Neuropsychologia 38 (2000) 1207±1215

Non-verbal semantic impairment in semantic dementia
Sasha Bozeat a, Matthew A. Lambon Ralph a,*, Karalyn Patterson a, Peter Garrard b,
John R. Hodges a, b
MRC Cognition and Brain Sciences Unit, 15 Chaucer Road, Cambridge CB2 2EF, UK
University Neurology Unit, Addenbrooke's Hospital, Cambridge, UK
Received 11 October 1999; received in revised form 8 February 2000; accepted 14 February 2000


The clinical presentation of patients with semantic dementia is dominated by anomia and poor verbal comprehension.
Although a number of researchers have argued that these patients have impaired comprehension of non-verbal as well as verbal
stimuli, the evidence for semantic deterioration is mainly derived from tasks that include some form of verbal input or output.
Few studies have investigated semantic impairment using entirely non-verbal assessments and the few exceptions have been
based on results from single cases ([3]: Breedin SD, Sa€ran EM, Coslett HB. Reversal of the concreteness e€ect in a patient with
semantic dementia. Cognitive Neuropsychology 1994;11:617±660, [12]: Graham KS, Becker JT, Patterson K, Hodges JR. Lost
for words: a case of primary progressive aphasia? In: Parkin A, editor. Case studies in the neuropsychology of memory, East
Sussex: Lawrence Erlbaum, 1997. pp. 83±110, [21]: Lambon Ralph MA, Howard D. Gogi aphasia or semantic dementia?
Simulating and assessing poor verbal comprehension in a case of progressive ¯uent aphasia. Cognitive Neuropsychology, (in-
This study employed sound recognition and semantic association tasks to investigate the nature of the verbal and non-verbal
comprehension de®cit in 10 patients with semantic dementia. The patients were impaired on both verbal and non-verbal
conditions of the assessments, and their accuracy on these tasks was directly related to their scores on a range of other tests
requiring access to semantic memory. Further analyses revealed that performance was graded by concept and sound familiarity
and, in addition, identi®ed signi®cant item consistency across the di€erent conditions of the tasks. These results support the
notion that the patients' de®cits across all modalities were due to degradation within a single, central network of conceptual
knowledge. There were also reliable di€erences between conditions. The sound-picture matching task proved to be more
sensitive to semantic impairment than the word-picture matching equivalent, and the patients performed signi®cantly better on
the picture than word version of a semantic association test. We propose that these di€erences arise directly from the nature of
the mapping between input modality and semantic memory. Words and sounds have an arbitrary relationship with meaning
while pictures bene®t from a degree of systematicity with conceptual knowledge about the object. 7 2000 Elsevier Science Ltd.
All rights reserved.

Keywords: Semantic dementia; Progressive ¯uent aphasia; Frontotemporal dementia; Non-verbal comprehension; Conceptual knowledge

1. Introduction severely a€ecting the temporal pole and the infero-lat-
eral temporal lobe, typically bilaterally but often asym-
metrically [14,33]. The ventromedial frontal cortex
The syndrome of semantic dementia is associated
bilaterally and the amygdaloid complex are also often
with circumscribed temporal lobe atrophy, most
a€ected [25,26]. The term semantic dementia was pro-
posed because the cases have a selective and progress-
* Corresponding author. Tel.: +44-1223-355294.
ive deterioration of conceptual knowledge. The
E-mail address: (M.A. selective nature of the semantic impairment was ®rst
Lambon Ralph). described by Warrington [35], who reported three

0028-3932/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 8 - 3 9 3 2 ( 0 0 ) 0 0 0 3 4 - 8

To compare the performance of the patients with Howard. visuospatial drawings of objects and animals that have a speci®c abilities and day-to-day memory [13. the evidence for a non-verbal semantic clinical neuropsychologist. The semantic term memory. Participants patient who was surprised that her husband could no longer understand the names of objects that he still Ten patients were identi®ed through the Memory used eciently everyday [34]. as a general measure of cognitive impairment [8]. General neuropsychology to test patients with pure word deafness [1.g. selective nature of the semantic de®cit in these patients The aim of this cross-sectional study was to provide has been con®rmed by their good performance on further evidence for the non-verbal conceptual impair- assessments of current day-to-day memory. noted comments from the wife of a 2. tasks can also provide a straightforward. which is a non-verbal assessment reported: anomia. More recent studies have highlighted to-picture test and a corresponding sound-to-word ver- the patients' progressive loss of knowledge about the sion thereby con®rming the presence of a non-verbal meanings of words. verbal ¯uency for the letters F. word-to-picture matching and draw. that of age and education matched controls. we verbal semantic impairments in single case studies of selected normal participants from the Medical patients with semantic dementia. being able to select the correct colour for the polar and inferolateral regions of one or both of only 10 out of 28 items. Assessments been assessed in one case with herpes simplex virus encephalitis (HSVE) but it has been used most often 2. etc).. both Graham et al. In this regard. all patients were given a number of stan- verbal input (e. impairment in single word compre- of semantic associations [3.. [12] and also Lambon Ralph.1208 S. phonology and syntax until very late in the course vious studies. non-verbal reason. de®nitions of pictured concepts.14]. where they were seen by a degraded knowledge for non-verbal as well as verbal senior neurologist (JRH).2.31]. A. Bozeat et al. aphasia. sound recognition participant panel. A number of researchers have reported in the context of ¯uent speech production. [18] and Tanabe et al. The patient showed severe impairment brain imaging by MRI showed focal atrophy involving on this task. ning together with the usual battery of screening blood The existing literature contains limited examples of tests to exclude treatable causes of dementia. schizophrenia. The semantic impairment [21]. the temporal lobes. / Neuropsychologia 38 (2000) 1207±1215 patients with progressive anomia and impaired word dementia patient IW was impaired on both this sound- comprehension.20. Research Council Cognition and Brain Sciences Unit's In addition to picture based tests.6]. short-term ment in semantic dementia based on a larger set of verbal memory. the Lambon Ralph and Howard [21] formulated a non. syntax. [34] emphasised that the impairment in Gogi aphasia is 2. UK.30]. sounds and a new semantic association task were used to the extent that the term `Gogi'. Rather The following battery of neuropsychological tests than requiring a subject to identify an object or animal was administered: the Mini-Mental State Examination by naming it in response to its characteristic sound. Using similar colouring tasks. dard psychiatric rating scales to exclude major func- ing of objects to dictation of their names) or output tional psychiatric disorders such as depression and (e. Structural known colour. recognition of environmental dominated by anomia and poor verbal comprehension. visuospatial skills. digit span subtest of the Wechsler Memory Scale- verbal assessment by asking the patient to match the Revised (WMS-R) [38] to assess auditory-verbal short sound to a picture of the target concept.g. S to . Sound recognition has 2. In addition to a battery of standard of the disease [14]. [3] hension and impoverished semantic knowledge with assessed a patient on colouring of black and white line relative preservation of phonology. including the Pyramids The clinical presentation of semantic dementia is and Palm Trees test. They all underwent CT or MRI scan- descriptions of famous people from face stimuli. Breedin et al. Although it has been and Cognitive Disorders Clinic at Addenbrooke's Hos- argued that patients with semantic dementia have pital. the All patients presented with a progressive loss of vast majority of which rely on pictures to probe vocabulary a€ecting expressive and receptive language knowledge. They all impairment on the picture version of the Pyramid and ful®lled the criteria for semantic dementia previously Palm Trees test [15]. a senior psychiatrist and a stimuli [14.23. Cambridge. In addition to a clinical impairment has tended to rely upon tasks that include assessment.33]. assessments of comprehension.33]. entirely non-verbal assessment of comprehension.2. Tanabe et al. picture naming. Method limited to the domain of language.1. Nightingale and Ellis [22] demonstrated non. patients and a more extended investigation than in pre- ing. is sometimes used to describe these patients in Japan [17. objects and concepts [13.1. non-verbal assessment of comprehension. Imura et al. or word meaning to test the patients' semantic memory non-verbally.

span. all items (targets and response choices) were presented as pictures.2. arity with each item and how often they hear each ing three categories of living things (animals.2. Some less familiar sounds were included visuospatial function in more detail [36]. The two patients. measured by Raven's matrices. Two other semantic assessments were also adminis.2.1.2.. 3. Subjects were 2. household items. tree. Rey Figure was good in all cases revealing preser- The target items for this test were the same 64 items as vation of non-verbal episodic memory. Camel and Cactus test. For example. General neuropsychology ing the target plus nine within-category foils. in which the subject is asked to produce as many exemplars as possible in one minute for each of the six 3. General semantic assessment all stimuli were words. represent. the scores are given as per- assessments were designed speci®cally for this study. Bozeat et al.1. vehicles and musical Space Perception battery were also used to assess instruments). given the advanced version of the test which is particu- Subjects were asked to choose one of four same-cat. on the letter ¯uency test. They all that is similar in meaning to the target word [37] and showed intact working memory as measured by for- the Pyramids and Palm Trees test [15] of associative ward and backward digit span. All patients had preserved visuospatial egory items that has an associative relationship with skills as indicated by their copying of the Rey Figure the target. / Neuropsychologia 38 (2000) 1207±1215 1209 test executive function. tools and vehicles). 3. those included in the battery of semantic tests described above. to IF. the 2. The assessment was administered in 3. items. word and match it to the target stimulus (picture or which is a collection of tests that use the same set of written word) from an array of 10 within-category stimulus items to assess semantic knowledge systemati. All showed reduced category ¯uency and some . chosen from the corpus of line 20 normal participants to rate both their overall famili- drawings by Snodgrass and Vanderwart [32]. It Familiarity with the items was established by asking contains 64 items.2. Concept familiarity ratings As shown in Table 2. There were three con- 2. Semantic tests ditions: matching sounds to pictures. JC and WM.2. in the other form. There was general preser- The stimuli are presented as either pictures or written vation of non-verbal problem-solving skills as words. Two new semantic advanced and the coloured. In this later assessment. Environmental Sounds test. subjects formed worse than expected on the backward digit are asked to choose one of two items that is most clo. larly dicult. birds and sound on a scale of 0 (never) to 6 (every hour). General semantic assessment. Various subtests from the Visual Object and human sounds. copy and immediate recall of the Rey Complex Ralph and Howard [21]. hend the task. The patients were asked to listen on each trial to a sound or a spoken given a selection of tasks from a semantic battery.3.2.2. All patients exhibited some impairment mid. centiles. foreign animals. the choice is between palm tree and pine tree ). The following subtests from general and the sound speci®cally.2. It contained 48 sounds from Figure to test visuospatial skills and episodic memory six categories (domestic animals. cally across di€erent input and output modalities. in one of the trials the subject and performance on the various subtests of the Visual was asked to match a camel to one of four types of Object and Space Perception battery. The 10 patients covered a broad spectrum of impair- tered: the concrete and abstract word synonym test ment as indicated by their performance on the Mini- which requires the subject to choose one of two words Mental State Examination (see Table 1).2. Results categories. for the target pyra. This test was based coloured matrices to assess non-verbal problem solving on the sound matching assessment used by Lambon [27.1. leading fruit) and three categories of artefacts (household to a measure for familiarity with both the concept in items.g. It should be noted that as two versions of the Raven's matrices were used.2.28]. Although DC per- semantic knowledge. this was probably due to a failure to compre- sely associated with the target (e. the semantic battery were administered: category ¯u- ency. sounds to written words and spoken words to picture. sun¯ower or rose. the Raven's advanced or 2. S. [29]. naming of the 64 line drawings. Recall of the vegetation: cactus (the target). who per- The ®rst was a test of semantic association based on formed below the level of the other patients were both the principle of the Pyramids and Palm Trees test [15]. spoken word-to-picture matching using picture arrays contain. in an attempt to create a more sensitive measure of early semantic impairment. Semantic tests two forms: in one. the patients included in this from a previous study [10] were already available for study also covered a wide range of semantic decline the 64 target items included in the Camel and Cactus from the mildly impaired patients such as JP through test.

2) Raven's matrices (percentiles) > 95a 50b NT 25b 95a 90a > 95a > 95a 90a 90a Rey Figure copy (36) 35 35 30 30 35 28 32 36 36 26 34 (2. supporting the view that they have a mean Ð 40.7 (2.5 24 15.8 (0. t…19† ˆ 1:16. NT Ð not tested. there was no main e€ect within the normal range on the picture version.5 14 12 NT 10.8 (0.s).2 (0. with the were impaired on both conditions of the Pyramids and patients performing worse than controls Palm trees test.The small di€erence ities of input. In this regard it is important to note favouring the word condition for the control subjects that signi®cant correlations were found between the was not statistically reliable (pictures: mean Ð 56. words: mean Ð 59.95. revealing this assessment to be a highly sensi. Post hoc paired-samples t-tests revealed that patients tended to achieve similar scores for both the the patients performed signi®cantly better on the pic- non-verbal and verbal versions of the Pyramids and ture than the word condition of this test (pictures: Palm Trees test. Scores on the picture condition of the Camel and prising given the almost uniformly chance level of Cactus test were signi®cantly correlated with all the performance on this test.1 (SD Ð 6.6) cube analysis (10) NT 10 NT 9 10 NT 10 NT 10 6 9.2.9) backward 5 7 3 4 4 5 2 5 5 5 4.2.8) Patients are ordered according to their performance on the naming and the category comprehension tests (see Table 2).5) number location (10) 10 9 10 10 9 10 10 NT 10 NT 8. 27† ˆ 4:18.3) position discrimination (20) NT 20 NT 20 20 NT 17 NT 20 20 19. n.1210 S. the patients in Tables the normal range on both conditions (see Fig. This analysis tive measure of semantic impairment.2 (11.25). which is not sur. words: mean Ð 37 (SD Ð generalised semantic impairment a€ecting all modal. b Raven's advanced progressive matrices.27). except for WM and JP being WM who scored at the bottom of the normal were impaired on comprehension as measured by the range on the picture version and JP who scored within word-to-picture matching (N. a Raven's coloured progressive matrices. it is clear that dition also correlated with the same semantic assess- eight out of 10 were impaired on the Camel and Cac. The stages of semantic dementia and their mild semantic patients performed very poorly on the concrete and de®cit was only revealed by certain assessments.3).5 NT 18. The patients revealed a signi®cant main e€ect of group. 0.5 NT 17. the two di€erent conditions of this test. crete and abstract word synonyms. p < 0:01). All patients.9) immediate recall 23. Bozeat et al.9 patients' scores for all the semantic tasks (r between (SD Ð 7. ments (r between 0. all pone-tailed < 0:05† except for the con. Apart from WM.8 (0. we note that the p ˆ 0:05).88 and 0. 27† ˆ 10:81.7 (1. the An analysis of variance was used to investigate the synonym scores were no better than expected by di€erences in performance of patients and controls for chance. t…7† ˆ 2:93.75 and 0.88. VOSP Ð Visual Object and Space Perception battery.8) dot counting(10) 10 10 10 10 10 10 10 10 10 10 9. the exception the naming test.2) Letter ¯uency (total: FAS) 27 29 31 23 7 20 16 NT 19 16 44.9 (0.94. semantic measures in Table 2. and the Environmental Sounds test. these two patients were in the early decreasing naming and comprehension scores).B. except the concrete and abstract word synonym test (r between 0. abstract word synonym test. p < 0:05). degree of anomia as indicated by their performance on tus test when compared to controls. 3. except for JP and WM who performed …F…1. As 1 and 2 are ordered as best as possible to re¯ect noted earlier.77 and 0.5) Digit span forward 6 8 6 8 6 8 7 5 6 5 6. / Neuropsychologia 38 (2000) 1207±1215 Table 1 General neuropsychology Test (maximum score) JP WM SL JC DS AT DC JW JH IF Control mean (SD) MMSE (30) 29 27 28 24 23 25 18 NT 9 11 28.6 (SD Ð 13.9 (2.2) VOSP incomplete letters (20) NT 19 20 20 20 NT 20 20 18 NT 19. 1).9). 16. Camel and Cactus test all pone-tailed < 0:01). Performance on the word con- Looking at the patients individually.3 (5. Before of condition …F…1. all pone-tailed < 0:05). . 27† < 1† but a signi®cant interaction considering the results of the Camel and Cactus test between group and condition …F…1.

The two patients who performed at chance on one or both versions of the Camel and Cac- tus test.83 and 1.1. The two horizontal lines denote the upper and lower 30.61 and cactus test. Item consistency across the two conditions of the Camel and Cactus test was investigated using a series of simultaneous logistic re- gression analyses.19]. Speci®cally.6) Word-picture matching (64) 64 63 60 58 58 57 36 23 18 18 63.64. Assessment of item consistency is complicated by the fact that some items are invariably harder than others.16. Accuracy varied available for JW and JH. (b) Performance on the word con- dition of the camel and cactus test. leading to impairments in all tasks that require activation of such knowledge. Bozeat et al. where the same terms/concepts are used.5) Word synonyms concrete (25) 13a 21 15a 12a 12a 16 14a NT 12a 13a 23.2.3) abstract (25) 14a 18 15a 8a 14a 14a 13a NT 13a 13a 23.18. S. Consistency between the picture and word conditions was assessed by predicting the former from the latter. Item consistency. The two horizontal lines denote a€ected by familiarity only in the word version of the the upper and lower limits of normal performance.3 (1. The possibility that consistency may vary across individual patients was assessed by repeating each regression analysis with an interactive term. and more fam- iliar items consistently accurate [2. between patients (Wald values 20. but there was no indication that consistency dual impairment to a single central system of concep. . (a) Performance on the picture condition of the camel and across the two conditions (Wald values 30. If this is the case.6) manmade 37 29 20 21 10 18 4 4 7 5 54. performance on di€erent semantic assessments should not only be correlated but.1) Pyramids and Palm Trees words (52) 48 48 46 44 46 45 25a 32 25a 28a 51.3 (12. and vice versa.1) pictures (52) 49 52 48 41 46 47 36 27a 37 22a 51. both the term semantic dementia implies a selective and gra. n.1 (1. These analyses revealed signi®cant consistency Fig.0 (2. Performance was signi®cantly limits of normal performance. performance in semantic dementia is known to be a€ected by the familiarity of each con- cept [9. tual knowledge.5. 1. patients should exhibit signi®cant item consistency across the di€erent modalities tested.7 (1.99. As noted in the Introduction. varied across subjects (Wald values 0. both p < 0:001). IF and DC.8 (10. / Neuropsychologia 38 (2000) 1207±1215 1211 Table 2 Assessment of semantic memory Test (maximum score) JP WM SL JC DS AT DC JW JH IF Control mean (SD) Category ¯uency living 42 38 23 15 3 14 6 3 5 2 60.2. included. patient by predicting task.01 and 73. p < 0:001). whether they involve verbal or non-verbal input or output modalities. One would therefore expect to ®nd some consistency simply on the basis that unfamiliar items will tend to be consistently inaccurate.21]. with two additional predictor variables included in the regression equation: familiarity and patient. 3.s). Data are not test (Wald=11.08.7 (0. p < 0:05).2(1. were not included in the consist- ency analysis.3) Naming (64) 59 57 45 43 17 17 11 9 6 1 62.4) a Score not signi®cantly better than expected by chance.

with the patients performing worse 39. Accuracy varied signi®cantly Analysis of variance revealed a signi®cant main between the patients (Wald values between 18.3 (SD Ð 11.s. p < 0:001).64). Performance was signi®cantly consistent across all Fig. which arose from a larger di€erence between word-to-picture matching and the two conditions requiring sound recognition in the patients than in the controls (patients: word±pic- ture: mean Ð 33.s). though not with the concrete and abstract synonym test.45. The one Looking at the patients individually. Environmental Sounds test values between p ˆ 0:08). (a) Performance on the sound-to-picture matching test.74)).3. Consistency between the three conditions of the Environmental Sounds test was also measured by a series of simultaneous logistic re- gression analyses.48)=8. sound±picture: mean Ð 21.75 (SD Ð 0. Signi®cant correlations were found between the patients' scores for all conditions of this test. The possibility that consistency may vary across individual patients was tested by repeating each regression analysis with an interactive term. 48† ˆ 81:36.97. whether measured as the famili. Paired-samples t-tests con®rmed that both the patients and controls performed signi®cantly better on the word-to-picture matching condition compared to both of the sound conditions (t values between 4.6 (SD Ð 8.65 and The two horizontal lines denote the upper and lower limits of normal 11.97.8 (SD Ð 7.57). with two ad- ditional predictor variables included in the regression equation: familiarity (either general familiarity or sound familiarity) and patient. all pone-tailed < 0:05). Bozeat et al. p < 0:01). Item consistency. all p < 0:05† or familiarity in general (Wald performance. The correlation between sound-to-picture matching and the word condition of the Camel and Cactus test just failed to reach signi®cance …r ˆ 0:59.25 (SD Ð 2. all p < 0:001).87. The possible pairings of the three conditions (Wald values two horizontal lines denote the upper and lower limits of normal performance. consistency between the sound-to-picture and sound-to-word conditions was assessed by predicting the former from the latter.s).89). controls: word±picture: mean Ð 47.01 and 23. (Wald=2.89. n.73 and 0. all p < 0:001). 2.26 and 48.2. (c) Performance on the word-to-picture matching test. but there was no indication that than the control subjects …F…1. formance on the word-to-picture matching task from ditions of the Environmental Sounds test except for performance on sound-to-picture matching. sound±picture: mean Ð 41.18 (SD Ð 2. n.1212 S. between 0. There was no signi®cant di€erence between sound-to-picture matching and sound-to-word matching (patients: t…9† ˆ 1:57. n. included. between 13.94. Accuracy was The two horizontal lines denote the upper and lower limits of normal a€ected by familiarity.83 and e€ect of group. the e€ect WM. For example.71. controls: t…17† ˆ 0:92.91 and 9. whose score fell below normal on sound-to-word of general familiarity failed to reach signi®cance matching only (see Fig. p < 0:001† and an interaction between group and con- dition (F(2. . all p < 0:001).1. all p < 0:05). / Neuropsychologia 38 (2000) 1207±1215 3. consistency varied across subjects (Wald values there was a main e€ect of condition …F…2. it is clear that exception to this summary is that. patient by predicting task. arity with the sound (Wald values between 4. the semantic battery tests and the Camel and Cactus test (between 0. sound±word: mean Ð 19. (b) Performance on the sound-to-word matching test. 3. and then repeated the other way around.54).2.16 and 34. performance. 2). pone-tailed ˆ 0:06).69). 24† ˆ 68:01. sound±word: mean Ð 41. when predicting per- they all scored outside the normal range on all con.

For example. Signi®cant correlations were found even when the network made comprehension errors. full semantic pattern is likely to bene®t comprehension cantly better on the picture than word condition of the in general. Systematicity can be thought of in terms of how well features of the input are able The aim of this study was to demonstrate that the to predict output patterns. literature on category speci®c de®cits [6. favouring picture over word input. A similar proposal can be found in the one or both conditions of the sound recognition task. Lambon Ralph and Howard [21] used a simple con- cantly below the control subjects on both the picture nectionist simulation to demonstrate the impact of version of the Camel and Cactus test and the sound. suggesting that this assessment is very sensitive to putational level of description. On the Environmental Sounds test. Even the two secondly. di€erent mappings between surface form and concep- to-picture matching condition of the Environmental tual knowledge. Overall. verbal domain. S.7]. Camel and Cactus test. there is no such predictability. Altogether. tions were removed there was no e€ect on the task tures and sounds to words. tematic and arbitrary mapping comes from another both the patients and controls were more successful on simulation reported by McGuire and Plaut [24]. Information available directly from system because the arbitrary mapping underlying word the perception of pictures and objects provides clues to comprehension is relatively sensitive to mild levels of both form and function [11]. mapping. dance [11] has some overlap with the theories of Cara. This would seem to suggest that even limited systema- Further analysis also revealed reliable di€erences ticity between input modality and a minor part of the between modalities. expected following impairment to a unitary semantic tual knowledge. . / Neuropsychologia 38 (2000) 1207±1215 1213 4. formance independent of modality. Although the discussion of these networks has been mazza and colleagues [4]. ance on the sound recognition tests. comprehension performance was comprehension is impaired in semantic dementia even always better when semantic representations were acti- in tasks where neither stimuli nor responses contain or vated by picture than by word input. that knowledge about the structural aspects mildest patients included in this study (JP and WM). damage. that cast only in terms of picture and word comprehension. di€erent conditions combined with di€erences in per. These results provide clear evidence that semantic system. support for the notion that semantic dementia is best When the quality of the two parts of the conceptual characterised as a progressive impairment to central representation was compared. like the patients formance on the di€erent conditions. who propose ®rstly. between the patients' scores on all the semantic tests as these nearly always involved activation of the correct well as item consistency across the di€erent modalities semantic region for pictures but this was much less of the Camel and Cactus test and the Environmental likely to occur for words. to damage. but. Following simulated damage to the Sounds test. There were no di€er. corresponding input feature. whereas in an arbitrary ration of semantic memory that extends to the non. even after controlling for the familiarity portion of the full distributed semantic representation of the items. each output feature can be predicted by a word comprehension de®cit but a generalised deterio. Furthermore. At a com. These computational models. had scores outside the normal range on relationships). this level of There is no puzzle in the pattern of correlations damage reduced correct performance to 78% on the between tests and consistency between items across task underpinned by arbitrary mapping. cant di€erence. the patients performed signi®. We propose that terms of the systematicity of the mapping between there may be two factors underlying this ®nding. of an object is closely linked to the semantic properties who did not show much impairment on the other that specify its function (the assumption of privileged semantic tests. supported by a systematic mapping. This pattern is described here. this can be cast in early/mild semantic impairment. though the di€erence was greater the rate of acquisition and the robustness of the system in the patients than the controls. Gibson's notion of a€or. Discussion di€erent representations. that supports semantic per. They the word-to-picture matching condition than either of found that task systematicity had a dramatic e€ect on the sound conditions. The patients performed signi®. exhibit impaired comprehension of easily explicable in terms of the notion that partially both word and picture stimuli as well as a signi®cant degraded knowledge will have di€erential e€ects on di€erence in accuracy favouring the pictures. In a perfectly systematic key impairment in semantic dementia is not simply a mapping. a result that was not echoed in Further support for the di€erential e€ects of sys- the control group. In this simulation only a Sounds test. The net- overt performance depending on the nature of the works demonstrate that such a di€erence should be mapping between input/output modalities and concep. objects bene®t from direct access to meaning from the same argument can be applied to explain perform- visual input (the assumption of privileged access) and. when just 1% of the connec- ences in either group between matching sounds to pic. this evidence provides strong was related systematically to the picture representation. require any words. Bozeat et al. both exhibited a signi®- conceptual knowledge.

impaired perceptual or associative knowledge? Evidence from a [3] Breedin SD. Cognitive Neuropsychology 1992. distinctiveness and intercorrelation: analyses of the semantic attributes of living and nonliving concepts. Monoi H. Arch Psychologie 1941. resented in the brain? The case of living category impairment.1214 S. Brain 1999. Levy JI. Naeser M. [8] Folstein MF. Alexander MP. acting and knowing: toward an object. Bransford J. Hodges JR. Patterson K.4:311±38. Gallon CJ. NJ: Erlbaum. 1962. Cognitive a defect at the prephonemic level. Nineteenth Annual Conference of the Cognitive Science Society. From objects to properties: evidence for spreading [31] Simons JS. Nightingale G. whole …t…47† ˆ 2:17. In: Shaw R. Functional neuroanatomy of the semantic system: Ð divisible tening. Cortex 1997. 1977. Cognitive Neuropsychology 1985. Modality-speci®c oper- Neuropsychology 1994. Howard D. [14] Hodges JR. Perceiving.105:271±300. Surface dyslexia and dysgraphia: dual [22] Lambon Ralph MA. Su€olk: Thames Valley Test Company. semantic activation in a case of semantic dementia. Journal of Psychiatric Research 1975. Moss HE. Patterson K.3:497±519. by what? Journal of Cognitive Neuroscience 1998. Ellis AW. Selective impairment in Kanji processing. traumatique. The multiple [24] McGuire S. 1997. [28] Raven JC. Acknowledgements [17] Imura T. Is a picture worth a thousand words? Evidence from concept de®- nitions by patients with semantic dementia. Gogi aphasia or semantic [1] Auerbach SH. Sa€ran EM. Mahwah. Piccini C. East formance in these cases. [21] Lambon Ralph MA. Pyramids and palm trees: a test of arbitrary mapping to meaning.2:49±80.10:766±77. Bury St Edmunds. and so it is important to note Sussex: Lawrence Erlbaum. A. [25] Mummery CJ. submitted. dementia. 1965. Brain 1992. Orchard-Lisle V. / Neuropsychologia 38 (2000) 1207±1215 Sounds. Patterson K. In: Proceedings of the Nineteenth Annual London: H.122:61±73. edi- [9. impairment. that the ratings of familiarity with the sound of an [13] Hodges JR. editors. Aphasie: characteristic symptoms in Japanese. Ellis A. McHugh PR. Neuropsychology. Lucchelli F. in press. Conference of the Cognitive Science Society.28:286±340. Patterson K. Shallice T. Asakawa K. 313±39. London: [7] Durrant-Peat®eld M. submitted. 83±110. Brain 1982. Funnell E. two? In: Allport DA. Advanced progressive matrices sets I & II. Allard T. Hodges JR. 1997. This research is sup. Bub D. editors. Neurocase 1998. Reversal of the concrete. Disrupted temporal lobe connections in semantic dementia. Price CJ. L'examen psychologique dans les cas d'encephalopathie University. [18] Imura T. Graham KS. H. Hodges JR.36:775±84. 1997. 1998. [12] Graham KS. [10] Garrard P. Nihon University Journal of Medicine 1971. Memory Semantic knowledge and episodic memory for faces in semantic 1995. p. Rapp BC. relatively unfamiliar. Patterson K. therefore. on semantic access from pictures and words. Coloured progressive matrices sets A. Patterson K. Oxbury S. Language perception and production: shared mechanisms in lis. [20] Lambon Ralph M.11:617±60. NJ. Psychiatra et Neurologia Japonica 1943. Patterson K. London: Academic Press. 193±8. Wise RJS. The syndrome of gogi (word meaning) practical method for grading the mental state of patients for aphasia. Cognitive [23] Lauro-Grotto R. [16] Howard D. Patterson K. we are not often [11] Gibson JJ. CA: Hillsdale. B. category-speci®c double dissociation. AB. Hodges JR. Folstein SE. [6] De Renzi E.9:209± living and non-living category-speci®c de®cits causally linked to 51. Funnell E. Patterson K. the whole. Romani C. [4] Caramazza A. Are semantic systems separately rep. K. Stanford [29] Rey.7:161±89. 1992. New York: Lawrence Erlbaum sound and seeing the animal or object at the same Associates. Memory 1995. Unlike words. Morton J. [19] Lambon Ralph M. single lexicon.33:593±622. The theory of a€ordances. and the sounds are. dementia? Simulating and assessing poor verbal comprehension Pure word deafness: analysis of a case with bilateral lesions and in a case of progressive ¯uent aphasia. Lost for time. Hodges JR. A test involving the com. Albert ML. 1975. like words.25:627±32. Hodges JR. Prinz W. Neurology clinicians. p < 0:05).115:1783±806. Are routes. have more of an arbitrary relation. Graham KS. Charting the progression item were signi®cantly lower than for the concept as a in semantic dementia: implications for the organisation of semantic memory. p. Hillis AE. Neuropsychology. makes them vulnerable to mild levels of semantic Cognitive Neuropsychology. The facilitation of picture naming in aphasia. Becker JT. [27] Raven JC.12:189±98. tor. with semantic dementia because both have a largely [15] Howard D. The distinc. MacKay W.19] that familiarity is an important predictor of per. Aphasia in Japanese We are indebted to the continuing support of the language. . ship with their meaning than pictures do and this Prototypicality. Howard D. Nogami Y. Patterson K. Brain and Language References 1999. [2] Behrmann M. Semantic bination of sounds and words as stimuli. K. required to access meaning solely from the sound of Hillsdale NY. Reading and writing: one lexicon or Stanford. ness e€ect in a patient with semantic dementia. Scheerer E. Hodges JR. Lewis.30:3±25. Case studies in the neuropsychology of memory. ``Mini-mental state'': a [30] Sasanuma S. Wise RJS. [5] Coltheart M. 67±82. p. [26] Mummery CJ.70:309±35. Naming ported by grants from the Medical Research Council in semantic dementia Ð what matters? Neuropsychologia (UK) and the National Institutes of Health (USA).47:196±218. Lewis. Graham N. In: Neuropsychology 1990. however. Bozeat et al. Graham KS.13:69±90. reading and writing.3:463±95. tionist model. Indeed we know from this and previous studies words: a case of primary progressive aphasia? In: Parkin A. Lambon Ralph MA. dementia: progressive ¯uent aphasia with temporal lobe atro- creates a particularly dicult situation for the patients phy. [9] Funnell E. Systematicity and specialisation in semantics hypothesis: multiple confusions? Cognitive semantics: a computational account of optic aphasia. 1987. We are more accustomed to hearing a an ecological psychology. tiveness of form and function in category structure: a connec. Coslett HB. Plaut DC. ations in semantic dementia. Cortex 1994. Tyler LK. patients included in this study. p. Price CJ. Franklin S.

Brain processes [38] Wechsler DA. Orpwood L. Bozeat et al. In: Neuropsychological Rehabilitation 1998. editors. et al. A standardized set of 260 pic. Goulding PJ. Wechsler memory scale Ð revised. Bury St Edmunds: Thames Valley Test Company. Journal of Experimental Psychology: [36] Warrington EK. hension: a concrete and abstract word synonym test. familiarity.6:174±215. S. Ikeda M. Selective loss of semantic memory for words. [35] Warrington EK. San Antonio: and memory. Nakagawa Y. 1987. Semantic dementia: a form 1991. Quarterly Journal of Experimental Psychology 1975. Yamada N. McGaugh JL. Kazui H. Sakata H.2:167±82. . Vanderwart M. [37] Warrington EK.27:635±57. of circumscribed cerebral atrophy. 141±52.8:143±54. The visual object and space percep- Human Learning and Memory 1980. Hashimoto M. p. Behavioural Neurology 1989. and visual complexity. Neary D. James M. / Neuropsychologia 38 (2000) 1207±1215 1215 [32] Snodgrass JG. Single word compre- [34] Tanabe H. tures: norms for name agreement. tion battery. McKenna P. Ishikawa K. Selective impairment of semantic memory. Psychological Corporation. 1996. image agreement. [33] Snowden JS.