Alzheimer’s disease, which we believe strengthens our main finding. One could argue that a possible category-effect in more closely matched AD patients cannot be ruled out since our AD and SD participants are not matched for the severity of their lexical-semantic deficit. However, we demonstrated that AD patients performed similarly to SD participants at the three subordinate conditions of the semantic sorting task. Furthermore, since AD is characterized by progressive cognitive impairments in almost all domains, contrasting with the relatively focal pattern of deficits observed in SD, matching AD and SD participants for lexical semantic abilities would have prevented from any interpretation of AD performances in terms of semantic knowledge. Further analysis showed that the finding of fruit and vegetables preservation in both the perceptual and functional conditions neither depended on cognitive variables, such as lexical frequency or lexical-semantic impairment levels, nor on disease-dependent variables, such as age at onset, disease duration, or severity of temporal atrophy. Most importantly, an intraindividual analysis showed an advantage of fruit and vegetables in more than two thirds of our SD patient sample, indicating that the great majority of SD participants genuinely displayed this effect and ruling out any possible mean biases due to outliers. Furthermore, follow-up data provided evidence that this advantage of fruit and vegetables observed at baseline was still present with the same magnitude at a group level after 2 years. At an individual level of analysis, this was true for 92% (functional features) and 75% (perceptual features) of the participants. Finally, 67% (functional features) and 80% (perceptual features) of SD participants who did not present with advantage of fruit and vegetables at baseline did present with it at follow-up, thereby ruling out the possibility of a random effect, unpredictable at the individual level. Since this effect was confirmed at an intraindividual level and turned out to be consistent at follow-up, we argue that the fruit and vegetables category enjoys a particular status within the semantic system that makes its items more resistant to the worsening semantic disruption observed in SD. As mentioned above, we were not able to find any previous reports of a category-specific effect in SD at a group level. Only four SD cases with category-semantic deficits have been documented in the literature (Patient IW; Lambon Ralph, Graham, et al., 1998; Patient MF; Barbarotto et al., 1995; Patient KH; Lambon Ralph et al., 2003; and Patient LI; Zannino et al., 2006). However, given that, as Lambon Ralph et al. (2003) pointed out, ‘‘the combination of semantic dementia and category specificity is something of a rarity’’ (p. 319), the contradiction between previous reports and the present findings merits further discussion. There are several obvious methodological explanations for this contradiction. In particular, our study was the first to assess category specificity in SD with such a large sample of participants, and one would expect a higher statistical power to reveal different findings. Moreover, Lambon-Ralph et al.’s (2003) study included only one patient (AN) in the very early stage of SD, while disease duration was not mentioned for the other participants. At variance with this last sample, our group was quite homogenous, only including participants in the early stages of SD. Furthermore, as stated in the Introduction, we looked for category-specific effects rather than focusing on a more general dichotomy between the living versus nonliving domains. Lastly, we should emphasize another major difference between the current study and the one conducted by the Cambridge group. Our use of a sorting task in

the assessment of semantic knowledge clearly avoided the recruitment of the expressive lexical-semantic abilities known to be impaired in SD patients (see, for example, Warrington, 1975). Taken together, these explanations could account for our finding of category-specific preservation in our SD sample. Nevertheless, one limitation of our task must be mentioned here: initially designed for clinical purposes in the early nineties (inclusion started in 1991), potential effects of confounding variables like conceptual familiarity were not controlled. One can easily argue that fruit and vegetables items could actually be easier to sort because of a higher familiarity. Indeed, future studies must carefully match items for familiarity since it is an important factor determining performance accuracy, as previously noted (LambonRalph, Graham, et al., 1998). However, since we failed to find any category effect in the control sample, or in AD patients group, we believe that a familiarity bias alone cannot account for our results. Furthermore, many previous reports in SD suggested that personal familiarity could be a key factor influencing semantic performances, rather than the overall familiarity standardized in the general population (Snowden, Griffiths, & Neary, 1994, 1995). In that perspective, one should ideally remove any potential personal familiarity bias, determined for each SD participant. This was clearly not possible in our retrospective study including 35 SD patients. Nonetheless, we speculate that personal familiarity characterizing one’s experience with fruit and vegetables might not be significantly different from the one with kitchenware. Our results precisely demonstrated a significant preservation of fruit and vegetables relative to kitchenware category for 74% of the SD participants in the ‘‘perceptual-words’’ condition at baseline. 4.2. Neuroanatomical considerations Whereas no previous study has ever documented a categoryspecific effect at a group level in SD, other pathologies (including stroke, paraneoplastic syndrome) are known to potentially give rise to the relative preservation of the fruit and vegetables category (Case EW; Caramazza & Shelton, 1998; Case KR; Hart & Gordon, 1992). Numerous cases of a disproportionate impairment of fruit and vegetables have also been documented (see, for example, Case FAV; Crutch & Warrington, 2003; Case MD; Hart, Berndt, & Caramazza, 1985; Case TU; Farah & Wallace, 1992; Case RS; Samson & Pillon, 2003), such that this pattern of performance seems to be the rule rather than the exception. Notably, an apparently selective deficit for fruit and vegetables was recently reported by Mendez (2007) in two empirical case studies of SD. Unfortunately, the author did not provide sufficient data for us to examine these cases in the light of previous studies. We therefore need to look elsewhere for an explanation for our finding of relative and selective preservation of fruit and vegetables knowledge. The neural substrates underlying category-specific impairments are probably the first factor to consider. A recent study investigated the relationship between stroke in posterior cerebral artery (PCA) territory and semantic category dissociations (Capitani et al., 2009). The authors outlined a possible dissociation between a selective deficit for fruit and vegetables in patients with left PCA stroke, who had lesions affecting the middle and posterior portions of the left fusiform gyrus (FG), and a disproportionate deficit for animals in patients with herpex simplex encephalitis (HSE), who typically present with lesions in the anterior temporal lobe areas. Based on this finding, these authors suggested that distinct brain regions may underlie knowledge about plant life and knowledge about animals: the middle and posterior portions of the left FG would appear to be critical for plant-life knowledge, whereas animal knowledge critically involves left anterior temporal areas. Furthermore, prior studies have underlined the major role of color knowledge in the correct identification of fruit and vegetables


compared with other object categories, including living ones (Crutch & Warrington, 2003; Warrington & McCarthy, 1987). As it happens, color knowledge retrieval seems to depend critically on the left posterior parts of the FG, as evidenced in a recent fMRI study in healthy individuals (Simmons et al., 2007; see also Chao & Martin, 1999; Price, Noppeney, Phillips, & Devlin, 2003). We therefore argue for the following neuroanatomical and cognitive account of our finding of relative preservation of fruit and vegetables in SD patients. In its early stages, SD is characterized by bilateral atrophy (most often predominant in the left hemisphere) of the anterior temporal lobes. Within the temporal lobe, several imaging studies converge to show a rostrocaudal gradient of dysfunction, with the anterior parts being more affected than the posterior ones (Brambati et al., 2009; Bright, Moss, Stamatakis, & Tyler, 2008; Chan et al., 2001; Desgranges et al., 2007). The same findings were also yielded focusing on FG. Indeed, recent neuroimaging studies of SD have highlighted abnormalities in the rostral parts of the FG (BA 20, 36; Desgranges et al., 2007; Mion et al., 2010) rather than the caudal ones. The latter has long been associated with semantic knowledge (Chao & Martin, 1999). Given the role of the left posterior parts of the FG in the retrieval of color knowledge, and given that color attribute knowledge could be of discriminatory value for the correct identification of fruit and vegetables (Crutch & Warrington, 2003), our finding of the relative preservation of this category could be accounted for by the relative preservation of the posterior FG at the beginning of the neurodegenerative process underlying SD. 4.3. Theoretical accounts of semantic knowledge We now need to discuss how this finding of selective preservation of fruit and vegetables knowledge can be related to actual theoretical accounts of semantic memory. One of the first conceptual accounts of category-specific deficits was the sensory-functional theory (SFT), developed in the eighties by Warrington and Shallice. According to the initial formulation of this theory, the identification of living entities depends mainly on sensory features, whereas functional features are critical for the representation of nonliving items. The authors did not make any predictions regarding more fine-grained dissociations. In particular, SFT does not predict any dissociation within living categories. However, this was clearly the case in the present study, where fruit and vegetables were better preserved than animals. Moreover, this relative preservation of that particular category of living entities occurred despite greater impairment for perceptual than functional features. These results are therefore at variance with SFT. Nonetheless, the more fine-grained distinctions between concepts proposed in the SFT variant framework (Crutch & Warrington, 2003; Warrington & McCarthy, 1987) could fit our interpretation of the dissociation between fruit and vegetables and animals, in that color attributes – which can be regarded as a separate sensory channel of knowledge acquisition according to the SFT variant could be more relevant for the identification of fruit and vegetables than for that of animals. Notwithstanding, as our task did not allow us to make any predictions regarding the relative influence of the different sensory channels involved in knowledge acquisition, our results cannot reasonably be interpreted within this particular theoretical framework. Furthermore, it should be borne in mind that this variant was recently challenged in an fMRI study demonstrating that both congenitally blind and sighted individuals display the same pattern of category-specific activation within the ventral stream (Mahon, Anzellotti, Schwarzbach, Zampini, & Caramazza, 2009). Another influential view is formulated in Caramazza and Shelton’s DSK hypothesis (1998), whose core postulate is the existence of distinct neural substrates underpinning different domains of

265

knowledge (animals, fruit and vegetables, conspecifics and, probably, tools) due to evolutionary pressures. Consequently, our finding of relative preservation of fruit and vegetables may be in accordance with this model’s main prediction, illustrated by the authors’ princeps case, EW. Nonetheless, as every feature or piece of information about a concept is supposedly aggregated within the same subsystem, DSK does not predict any dissociation between the attributes of knowledge within a single category. There was, however, just such a dissociation in the tools category in the present study, with functional features being better preserved than perceptual ones. Therefore, our results failed to confirm all the predictions of the classic DSK hypothesis. At odds with these theoretical accounts of semantic memory, based mainly on multiple semantic stores, two well-known models suggest the existence of a single store where semantic knowledge is organized according to statistical regularities in the co-occurrence of object features, but also according to their distinctiveness (Computational Account : Devlin et al., 1998; Gonnerman et al., 1997. Conceptual Structure Account: Davis, Moss, de Mornay Davies, & Tyler, 1999; Taylor, Moss, & Tyler, 2007; Tyler & Moss, 2001). The Conceptual Structure Account (CSA) actually makes predictions that contrast with our results. According to this model, based on the neuropsychology of neurodegenerative disorders, fruit and vegetables should be more vulnerable to brain damage, as their features are more weakly intercorrelated and less distinctive than those of animals. Therefore, the present study seemed to refute the CSA hypothesis, as a living category (i.e., fruit and vegetables) was found to be relatively preserved despite dramatic semantic disruption. Lastly, Devlin and Gonnerman’s Computational Account (CA) makes interesting predictions regarding our findings. According to this model, living entities are characterized by more shared and intercorrelated features than nonliving ones, making the living entities more resistant to mild-to-moderate levels of semantic disruption. This assumption is not entirely in agreement with the pattern of deficits we observed in our SD participants since their knowledge about animals was as impaired as those of tools and kitchenware. However, a careful examination of Devlin et al.’s data (1998; see graph on p. 84) reveals that the mean number of correlated features is higher for fruit and vegetables than it is for animals, even if the level of significance of this more fine-grained dissociation among biological entities was not reported. If we follow the core prediction of the model, this higher number of correlated features should result in knowledge about fruit and vegetables being more resistant in the face of gradual semantic disruption, thus accounting for our main finding. Another CA’s main prediction developed in Gonnerman et al. (1997) concerned the differential time course of living and nonliving domains of knowledge as semantic memory decline. Gonnerman et al. (1997) predicted that the worsening of the semantic deficit for biological entities would follow a nonlinear curve, whereas the semantic knowledge breakdown for artefacts would follow a linear function. Thus, biological kinds would be more resistant to semantic disruption until a critical point was reached, where the reverse pattern of deficit would be observed (i.e., living things more affected than nonliving ones). We should therefore have observed a difference in the magnitude of the category effect at follow-up. At variance with those predictions, we failed to find any significant interaction between category and time of testing (baseline versus 24-months follow-up). Besides, both the persistent preservation of fruit and vegetables at follow-up and the absence of any reversal pattern of deficits could clearly advocate against the assumptions of CA’s model. Alternatively, we can also argue that the follow-up assessment may have been undertaken too soon in the present study for us to be able to observe the expected reverse pattern of performance.

266


Overall, we thus believe that this last theoretical account of semantic knowledge organization is the one that seems to fit our data best. However, future research testing CA’s predictions in SD should concentrate on two main aims: investigating the existence of a ‘‘narrow window’’ – where living entities could be preserved in face of a deficit for nonliving ones – on the one hand, and exploring whether a longer follow-up allows to observe the reverse pattern (i.e. a greater impairment of living rather than nonliving entities) on the other hand. 5. Conclusion The aim of this study was to explore the potential effect of different knowledge categories on gradual semantic breakdown in a large cohort of patients with SD. We observed relative preservation of knowledge about fruit and vegetables, at both baseline and 2-years follow-up, despite significant semantic memory impairment. To the best of our knowledge, this is the first time that such category-specific preservation has been reported in SD. We argue for a twofold interpretation, at both cognitive and neuroanatomical levels. We suggest that the relative preservation of fruit and vegetables can be accounted for by the importance of color knowledge in their discrimination, which in turn may be preserved because of the initial preservation of the posterior FG in SD. The left posterior FG has long been documented as an important area for color knowledge retrieval. Furthermore, on reviewing the main predictions of the most influential models of semantic memory, we found that our findings best fitted the core predictions of Devlin and Gonnerman’s CA, (Devlin et al., 1998; Gonnerman et al., 1997). Further studies are now needed to investigate the role of color knowledge in the preservation or disruption of semantic knowledge in SD. With regard to the methodological caveats inherent to semantic knowledge assessment in SD participants, future studies must avoid the limitations arising from the explicit verbal responses and controlled processes required for the great majority of semantic tasks, by favouring implicit assessment paradigms. Finally, our hypothesis regarding the relative preservation of posterior FG clearly requires further investigation, which is currently ongoing with a subgroup of our SD sample. Acknowledgments We are particularly indebt to Prof. Guido Gainotti who provided helpful comments and advices on a previous version of the study. We are also grateful to Elizabeth Portier for the English translation. Finally, we wish to thank patients and their families for their willingness and their endless patience to devote such efforts to this study. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bandl.2013.01. 003. References Agniel, A., Joanette, Y., Doyon, B., & Duchein, C. (1992). Protocole Montréal-Toulouse, Evaluation des Gnosies Visuelles et Auditives (PEGVA). Isbergues: L’Ortho Edition. Barbarotto, R., Capitani, E., Spinnler, H., & Trivelli, C. (1995). Slowly progressive semantic impairment with category specificity. Neurocase, 1, 107–119. Beauvois, M. F. (1982). Optic aphasia: A process of interaction between vision and language. Philosophical Transactions of the Royal Society of London. Series B, Biological of Sciences, 298(1089), 35–47. Belliard, S., Bon, L., LeMoal, S., Jonin, P. Y., Vercelletto, M., & LeBail, B. (2007). La démence sémantique [Semantic dementia]. Psychologie et Neuropsychiatrie du Vieillissement, 5(2), 127–138.

Belliard, S., Merck, C., Jonin, P. Y., Lemoal, S., & Vercelletto, M. (2011). Démence sémantique : démographie et données neuropsychologiques initiales chez 82 patients. [Semantic dementia: Demography and neuropsychological assessment in a cohort of 82 cases]. Revue de Neuropsychologie, 3(4), 257–265. Benton, A. L., Sivan, A. B., Hamsher, K. deS., Varney, N. R., & Spreen, O. (1994). Contributions to neuropsychological assessment: A clinical manual (2nd ed.). New York: Oxford University Press. Binney, R. J., Embleton, K. V., Jefferies, E., Parker, G. J., & Ralph, M. A. (2010). The ventral and inferolateral aspects of the anterior temporal lobe are crucial in semantic memory: Evidence from a novel direct comparison of distortioncorrected fMRI, rTMS, and semantic dementia. Cerebral Cortex, 20(11), 2728–2738. Bozeat, S., Lambon Ralph, M. A., Patterson, K., Garrard, P., & Hodges, J. R. (2000). Non-verbal semantic impairment in semantic dementia. Neuropsychologia, 38(9), 1207–1215. Brambati, S. M., Rankin, K. P., Narvid, J., Seeley, W. W., Dean, D., Rosen, H. J., et al. (2009). Atrophy progression in semantic dementia with asymmetric temporal involvement: A tensor-based morphometry study. Neurobiology of Aging, 30(1), 103–111. Bright, P., Moss, H. E., Stamatakis, E. A., & Tyler, L. K. (2008). Longitudinal studies of semantic dementia: The relationship between structural and functional changes over time. Neuropsychologia, 46(8), 2177–2188. Capitani, E., Laiacona, M., Mahon, B., & Caramazza, A. (2003). What are the facts of semantic category-specific deficits? A critical review of the clinical evidence. Cognitive Neuropsychology, 20(3), 213–261. Capitani, E., Laiacona, M., Pagani, R., Capasso, R., Zampetti, P., & Miceli, G. (2009). Posterior cerebral artery infarcts and semantic category dissociations: A study of 28 patients. Brain, 132(Pt 4), 965–981. Caramazza, A., Hillis, A. E., Rapp, B. C., & Romani, C. (1990). The multiple semantics hypothesis: Multiple confusions? Cognitive Neuropsychology, 7, 161–189. Caramazza, A., & Mahon, B. Z. (2003). The organization of conceptual knowledge: The evidence from category-specific semantic deficits. Trends in Cognitive Sciences, 7(8), 354–361. Caramazza, A., & Shelton, J. R. (1998). Domain-specific knowledge systems in the brain: The animate–inanimate distinction. Journal of Cognitive Neuroscience, 10(1), 1–34. Chan, D., Fox, N. C., Scahill, R. I., Crum, W. R., Whitwell, J. L., Leschziner, G., et al. (2001). Patterns of temporal lobe atrophy in semantic dementia and Alzheimer’s disease. Annals of Neurology, 49(4), 433–442. Chao, L. L., & Martin, A. (1999). Cortical regions associated with perceiving, naming, and knowing about colors. Journal of Cognitive Neuroscience, 11(1), 25–35. Crutch, S. J., & Warrington, E. K. (2003). The selective impairment of fruit and vegetable knowledge: A multiple processing channels of fine-grain category specificity. Cognitive Neuropsychology, 20(3), 355–372. Davis, M. H., Moss, H. E., de Mornay Davies, P., & Tyler, L. K. (1999). Spot the difference: Investigations of conceptual structure for living things and artefacts using speeded word-picture matching. Brain and Language, 69(3), 411–414. De Renzi, E., & Faglioni, P. (1978). Normative data and screening power of a shortened version of the Token test. Cortex, 14(1), 41–49. Deloche, G., & Hannequin, D. (1997). Test de dénomination orale d’images. DO 80. ECPA. Desgranges, B., Matuszewski, V., Piolino, P., Chételat, G., Mézenge, F., Landeau, B., et al. (2007). Anatomical and functional alterations in semantic dementia: A voxel-based MRI and PET study. Neurobiology of Aging, 28(12), 1904–1913. Devlin, J. T., Gonnerman, L. M., Andersen, E. S., & Seidenberg, M. S. (1998). Categoryspecific semantic deficits in focal and widespread brain damage: A computational account. Journal of Cognitive Neuroscience, 10(1), 77–94. Farah, M. J., & Wallace, M. A. (1992). Semantically-bounded anomia: Implications for the neural implementation of naming. Neuropsychologia, 30(7), 609–621. Fodor, J. (1983). The modularity of mind. Cambridge, MA: MIT Press. Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). ‘‘Mini-Mental State’’: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189–198. Gainotti, G. (2010). The influence of anatomical locus of lesion and of genderrelated familiarity factors in category-specific semantic disorders for animals, fruits and vegetables: A review of single-case studies. Cortex, 46(9), 1072– 1087. Gainotti, G. (2011). The organization and dissolution of semantic-conceptual knowledge: Is the ‘amodal hub’ the only plausible model? Brain and Cognition, 75(3), 299–309. Gainotti, G. (2012). The format of conceptual representations disrupted in semantic dementia: A position paper. Cortex, 48(5), 521–529. Gonnerman, L. M., Andersen, E. S., Devlin, J. T., Kempler, D., & Seidenberg, M. S. (1997). Double dissociation of semantic categories in Alzheimer’s disease. Brain and Language, 57(2), 254–279. Goodglass, H., & Kaplan, E. (1972). The assessment of aphasia and related disorders. Philadelphia: Lea and Febiger. Hart, J., Jr., Berndt, R. S., & Caramazza, A. (1985). Category-specific naming deficit following cerebral infarction. Nature, 316(6027), 439–440. Hart, J., Jr., & Gordon, B. (1992). Neural subsystems for object knowledge. Nature, 359(6390), 60–64. Hodges, J. R., Patterson, K., Oxbury, S., & Funnell, E. (1992). Semantic dementia: Progressive fluent aphasia with temporal lobe atrophy. Brain, 115(Pt 6), 1783–1806. Lambon Ralph, M. A., Graham, K. S., Ellis, A. W., & Hodges, J. R. (1998). Naming in semantic dementia—What matters? Neuropsychologia, 36(8), 775–784.

C. Merck et al. / Brain & Language 124 (2013) 257–267 Lambon Ralph, M. A., Howard, D., Nightingale, G., & Ellis, A. W. (1998). Are living and non living category-specific deficits causally linked to impaired perceptual or associative knowledge? Evidence from a category-specific double dissociation. Neurocase, 4, 311–338. Lambon Ralph, M. A., Patterson, K., Garrard, P., & Hodges, J. R. (2003). Semantic dementia with category specificity: A comparative case-series study. Cognitive Neuropsychology, 20(3), 307–326. Lambon Ralph, M. A., Pobric, G., & Jefferies, E. (2009). Conceptual knowledge is underpinned by the temporal pole bilaterally: Convergent evidence from rTMS. Cerebral Cortex, 19(4), 832–838. Laws, K. R., Gale, T. M., Leeson, V. C., & Crawford, J. R. (2005). When is category specific in Alzheimer’s disease? Cortex, 41(4), 452–463. Luzzi, S., Snowden, J. S., Neary, D., Coccia, M., Provinciali, L., & Lambon Ralph, M. A. (2007). Distinct patterns of olfactory impairment in Alzheimer’s disease, semantic dementia, frontotemporal dementia, and corticobasal degeneration. Neuropsychologia, 45(8), 1823–1831. Mahon, B. Z., Anzellotti, S., Schwarzbach, J., Zampini, M., & Caramazza, A. (2009). Category-specific organization in the human brain does not require visual experience. Neuron, 63(3), 397–405. McCarthy, R. A., & Warrington, E. K. (1988). Evidence for modality-specific meaning systems in the brain. Nature, 334(6181), 428–430. McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan, E. M. (1984). Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS–ADRDA work group under the auspices of department of health and human services task force on alzheimer’s disease. Neurology, 34(7), 939–944. Mendez, M. F. (2007). Impaired knowledge for fruits and vegetables in semantic dementia. The Journal of Neuropsychiatry and Clinical Neurosciences, 19(3), 349–350. Mion, M., Patterson, K., Acosta-Cabronero, J., Pengas, G., Izquierdo-Garcia, D., Hong, Y. T., et al. (2010). What the left and right anterior fusiform gyri tell us about semantic memory. Brain, 133(11), 3256–3268. Moreaud, O., Belliard, S., Snowden, J., Auriacombe, S., Basaglia-Pappas, S., Bernard, F., et al. (2008). Démence sémantique: réflexions d’un groupe de travail pour des critères de diagnostic en français et la constitution d’une cohorte de patients. [Semantic dementia: Reflexions of a French working group for diagnostic criteria and constitution of a patient cohort]. Revue Neurologique, 164(4), 343–353. Neary, D., Snowden, J. S., Gustafson, L., Passant, U., Stuss, D., Black, S., et al. (1998). Frontotemporal lobar degeneration: A consensus on clinical diagnostic criteria. Neurology, 51(6), 1546–1554. Nelson, H. E. (1976). A modified card sorting test sensitive to frontal lobe defects. Cortex, 12(4), 313–324. New, B., Pallier, C., Brysbaert, M., & Ferrand, L. (2004). Lexique 2: A new French lexical database. Behavior Research Methods, Instruments, and Computers, 36(3), 516–524. Osterrieth, P. A. (1944). Le test de copie d’une figure complexe: contribution à l’étude de la perception et de la mémoire [The test of copying a complex figure: A contribution to the study of perception and memory]. Archives de Psychologie, 30, 286–356. Patterson, K., Nestor, P. J., & Rogers, T. T. (2007). Where do you know what you know? The representation of semantic knowledge in the human brain. Nature Reviews. Neuroscience, 8(12), 976–987. Pobric, G., Jefferies, E., & Ralph, M. A. (2010). Amodal semantic representations depend on both anterior temporal lobes: Evidence from repetitive transcranial magnetic stimulation. Neuropsychologia, 48(5), 1336–1342. Price, C. J., Noppeney, U., Phillips, J., & Devlin, J. T. (2003). How is the fusiform gyrus related to category-specificity? Cognitive Neuropsychology, 20(3), 561–574.

267

Quillian, M. R. (1966). Semantic memory. Unpublished doctoral dissertation, Carnegie institute of Technology, 1966. Reprinted in part. In M. Minsky (Ed.), Semantic information processing. Cambridge, Mass: MIT Press. Raven, J., Raven, J. C., & Court, J. H. (1998). Raven manual: Standard progressive matrices. Oxford, England: Oxford Psychologists Press. Reitan, R. M. (1958). The validity of the trail making test as an indicator of organic brain damage. Perceptual and Motor Skills, 8, 271–276. Rogers, T. T., Hocking, J., Noppeney, U., Mechelli, A., Gorno-Tempini, M. L., Patterson, K., et al. (2006). Anterior temporal cortex and semantic memory: Reconciling findings from neuropsychology and functional imaging. Cognitive, Affective, and Behavioural Neurosciences, 6(3), 201–213. Samson, D., & Pillon, A. (2003). A case of impaired knowledge for fruit and vegetables. Cognitive Neuropsychology, 20(3), 373–400. Simmons, W. K., Ramjee, V., Beauchamp, M. S., McRae, K., Martin, A., & Barsalou, L. W. (2007). A common neural substrate for perceiving and knowing about color. Neuropsychologia, 45(12), 2802–2810. Snowden, J. S., Goulding, P. J., & Neary, D. (1989). Semantic dementia: A form of circumscribed cerebral atrophy. Behavioural Neurology, 2, 167–182. Snowden, J. S., Griffiths, H. L., & Neary, D. (1994). Semantic dementia: Autobiographical contribution to preservation of meaning. Cognitive Neuropsychology, 11, 265–288. Snowden, J. S., Griffiths, H. L., & Neary, D. (1995). Autobiographical experience and word meaning. Memory, 3, 225–246. Snowden, J. S., Thompson, J. C., & Neary, D. (2004). Knowledge of famous faces and names in semantic dementia. Brain, 127(Pt 4), 860–872. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643–662. Taylor, K. I., Moss, H. E., & Tyler, L. K. (2007). The conceptual structure account: A cognitive model of semantic memory and its neural instantiation. In J. Hart & M. A. Kraut (Eds.), Neural basis of semantic memory (pp. 265–301). Cambridge: Cambridge University Press. Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), Organization of memory (pp. 381–403). New York: Academic Press. Tyler, L. K., & Moss, H. E. (2001). Towards a distributed account of conceptual knowledge. Trends in Cognitive Sciences, 5(6), 244–252. Violon, A., & Wijns, C. (1984). Le test de la Ruche. Test de perception et d’apprentissage progressif en mémoire visuelle. Braine le Château (Belgium): Editions l’Application des techniques modernes SPRL. Visser, M., Embleton, K. V., Jefferies, E., Parker, G. J., & Ralph, M. A. (2010). The inferior, anterior temporal lobes and semantic memory clarified: Novel evidence from distortion-corrected fMRI. Neuropsychologia, 48(6), 1689–1696. Visser, M., Jefferies, E., & Lambon Ralph, M. A. (2010). Semantic processing in the anterior temporal lobes: A meta-analysis of the functional neuroimaging literature. Journal of Cognitive Neuroscience, 22(6), 1083–1094. Warrington, E. K. (1975). The selective impairment of semantic memory. The Quarterly Journal of Experimental Psychology, 27(4), 635–657. Warrington, E. K., & McCarthy, R. A. (1987). Categories of knowledge: Further fractionations and an attempted integration. Brain, 110(Pt 5), 1273–1296. Warrington, E. K., & Shallice, T. (1979). Semantic access dyslexia. Brain, 102(1), 43–63. Warrington, E. K., & Shallice, T. (1984). Category specific semantic impairments. Brain, 107(Pt 3), 829–854. Wechsler, D. (1981). Wechsler adult intelligence scale revised manual. New York: The Psychological Corporation. Zannino, G. D., Perri, R., Pasqualetti, P., Di Paola, M., Caltagirone, C., & Carlesimo, G. A. (2006). The role of semantic distance in category-specific impairments for living things: Evidence from a case of semantic dementia. Neuropsychologia, 44(7), 1017–1028.