This article was downloaded by: [The University of Manchester Library]On: 09 December 2014, At: 17:30Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK
Neurocase: The Neural Basis of CognitionPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/nncs20
The Heterogeneity of Category-Specific SemanticDisorders: Evidence from a New CaseCristina Rosazza , Emilia Imbornone , Marco Zorzi , Elisabetta Farina , Leonora Chiavari &Stefano F. CappaPublished online: 09 Aug 2010.
To cite this article: Cristina Rosazza , Emilia Imbornone , Marco Zorzi , Elisabetta Farina , Leonora Chiavari & Stefano F.Cappa (2003) The Heterogeneity of Category-Specific Semantic Disorders: Evidence from a New Case, Neurocase: The NeuralBasis of Cognition, 9:3, 189-202
To link to this article: http://dx.doi.org/10.1076/neur.9.3.189.15557
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
The Heterogeneity of Category-Specific SemanticDisorders: Evidence from a New Case
Cristina Rosazza1, Emilia Imbornone2, Marco Zorzi1, Elisabetta Farina2, Leonora Chiavari2
and Stefano F. Cappa1
1Department of Psychology, Universita Vita Salute San Raffaele, Milan, Italy and 2Unita di Neurologia Riabilitativa, IRCCS SantaMaria Nascente, Fondazione Don Carlo Gnocchi Onlus, Milan, Italy
Abstract
We report a new case of category-specific semantic impairment, affecting living entities, in a patient with traumatic braindamage. In the present investigation we attempted to replicate as closely as possible the testing procedures whichhave been developed by Caramazza and Shelton (1998) to evaluate EW, a patient with a selective semantic disorder forthe animal category. The results in our patient indicated a different performance profile, characterised by a moreextensive semantic disorder for living entities, and by a more severe loss of specific visual rather than functionalknowledge. These findings concur with other evidence indicating that category-specific semantic disorders areheterogeneous, reflecting different mechanisms of impairment, most likely associated with different neurobiologicalunderpinnings.
Introduction
Category-specific semantic impairments have attracted con-
siderable attention in neuropsychology because of their con-
tribution to the understanding of the organisation, of the
mechanisms and of the neuroanatomical bases of semantic
memory. Various patterns of deficit have been pointed out.
Coltheart et al. (1998) suggested that selective deficits of
semantic memory can be classified according to three distinctclasses: (i) category-specific semantic impairments, (ii) input-
modality specific semantic impairments and (iii) attribute-
specific semantic impairments, with some patients showing
more than one type of selectivity. Among the examples of the
first class there are impairments related to the distinction
between abstract and concrete words (Warrington, 1975),
as well as the most frequently reported dissociation between
knowledge of living things and man-made artefacts(Warrington and Shallice, 1984; Laiacona et al., 1997; Cappa
et al., 1998; Caramazza and Shelton, 1998; Samson et al.,
1998; Gainotti, 2000). In patients belonging to the second
class of selective semantic impairment, the ability to perform
semantic tasks depends on the modality of stimulus input, i.e.
for example, pictures or words (McCarthy and Warrington,
1988). Finally, attribute-specific semantic impairments are
characterised by the patient’s inability to retrieve specificsemantic attributes in semantic memory (for example, visual
information about objects), whereas other semantic attributes
are accessible (Coltheart et al., 1998). Although patients with
an isolated, selective attribute impairment are rare (Coltheart
et al., 1998; Lambon Ralph et al., 1998), there are some cases
of combined attribute-categorical impairments. For example,
Michelangelo (Sartori and Job, 1988), L.A. (Silveri and
Gainotti, 1988), Giulietta (Sartori et al., 1993) and Felicia
(De Renzi and Lucchelli, 1994) represent cases of category-
specific deficit restricted to living things in association withattribute-specific impairments for visual knowledge.
As detailed above, data from the literature agree that dif-
ferent semantic categories can be damaged in isolation, but
several explanations have been suggested to account for them.
These explanations can be divided into two broad categories:
reductionist theories and non-reductionist theories. Among the
reductionist theories, the first to be proposed was the sensory/
functional theory (SFT): Warrington and Shallice (1984)suggested that semantic memory is organised by modality
(visual, olfactory, motor/functional . . . ), i.e. according to the
type of semantic information, rather than category per se.
According to this account, visual (sensory) and functional
features have different weights in the identification of mem-
bers of living and non-living categories, respectively: as a
consequence, damage to visual semantic subsystem results in
impairment of living things, whereas damage to the functionalsubsystem results in impairment of non-living things.
Recently, Moss, Tyler and colleagues proposed another reduc-
tionist model (Durrant-Peatfield et al., 1997; Moss and Tyler,
Neurocase 1355-4794/03/0903–189$16.002003, Vol. 9, No. 3, pp. 189–202 # Swets & Zeitlinger
Correspondence to: Stefano F. Cappa, M.D., Vita Salute San Raffaele S. Raffaele University, DIBIT Via Olgettina 58, 20132 Milan, Italy.Tel: þ39 0226434887 (secr 4784); Fax: þ39 0226434892; e-mail: [email protected]
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
2000). They emphasise the concept of intercorrelation between
perceptual and functional features, and the different role of
shared semantic properties versus distinct semantic properties.
A similar proposal had been already put forward by De Renzi
and Lucchelli (1994). Their patient had a deficit in the retrieval
of perceptual attributes. Her performance with artefacts was
better, according to the authors, because non-living items can
access their structural representation, since shape and functionare in a close correspondence. Similarly, Laiacona et al. (1997)
claimed that living entities are more vulnerable than non-living
ones because perceptual and functional properties show a
lesser degree of correlation. Moss and Tyler’s proposal is more
articulated. They suggest that for living things the shared
functional (biological) properties (e.g. can see, can run, can
hear, etc.) and the shared perceptual properties (e.g. has eyes,
has legs, has ears) are highly intercorrelated. On the other hand,the distinctive properties (e.g. has a mane, is pink, chases mice)
tend to be weakly correlated and therefore very vulnerable to
damage. For artefacts, the pattern is reversed: non-living items
have strong correlations between pairs of individual, distinc-
tive form and function properties (e.g. has a serrated edge – can
cut), whereas the shared properties are fewer and less
correlated than those of living things. When the corresponding
neural network model is lesioned by a random removal ofconnections, category-specific impairments can arise: with
mild degrees of damage, non-living entities are less affected,
because of the presence of the strong form-function intercor-
relations among the distinctive features of these items. With
more severe levels of lesioning, artefacts are more affected
because the model can only operate on shared properties and
living items are more resistant because they are supported by a
greater degree of shared, intercorrelated properties.An influential, non-reductionist theory is the domain-specific
knowledge hypothesis proposed by Caramazza and Shelton
(1998). According to these authors, semantic memory is orga-
nised categorically in the brain and separate neural systems may
have become specialised, under evolutionary pressures, for the
recognition of animals (that are potential predators and a source
of food) and plants (that are a source of food and medicine). On
this view the only ‘‘true’’ category-specific deficits will be thosethat involve the category of animals, plant life and by contrast
artefacts. These, and only these, three categories form the basis
for the organisation of conceptual knowledge.
Problematic findings exist for each of these theories. The
reductionist theories have clear limitations. As for the sen-
sory-functional theory (SFT), patient EW reported by
Caramazza and Shelton (1998) showed a deficit restricted
to the category of animals, whereas she was equally impairedon functional and perceptual notions about animate entities.
There was no evidence that the two patients studied by
Laiacona et al. (1997) with a deficit for living entities showed
a greater loss of visual than functional information. Further,
there is some evidence of cases with defective visual knowl-
edge that do not display a category-specific effect (Coltheart
et al., 1998; Lambon Ralph et al., 1998). These cases run
against the SFT.
Moreover, the model of Moss, Tyler and colleagues
(Durrant-Peatfield et al., 1997; Moss and Tyler, 2000), in
which an initial advantage for non-living entities is replaced
by an advantage for living ones, has not been empirically
supported (Garrard et al., 2001).
On the other hand, the domain-specific knowledge hypoth-
esis does not account for a number of cases of category
specific impairments restricted to living things in whichperceptual information was specifically lost (Sartori and
Job, 1988; Silveri and Gainotti, 1988; Sartori et al., 1993;
De Renzi and Lucchelli, 1994), as well as for the impairment
of categories which do not have evolutionary significance,
such as musical instruments, body parts or gems (but see
Caramazza and Shelton, 1998, for a criticism of the evidence).
Different hypotheses have been proposed in order to account
for category-specific disorders, but only a few efforts havebeen made to capture the heterogeneity of the cases described
in the literature (Humphreys and Forde, 2001). Each hypoth-
esis tends to focus on selected cases, without taking into
account the differences among patients at the psycholinguistic
(categories impaired, task effects) and neural level (aetiology,
localisation of the brain lesion).
At this stage an important question is still open: does a
unique interpretative hypothesis apply to each patient withcategory-specific semantic impairment? It is critically impor-
tant to understand if variability gives us clues to the different
mechanisms of impairment or if it is only the effect of
nuisance variables, which obscure an underlying unique
mechanism. Another important factor to take into account
from this point of view is the way in which the patient is
tested. It is possible that different methodologies affect the
results, because of differences in processing requirements,insufficiently detailed assessment or inadequate control of
nuisance variables.
With this problem in mind, we decided to submit a brain-
damaged patient, MA, with a deficit for living things, to the
same testing procedures employed by Caramazza and Shelton
(1998). The reasons for this choice were twofold. In the first
place, the procedures employed in this case study are excep-
tionally detailed and well-controlled. Second, the results ofthe investigation, which showed a selective disorder for
animals, without any differential impairment in perceptual/
functional knowledge, led the authors to reject reductionist
approaches, in favour of the ‘‘domain-specific’’ account.
The aim of this study is to replicate the methodology used by
Caramazza and Shelton (1998) to another case of category-
specific semantic impairment in order to assess, first, whether
their results could be generalised to our case and second,whether there is theory, either reductionist or non-reductionist,
that could explain the different patterns of category-specific
deficits shown in the literature, included ours.
Case report
In July 1999 MA, a 28-year-old, left-handed student, who was
about to take the final examination for his engineering degree,
190 C. Rosazza et al.
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
had a serious motorbike accident and suffered severe facial
and head trauma. A CT scan immediately after admission to
hospital showed multiple facial and right orbital fractures and
a large right frontal haematoma.
He was given to emergency surgery to reconstruct the facial
and orbital fractures; he remained in a coma for eight days andthis was followed by gradual neurological improvement. One
month later he had another operation to further reconstruct the
facial fractures, which had severely affected the right orbital
cavity, and to evacuate a small left frontal hygroma. A follow-
up CT showed a small right frontal hygroma.
At the end of August MA was discharged to a rehabilitation
hospital; there he started to communicate verbally, even if
spontaneous speech was reported to be limited; he did notshow motor deficits.
In November 1999 MA was given to an initial neuropsy-
chological examination at the Neuropsychology Laboratory
of the University of Brescia. He was disoriented in time and
place, and was severely amnesiac for past as well as recent
memories concerning personal and public events. He could
read words and sentences, but not numbers and he made
several mistakes at sums.
In December he went back home; his mother reported
that he was improving continuously, even if his temper
had changed: he was calmer and he always needed to be
stimulated.
In February 2000, he was admitted to Don Gnocchi Day
Hospital, where he was given a complete neuropsychologicalassessment (see below). An EEG was characterised by diffuse
slow activity. A CT scan showed a hypodense right frontal
area, a small bilateral frontal hygroma, and multiple small
frontal bilateral hypodensities (Fig. 1).
On the first of March he began a cognitive rehabilitation
program. A follow-up EEG in June showed a reduction of the
slow activity that was restricted in the right temporal regions.
Single photon emission tomography (SPECT) showed a re-duction of perfusion in the right frontal lobe, as well as in the
left mesial temporal region.
Unfortunately, the etiology of brain damage (severe head
injury) prevents any detailed consideration of the anatomical
locus of damage. While the location of structural lesions and
functional effects appear to be at variance with other patients
with similar neuropsychological profiles, it is likely that
axonal damage may have affected other brain regions.
Fig. 1. CT scan sections showing hypodense right frontal area and multiple small frontal bilateral hypodensities.
Category-specific semantic disorders 191
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
Neuropsychological assessment
MA was given two complete general neuropsychological
examinations in February and in June 2000. A summary ofthe neuropsychological tests administered in the two sessions
is reported in Table 1. At the first assessment the patient was
well oriented in time and place: he was given the Mini-Mental
State Examination (Folstein et al., 1975) on which he
obtained a raw score of 27 and a corrected score of 25.03
(cut off 24). He performed well on Raven Progressive
Matrices (Raven, 1962), scoring 26.5 (raw score: 29; cut
off 18). Language was appropriate, although spontaneousspeech was poor; reading and writing were normal, but
slower. MA performed slightly below the normal range on
Token Test (De Renzi and Faglioni, 1978): raw score: 29,
corrected score: 25.75; cut off 26.5. His performance was
normal or marginally defective on oral and written calculation
and number processing examinations. He scored within the
normal range on short-term verbal memory (digit span for-
ward and backward, 5 and 3 respectively) as well as spatialmemory (Corsi raw score: 5, corrected score: 4.5; cut off:
3.75). The patient showed no apraxic deficits and he was able
to copy drawings: on the Rey–Osterreith figure copy
(Osterrieth, 1994) he scored 31.87 (raw score 33; cut off:
28). MA was severely impaired in all the tests that challenged
his long-term memory abilities. He got zero scores on Logical
Memory (De Renzi et al., 1977), on Rey–Osterreith figure
recall and on Rivermead Behavioural Memory Test (Wilson
et al., 1985). His performance on the Naming Test and
Questionnaire (Laiacona et al., 1993b) was severely impaired
with living (37% and 61% correct, respectively) and non-
living items (67% and 87% correct, respectively). This last
result confirmed that his semantic memory was severelyaffected; his impairment was already evident at informal
questioning. MA was not able to define a large set of well-
known living and non-living items, e.g. hippopotamus and
pulley. His categorical fluency was defective as well as his
frontal lobe functions, tested with the Wisconsin Card Sorting
Test (Grant and Berg, 1948) and the Weigl test. Impairment
was also evident on attentional tasks (Attentive Matrices and
Trail Making Test). On Benton Face Recognition test (Bentonet al., 1992) his performance was borderline (39, cut off: 39).
In June 2000, MAwas given to a second assessment. He had
improved on MMSE (raw score: 29; corrected score: 27.03;
cut off: 24) and on the Raven Test (raw score 34; corrected
score: 30; cut off: 18), and he achieved normal scores on the
Token Test (raw score 34; corrected score: 30.75; cut off:
26.50). His attention improved as well: he scored within the
normal range on Attentional Matrices and on test B and B-A
Table 1. Neuropsychological tests administered in the two sessions (February and June 2000)
First assessment Second assessment
Raw score Corrected (cut off) Raw score Corrected (cut off)
MMSE 27 25.03 (24) 29 27.03 (24)Raven 29 26.15 (18) 34 30 (19)Token test 29 25.75� (26.5) 34 30.75 (26.5)
Digit spanforward 5 6backward 3 4
Corsi’s span 5 4.5 (3.75) 4 3.5� (3.75)
Rey figurecopy 33 31.87 (28)recall 0 0� (6.2)
Logical memory 0 0� (8)Autobiographical memory 16/45�Rivermead 26/93 5/24� (18)
WCSTCorrected R 0 25Categories 0 �1� 0 �1�Persevarative errors 63 < 1� 19 5�
Weigl 5 1.5� (4.5) 8 4.5a (4.5)Categorical fluency 23� 15 (25)Phonemic fluency 35 25 (17)Attentional matrices 34 19.25� (31) 47 32.25 (31)
Trail Making TestA 10900 12500� (9300) 9700 113 (9300)B 41300 47200� (28200) 18500 24400 (28200)A-B 30400 34700� (18600) 8600 12800 (18600)
Benton’s faces 39a
Famous faces 25/45�
�Impaired performanceaBorderline.
192 C. Rosazza et al.
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
of Trail Making Test. Rey–Osterreith figure copy was perfect
(36/36), however, MA continued to underestimate his condi-
tion and was deeply anosognosic. Frontal lobe functions were
defective on categorical (but not phonemic) fluency. His
performance was borderline on the Weigl Test (raw score:
8, corrected score: 4.5; cut off: 4.5) and still impaired on the
WCST. On short memory tests, he showed a slightly worse
performance on the Corsi test (raw score 4; corrected score:3.5; cut off: 3.75), but an improvement on digit span forward
(6) and backward (4). By and large, his deficits beyond
semantic memory were restricted to long-term episodic mem-
ory. His performance on Logical Memory (0 scores) as well as
on an autobiographical memory questionnaire (Borrini et al.,
1989) (16/45) was severely impaired. He was also severely
defective in naming faces of famous people (25/45). In
summary, at this time there was no evidence of mentaldeterioration or language disturbance. MA showed a severe
amnesia and naming and recognition deficits.
Experimental study
Control group
Five normal male subjects, matched for age (mean age¼ 26
years), gender, education level (18 years of schooling) and
type of studies (one of them had just graduated in engineering
and the others were about to conclude their engineering
degree, as MA) were given the test specifically constructed
for the present study.
Experimental procedures
In this study we carefully replicated the procedures used byCaramazza and Shelton (1998) to investigate MA’s naming
and recognition deficit, through a number of specific tasks,
prepared ad hoc. In particular, for each of the following tasks,
all subjects were asked to decide whether statements about
objects were true or false, and then we selected only those
answers that 4/5 control subjects gave correct, removing
wrong and ambiguous sentences. Only the standardised tests
as the VOSP (Warrington and James, 1991), the BORB(Riddoch and Humphreys, 1993) and the naming test (Lotto
et al., 2001) were not given to the control group. This
experimental study has been carried out in September
2000, during the second assessment.
Task 1 – visuo-perceptual abilitiesGiven the severe ocular damage, we deemed necessary to
perform an in-depth neuro-ophthalmologic examination todetermine the integrity of MA’s basic visual abilities. Visual
acuity was in the normal range and his perceptual abilities
assessed with colour tests (colour naming, recognition and
grouping) were perfect. Lang Test revealed lack of stereopsis.
MA has a restriction of visual field in both eyes, which he
compensates by moving his head and eyes.
The VOSP (Warrington and James, 1991) was administered
to investigate his performance on visual perceptual tasks more
comprehensively: he achieved normal scores on all the subset
except on Silhouettes. Some tests of the BORB battery
(Riddoch and Humphreys, 1993) were administered as well:
his normal performance on test 5 (33/40) ruled out appercep-
tive agnosia. Test 7, 8, 11 and 12 were given to assess his
ability to perform complex visual matching and categorisation
tasks: MA performed within the normal range on all the tasks.
CommentWe can conclude that MA’s impairment is not the result of
defective vision or impaired visual processing of complex
objects.
Task 2 – naming picture and controllingfor nuisance variablesThe patient was given in random order 266 standardisedpictures in black and white from the Lotto et al. (2001)
set, which includes 13 categories i.e. mammals, birds, fruits,
vegetables, flowers, housewares, buildings, vehicles, furni-
ture, clothes, weapons, musical instruments, receptacles and
mix. Examples from mix category are: candle, radio, pipe,
rucksack, alarm etc.; from this group we formed a fourteenth
category of tools. Naming was evaluated again about one
month later.
ResultsThe patient performance is summarised in Table 2. A response
was considered correct if MA provided the Lotto et al.’s
(2001) name or any acceptable response produced by subjects
more frequently than the correct name, according to norma-
tive data.
MA showed a clear naming impairment, affecting mostitem categories. However, he showed an overall category
effect, with living items more affected than non-living items.
Table 2. Picture Naming task. Percent of MA’s correct responses
Categories 18 Naming 28 Naming
Mammals 12/21 (57%) 16/21 (72%)Birds 6/22 (27%) 8/22 (39%)Tot. animals 18/43 (42%) 24/43 (56%)
Fruits 13/21 (62%) 13/21 (62%)Vegetables 6/17 (35%) 5/17 (29%)Flowers 1/14 (7%) 2/14 (14%)Tot. plants 20/52 (38%) 20/52 (38%)Tot. living 38/95 (40%) 44/95 (46%)
Buildings 13/20 (65%) 13/20 (65%)Clothes 25/29 (86%) 26/29 (90%)Tools 3/6 (50%) 4/6 (66%)Vehicles 16/25 (64%) 17/25 (68%)Furniture 7/12 (58%) 9/12 (75%)Houseware 11/11 (100%) 10/11 (91%)Musical Instruments 8/15 (53%) 10/15 (67%)Receptacles 5/11 (45%) 6/11 (54%)Weapons 5/16 (31%) 6/16 (37%)Mix 17/26 (65%) 20/26 (77%)Tot. inanimate 130/223 (58%) 141/223 (63%)Tot. non-living 110/171 (64%) 121/171 (71%)
Category-specific semantic disorders 193
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
In order to assess whether MA was impaired in processing
living things or the more restricted category of animals, data
were entered into a logistic regression analysis with response
accuracy as the dependent variable, and category membership
animate/inanimate and living/non-living as independent vari-
ables. The results revealed that animate did not have a
significant effect on performance in both evaluations. On
the contrary, the living factor had a significant effect in thefirst naming (Wald¼ 9.627, p< 0.002) as well as in the
second naming (Wald¼ 17.653, p< 0.0001), mainly due to
the severely defective performance with items belonging to
the categories of birds, vegetables and flowers.
We also compared the separate sets of animals, plant life
(fruit, vegetable, flowers and plants) and non-living items, by
entering 4 nuisance variables in order to ensure that living/
non-living dissociation with both animals and plants was notdue to uncontrolled stimulus factors. So we performed a
logistic regression analysis by entering response accuracy
as the dependent variable, and animals and plant life together
with other four additional factors i.e. name frequency, concept
familiarity, typicality and age of acquisition, as the indepen-
dent variables. The outcome of this analysis indicated a
significantly inferior performance on both animals and plant
life on the first naming (Wald¼ 12.37, p< 0.0004 andWald¼ 11.126, p< 0.0009, respectively) as well as on the
second naming (Wald¼ 7.509, p< 0.006 and Wald¼ 18.558,
p< 0.0001, respectively).
Age of acquisition has been shown to play an important role
in both naming performances (Wald¼ 21.007, p< 0.0001 and
Wald¼ 19.272, p< 0.0001). Neither familiarity, nor fre-
quency, nor typicality appear to have a significant effect on
both naming performances. The same regression analysis, withanimals, plants and nuisance variables, but without the mix
category belonging to the non-living domain showed the same
results. A significant effect of animals and plants on the first
naming (Wald¼ 11.012, p< 0.0009 and Wald¼ 10.814,
p< 0.001) as well as on the second naming (Wald¼ 6.602,
p< 0.10 and W¼ 17.299, p< 0.0001) and a significant effect
of age of acquisition on both naming performances
(W¼ 16.810, p< 0.0001 and W¼ 19.206, p< 0.001) wereobserved. This further analysis confirmed the living effect
also when the unspecified collection of items, i.e. the mix
category, was excluded.
We can definitely say that naming performance appeared to
be significantly worse for living than non-living entities once
frequency, concept familiarity, typicality and age of acquisi-
tion were accounted for. Thus, whereas age of acquisition had
a powerful influence on naming performance, it cannot byitself account for the category effect.
An analysis of erroneous responses indicated that when
MA is unable to name a stimulus correctly, his verbal
responses are consistently different according to the nature
of the stimulus. When the picture represents a non-living item,
he describes its function, but in presence of living items, he
gives either the name of superordinate category, or a semantic
paraphasia (coordinate), but never a perceptual description.
CommentNaming results show that MA’s deficit concerns the category
of living things (plus the category of weapons), not only the
category of animals, as in the case reported by Caramazza and
Shelton (1998). MA’s naming performance cannot be
explained by nuisance variables such as name frequency,
concept familiarity, typicality and age of acquisition since
MA’s difficulties persist even when these factors are con-trolled. The deficit for the non-living category weapons is
difficult to interpret: items within this group are very different
from each other, both from the perceptual and the functional
point of view. Examples of unrecognised items are: bomb,
crossbow, sword, whip, machine gun, sling and bow. He made
some anomic errors across the two sessions (bomb, sling,
crossbow and whip on both the submissions and bow), as
indicated by his ability to provide detailed information aboutweapons’ function (e.g. sling> ‘you put a stone and you
throw it’; crossbow> ‘it launches spears’; bomb> ‘you fire
it at enemies, you kill a lot of people’).
Task 3 – categorical fluencyMA and control subjects were asked to name in 1 min as many
items as possible from each of 14 categories used in the naming
task (mammals, birds, fruits, vegetables, flowers, kitchenware,buildings, vehicles, furniture, clothes, weapons, musical
instruments, receptacles and tools), plus other 4 categories:
insects, sea animals, animals (in general), and body parts.
ResultsData reported in Table 3 show a general impairment in all the
categories. This result is probably due to the coexistence of a
frontal deficit with a semantic deficit.
Task 4 – discrimination between real and unrealanimals and real and unreal objectsParticipants were asked to discriminate between pictures of
real objects from pictures of non-existing ones. Such an object
Table 3. Categorical fluency task. Number of MA’s and controls’ responses
Categories Controls range MA
Mammals 15–28 6Birds 18–24 4Insects 8–11 2Sea animals 15–27 3Animals (in general) 24–27 14Fruits 20–24 7Vegetables 16–22 4Flowers 8–19 1Buildings 11–25 3Clothes 22–30 5Tools 13–17 3Vehicles 21–24 7Furniture 14–19 3Kitchenware 15–21 3Musical instruments 19–23 5Receptacles 11–22 2Weapons 15–25 3Body parts 30–32 9
194 C. Rosazza et al.
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
decision task is intended to directly investigate structural
knowledge about objects. A set of 33 pictures was con-
structed, 18 of animals (11 unreal versus 7 real) and 15
objects (9 unreal versus 16 real). Unreal pictures were made
up by joining parts from two different animals or two different
objects to create plausible but non-existent items. In order to
test whether the sets of animals and artefacts were of compa-
rable level of difficulty, the stimuli were initially given to thecontrol subjects: a picture of an animal was then removed.
Moreover MA was submitted to test 10 of the BORB
(Riddoch and Humphreys, 1993), an object decision task,
requiring him to distinguish between drawings of real objects
and drawings of imaginary objects.
ResultsOn the set of pictures we constructed, MA’s performance wasslightly more impaired with animals (13/17) rather than non-
animals (14/15). Although the difference was not significant, a
trend to accept existing items as unreal is present. Controls did
not show any difference. On test 10 of the BORB, he scored 24/
32, 20/32, 22/32 and 22/32, showing a general impairment in
distinguishing real items from unreal ones. Moreover in order
to analyse the difference in MA’s performance between ani-
mals and objects, we combined the 32 pictures we created withBORB stimuli, since BORB items have fewer object pictures
than animal pictures. Results have shown that the difference is
significant (X2¼ 6.828, p< 0.009) and that the patient tends to
reject real animals. Since items were unbalanced to animals’
advantage, we also performed a logistic regression analysis: it
confirmed the previous results, namely, a defective perfor-
mance with animals (Wald¼ 6.123, p< 0.013).
Task 5 – part decision taskPictures from Lotto et al. (2001) and from Snodgrass and
Vanderwart (1980) were used to create a set of 42 stimuli, 20
animals and 22 non-animals. Each ‘‘body’’ with a part missing
was presented to the participants and arranged on a sheet of
paper with two ‘‘heads’’ (one target and one distracter);
participants were asked to match the ‘‘body’’ with the correct
choice. Two trails were performed. Pictures were balanced sothat each ‘‘head’’ was correctly paired with its ‘‘body’’ on one
trail and incorrectly paired with another ‘‘body’’ on the
second trial. The position of the target was varied over the
stimulus chosen; examples are presented in Fig. 2. No pictures
were removed after having given them to the control group.
ResultsMA was only slightly impaired in selecting the correct head;however, his performance was worse with animals (17/20),
than with objects (21/22), although the difference was not
significant. Only one control subject made a mistake with one
animal item.
CommentResults from object decision task, BORB and part deci-
sion task show that MA has defective processing of visual
information about animals and objects. In comparison to EW,
the disorder of recognition appears to be milder, and less
selective.
Assesment of semantic knowledge aboutliving vs. non-living attributes
Up to now we have shown that MA has an impairment in
naming and recognising, which affects more severely living
entities, and is not caused by a general visual processing
deficit, but appears to involve some loss of visual properties
(structural knowledge). Here we provide evidence that MA’sdeficit is due to a disruption of semantic memory and that the
patient is more impaired in judging if a living, as compared to
a non-living, entity has a particular attribute. MA was given
seven extensive questionnaires to evaluate his conceptual
knowledge about living and non-living properties. Question-
naire statements were divided into visual/perceptual versus
associative/functional in order to assess whether MA is more
impaired in retrieving the visual attributes of living entities.
Task 1 – questionnaire on central attributesIn this task 42 animals (20 mammals, 9 birds, 6 seabirds, 5
insects, spider and turtle), 15 fruits, 15 vegetables and 42
artefacts (10 tools, 10 vehicles, 10 furniture, 9 clothes and 3
kitchenware) were selected to test MA’s knowledge about the
most central attributes of the stimuli. By central attribute, we
mean the most representative features of each item, selectedon the basis of a pre-test. In this pre-test, between 4 and 16
attributes for each of 114 stimuli were determined (mean
number of attributes per stimulus¼ 8.3). Twenty normal
subjects (matched as much as possible for age and scholarship
for MA) were asked to rate how much the attributes, divided
into visual and functional, were representative of the stimulus
Fig. 2. Examples of the items used in the Part decision task.
Category-specific semantic disorders 195
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
at issue. From these judgements, we used the attributes that
the largest number of subjects agreed to be ‘‘central’’ in
defining the stimulus.
The questionnaire was given to the 5 control subjects,
after 30 statements were removed; the task consistedof 433 statements, 209 describing visual properties (e.g. a
giraffe has a long neck), 224 functional properties (e.g. wine is
made from grapes), half were false, the other half were true.
Participants were asked to say whether a statement about an
object was true or false.
ResultsA logistic regression was performed, using the responseaccuracy as the dependent variable and the binary variable
living/non-living as the independent factor. Results showed
that MA has difficulties in answering statements about living
entities (Wald¼ 8.924, p< 0.028). Control subjects do not
show any difference. As can be seen in Table 4, MA’s
performance tends to be significantly worse with false ques-
tions than with true ones, namely he tends to accept false
statements as true.Moreover, by entering in the regression analysis living/non-
living, perceptual/functional and true/false as independent
factors, only living (Wald¼ 4.706, p< 0.030) and true/false
(Wald¼ 9.808, p< 0.002) variables were significant.
Although perceptual features were more impaired than func-
tional ones, the difference was not significant (Wald¼ 1.7,
p< 0.19). The results have been confirmed by another regres-
sion analysis: we divided the questions into individual cate-gories, i.e. animals, plant life (fruits and vegetables) and
objects. We entered animals, plants, perceptual/functional
and true/false into a logistic regression and results showed
again that animals (Wald¼ 7.339, p< 0.007), plants
(Wald¼ 4.706, p< 0.03) and true/false (Wald¼ 9.808,
p< 0.0017) were significant.
CommentResults show that MA has a significant deficit of knowledge
for central attributes of animals, fruit and vegetables.
Task 2 – questionnaire on food/non-food animalsIn this task 16 food animals (duck, lobster, horse, deer, rabbit,
mussel, hen, cook, pig, cow, goose, sword-fish, ostrich, turkey,
turtle and sheep) and 26 non-food animals (bee, eagle,
donkey, whale, caterpillar, camel, dog, kangaroo, sea horse,
elephant, butterfly, ant, cat, giraffe, lion, fly, bear, robin,
spider, rhino, squirrel, shark, tiger, mouse, zebra and pigeon)
were selected. After the stimuli were given to the control
subjects, 3 items were removed (sheep, pigeon and turtle).
Participants were asked to tell if the animal or one of its parts
is eaten according to our culture.
ResultsMA had severe difficulties answering questions concerning
whether or not the animal was a food animal (X2¼ 10.363,
p< 0.001). He could recognise food animals, but he made
serious mistakes with non-food ones, saying that dog, ant,
spider, cat, butterfly and mouse can be eaten. This finding
indicates that there is a considerable loss of functional infor-
mation about the animal category.
Task 3 – questionnaire on specific-general attributesThis test was created in order to investigate the level of
knowledge that is damaged in MA. Knowing a general
feature, for instance that something has four legs or breathes,
is sufficient to categorise it as an animal, but it is does not help
in deciding whether it is a dog or an elephant. Following
Caramazza and Shelton (1998), we selected 69 animals (21mammals, 16 birds, 14 insects, 17 sea animals like whale,
octopus, shark, swordfish, mussel) and 45 non-animals (15
fruits, 15 vegetables and 15 buildings) to test MA’s knowledge
about general and specific attributes of stimuli. By general, we
mean those features of a category shared by all or most
members of that category (e.g. cats have eyes); by specific,
we mean distinctive attributes of one or a few members of the
category (e.g. a swimming pool has a springboard). The initialquestionnaire included 789 statements; after having submitted
them to control subjects, the task consists of 724 questions,
419 of which specific (210 perceptual and 209 functional) and
305 general (136 perceptual and 169 functional), matched for
true and false.
ResultsPerformance is summarised in Table 5. The results of thelogistic regression analysis revealed that MA has greater
difficulties with animals (Wald¼ 4.835, p< 0.028), while
his performance with plants was not significant (Wald¼0.322, p> 0.685). This difference in comparison with the
results of other tests is most likely due to the fact that, in order
to follow the procedures used by Caramazza and Shelton
(1998), that were directed only to test conceptual knowledge
about animals versus non-animals, we included a smallnumber of non-living items (15) compared to living ones
(99). This might have resulted in a reduced sensitivity of the
task in the case of plant stimuli. Questions about animals,
plants and objects were heavily unbalanced (433, 196, 95,
respectively). We also performed a logistic regression analysis
by entering, as independent variables, animals, plants, percep-
tual/associative and true/false; results revealed that animals
have a significant effect (Wald¼ 5.01, p< 0.025) and that a
Table 4. Central Attributes Questionnaire: proportion of MA’s correct
responses
Animals Inanimate Living Non-living
PerceptualTrue 26/37 (0.70) 69/79 (0.87) 53/70 (0.76) 42/46 (0.91)False 26/43 (0.60) 33/50 (0.66) 40/66 (0.61) 19/27 (0.70)
FunctionalTrue 28/39 (0.72) 68/76 (0.89) 56/70 (0.80) 40/45 (0.89)False 34/43 (0.79) 47/66 (0.71) 46/66 (0.70) 35/43 (0.81)
196 C. Rosazza et al.
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
trend for perceptual properties is present (Wald¼ 3.45,
p< 0.063). The results of a logistic regression performed
with animals, perceptual/functional and specific/generalshowed that MA’s performance is much worse with specific
attributes than general attributes (Wald¼ 16.086, p< 0.0001)
and with animals (Wald¼ 8.948, p< 0.0028); a trend for
perceptual properties is still present (Wald¼ 2.756,
p< 0.097). Control subjects did not show any significant
differences.
Looking at the only general statements, no significant
effects exist. Looking at the only specific statements, onecan note that MA has greater difficulties in answering ques-
tions about animals (Wald¼ 3.59, p< 0.058). His perfor-
mance is significantly worse with perceptual attributes as
compared with functional attributes (Wald¼ 8.345,
p< 0.004) when we enter into logistic regression animals,
perceptual/functional, plants and true/false. The last two
factors were not significant. By entering only plants and
perceptual/associative, results show that plants are not sig-nificant, but perceptual properties are more impaired than
functional ones (Wald¼ 8.09, p< 0.004). Multinomial logis-
tic regression did not show any interactions. Again, control
subjects did not show any significant difference.
CommentSince the living group was about 7 times more numerous than
non-living group, in this task it was not possible to show
a significant impairment for living entities. MA is more
impaired in answering statements about animals, and his
performance appeared to be basically worse with perceptual
attributes. Moreover MA has no difficulty with questions
concerning general attributes, while his knowledge of specific
attributes is significantly damaged. When we considerjust specific statements, MA appeared to be significantly
more impaired in retrieving visual attributes than functional
attributes. However, MA has difficulties with perceptual
properties of all the stimuli, not only with those of animate
entities.
Task 4 – questionnaire on shared attributesCaramazza and Shelton (1998) created this task to rule out the
possibility that their patient’s deficit for animals reflected a
selective deficit in processing certain properties that may be
differentially distributed across semantic categories. Namely,
they wanted to rule out the possibility that the problem
concerns attributes themselves and not the semantic category
of animals.
Task 4a – questionnaire on size judgementsWe selected 21 pairs of animals, 20 pairs of fruits and
vegetables and 20 pairs of objects. For each pair a precise
discrimination was required to judge item sizes (e.g. sheep-deer, carrot-cob, spoon-ladle). No items were removed after
the submission to control subjects. Participants were asked to
indicate which item of the pair was bigger.
ResultsMA made 5 mistakes, all with fruits and vegetables. Fruits and
vegetables are significantly more impaired than animals and
objects, categories without mistakes.
Task 4b – questionnaire on other shared attributesWe selected 7 attributes (big, small, number of legs, surface
texture, short, tall, colour) that are shared between animals
and non-animals, and 12 items, 6 of which animals and 6
artefacts. Each item that had one of these attributes was paired
with another item not having this attribute (e.g. an item was
rhino: 1. is it bigger than a sheep? 2. is it smaller than anelephant? 3. does it have smoother skin than a rabbit? etc.).
Two questions were removed because they were not appli-
cable to each object (e.g. numbers of legs) and two questions
were removed after having been given to the control group. So
the task consisted of 80 questions, 40 on animals and 40 on
non-animals. Participants were asked to answer questions in
the affirmative or not.
ResultsMA made 4 mistakes with animals and 3 mistakes with non-
animals. The difference is not significant.
CommentMA’s preserved ability to indicate the bigger animal of
the pair seems unusual, but other cases are described in the
literature. Michelangelo, the patient reported by Sartori andJob (1988), showed a deficit affecting living entities, a greater
impairment for perceptual attributes, an inability to distin-
guish real animals from unreal ones, but his performance on
size judgements was perfect (100%). A case of a patient
(Jennifer) with a deficit for living things, reported by Samson
et al. (1998), did not have difficulties on this kind of task.
The authors proposed that size might be a general physical
attribute that is not very discriminating. Coltheart et al.(1998) suggest that size is a semantic attribute that is not
perceptual. The patient AC they studied had a selective deficit
for visual attributes of stimuli, but he performed very well on
size judgement task (20/24). According to the authors, when
one forms a visual image of an animal, that image does not
have a size. Other possibilities should also be taken into
account. In the first place, the comparison between pairs of
items could be less demanding then a verification task.
Table 5. Specific-General Attributes Questionnaire: proportion of MA’s
correct responses
General Specific
Perceptual Functional Perceptual Functional
Animals 74/88 (0.84) 70/94 (0.74) 68/123 (0.55) 95/128 (0.74)Inanimate 42/48 (0.87) 66/75 (0.88) 64/87 (0.73) 63/81 (0.78)
Living 99/117 (0.85) 120/151 (0.79) 112/181 (0.62) 134/181 (0.74)Non-living 17/19 (0.89) 16/18 (0.89) 20/29 (0.69) 24/28 (0.86)
Category-specific semantic disorders 197
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
Namely, asking whether the dolphin is grey may place a
greater demand than asking whether the dolphin has the
same colour as the shark’s. In the second case you may just
think whether the two animals are similar, whereas in the first
case you are required to access its right colour. Moreover,
when a subject is asked to make a size judgement task, some
encyclopaedic information could be retrieved: for example,
if you are asked to say whether the horse is bigger thanthe donkey or the spoon than the ladle, you may just rely
on learned notions, instead of thinking about their visual
properties.
Task 5 – questionnaire on visible and not visibleattributesThis task was created to investigate MA’s knowledge about
properties of pictured animals, both when the properties werevisible and when they were not visible in the picture. Twenty
animals were selected and one to three pictures of these
animals were developed. A different view of the animal
was represented in each picture, so that only experimentally
relevant visual attributes were shown. For example in one
picture just a horse muzzle appeared and in another one, just
the tail was shown. Some pictures were coloured; others were
in black and white. We developed two to six questions foreach picture; 45 questions could be answered based on the
information in the picture and 77 attribute questions were not
answerable based on the information in the picture. The
aim was to test whether MA was able to answer the same
question both when the information was available in the
picture and when it was not given in the picture. Different
pictures of the same animal were spaced out throughout the
submission such that subjects could not rely on informationpresent in the previous picture. After having administered the
task to control subjects, 3 questions had been removed.
Participants were shown a picture of the animal (e.g. an
elephant) and told. ‘‘This is a picture of an elephant; I will
ask you some questions about elephants and you can use the
information available in the picture to answer them’’. Ques-
tions were closed.
ResultsMA had no difficulties in answering when the attribute was
visible (39/42), but if the information required was not
available in the picture, his performance worsened signifi-
cantly: 55/73 (X2¼ 5.479, p< 0.019).
CommentsThese results confirm that the disorder is specific for notvisible attributes, as in the case of EW.
Discussion
In the present paper we report a further case of a patient
showing a category-specific naming impairment affecting
living things. We show that in this case the living/non-living
dissociation is a semantic deficit that is independent of the
modality in which information is provided and the patient’s
response is expressed. The main results of this case study can
be summarised as follows (see Table 6):
1. MA’s deficit is not due to defective visual abilities or
impaired visual processing of complex objects, as testified
by his performance on tests such as the VOSP and the
BORB, and is characterised by the absence of considerable
language impairment.
2. MA has a category-specific deficit concerning the category
of living entities, and not only the category of animals: the
patient is more impaired in naming animals, fruits, vege-tables, flowers and plants; the category of weapons is
damaged as well. Overall, non-living entities are more
preserved compared to living things (on average 67%
versus 41%). The category-specific nature of the deficit
remains significant when potential confounding variables
such as name frequency, concept familiarity, typicality and
age of acquisition are controlled for. In particular, the
living (not the animate) effect persists, even when all fourfactors are entered in a logistic regression analysis at the
same time.
3. We replicated carefully the procedures employed by
Caramazza and Shelton (1998) to assess the possibility
to generalise their findings; the tasks were thus closely
comparable, but different results emerged.
4. MA has a category-specific deficit for living things that,
only in the case of specific features, is associated with a lossof perceptual more than functional attributes. The informa-
tion about perceptual features available to MA is unlikely to
be sufficient to discriminate between real and unreal ani-
mals, but may be sufficient to distinguish a real artefact
from an unreal one. Moreover MA had no difficulties with
the general properties of the stimuli: he knew what, for
example, a vegetable is, but he got confused with detailed
distinctions among category members.
Some authors have suggested that a number of reported
cases of deficits for living things may be accounted for by
nuisance variables such as low familiarity or low frequency
(Funnel and Sheridan, 1992); however, this explanation can beruled out in the case of our patient’s performance. Neither
familiarity nor frequency has significant effects; only age of
acquisition influenced MA’s naming performance, although
the patient still showed a highly reliable category effect. The
results of this study reinforce the importance of age of
acquisition in semantic impairments (Morrison et al., 1992;
Lambon-Ralph et al., 1998; Moss and Tyler, 2000, Garrard
et al., 2001). Age of acquisition is a strong predictor becauseearly-acquired words are more likely to be named than late
acquired ones. However, it cannot be claimed that MA’s
deficit is due to an age-to-acquisition effect, since it cannot
account by itself for the category effect.
Once we have ruled out the possibility that name frequency,
concept familiarity, typicality and age of acquisition were
responsible for MA’s decrement in performance for living
items, we can move on to compare our results with Caramazza
198 C. Rosazza et al.
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
and Shelton’s (1998). EW, who had been tested using the same
procedure.
The patient reported by these authors showed a category-specific deficit only for animate entities: on naming, on object
decision task, on sound identification, and on the several
questionnaires that tested her semantic knowledge, EW was
never impaired with fruits, vegetables, buildings or vehicles,
but only with animals. On the basis of this case, Caramazza and
Shelton (1998) proposed that there are selective neural
mechanisms responsible for the recognition of three broad
semantic domains: animals, plant life and artefacts.This hypothesis assumes and predicts a strong correlation
between cerebral substrates that have been damaged and the
impaired domains of knowledge. Studies on brain-damaged
patients and on normal subjects have provided evidence for
distinct brain substrates of knowledge concerning different
categories (Gainotti, 2000). Many patients with category-
specific disorders affecting living items have bilateral damage
to the antero-medial and inferior parts of the temporal lobes;patients with deficit for non-living things often have left
fronto-parietal damage (Gainotti et al., 1995; Gainotti
2000). However, there are exceptions to this pattern. Tranel
et al. (1997) found that defective knowledge about animals
was associated with medial occipito-temporal lesions bilat-
erally, while deficit for tools was associated with the left
occipito-parietal-temporal junction. Finally, data from PET
and fMRI studies have provided other evidence about corre-lation between activated regions and categories. The most
consistent finding is that tools elicit greater activity in the left
posterior middle temporal gyrus across different tasks (Chao
et al., 1999; Perani et al., 1999; Martin and Chao, 2001). On
the other hand, animals activated different areas less consis-
tently, in particular the left fusiform gyrus and the more lateral
aspect of the fusiform gyrus. Moreover, neuroimaging studies
have shown that the response to an object category is not
limited to the region that responds maximally to that category,but it involves other regions that respond maximally to other
categories. What emerges is a continuous representation of
object features, instead of a collection of category-specific
modules (Ishai et al., 1999).
Another prediction made by Caramazza and Shelton (1998)
is that category specific deficits exist only for those categories
that have an evident evolutionary value (animals, plant life
and artefacts): consequently other finer-grained distinctionsare considered to be artefactual. This theory cannot account
for some impairments that do not respect this tripartition: L.A.
(Silveri and Gainotti, 1988) showed a deficit for animals,
fruits, vegetables and also food, while her performance was
better with tools and body parts. Patient JP studied by Siri et al.
(submitted) has a deficit restricted to fruits, vegetables, birds
and musical instruments, while his performance with animals
was spared. It is difficult to understand how the neural systemspecialised for animals is preserved in presence of an impair-
ment of the category of birds. In the present study as well, MA
is more impaired with living things and with weapons within
non-living categories on naming; unfortunately we did not
include this category in our questionnaires.
Finally, Caramazza and Shelton’s hypothesis (1998) pre-
dicts that perceptual attributes are damaged just as much as
functional ones. As already said, there are a number of casesof patients with a different pattern of impairment (Basso et al.,
1988; Sartori and Job, 1988; Silveri and Gainotti, 1988;
Sartori et al., 1993; De Renzi and Lucchelli, 1994) to which
we can add MA, who, at least for specific features, is more
impaired in perceptual than functional features. Caramazza
and Shelton (1998) have questioned the evidence supporting
Table 6. Comparison of the most important findings in MA and EW (Caramazza and Shelton, 1998)
Naming EW MA
Animals Plants Objects Animals Plants Objects
16/47 (34%) 24/24 (100%) 162/179 (90%) 21/43 (49%) 20/52 (38%) 115/171 (67%)
Object decision task 36/60 (60%) 55/60 (92%) 86/129 (67%) 28/31 (90%)
Central attributes questionnairePerceptual 64/98 (65%) ? 91/98 (93%) 52/80 (65%) 41/56 (73%) 61/73 (83%)Functional 37/57 (65%) ? 56/57 (98%) 62/82 (76%) 40/54 (44%) 75/88 (85%)
Specific-general questionnaireGeneral
Perceptual 130/130 (100%) 60/60 (100%) 74/88 (84%) 42/48 (87%)Functional 129/130 (99%) 75/75 (100%) 70/94 (74%) 66/75 (88%)
SpecificPerceptual 74/100 (74%) 100/100 (100%) 68/123 (55%) 64/87 (73%)Functional 115/150 (77%) 149/150 (99%) 95/128 (74%) 63/81 (78%)
Size judgement 13/18 (72%) ? 17/18 (94%) 21/21 (100%) 15/20 (75%) 20/20 (100%)
Other shared attributes 32/42 (76%) 42/42 (100%) 36/40 (90%) 37/40 (92%)
Category-specific semantic disorders 199
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
these cases on the basis of defective matching of the items in
terms of familiarity, frequency and visual complexity. It is
noteworthy that a similar criticism has been formulated by
Borgo and Shallice (2001) about a case study showing
defective visual knowledge in the absence of a category effect
(Lambon Ralph et al., 1998). This criticism should not apply
to the present case, insofar as the replication of the procedure
used by Caramazza and Shelton (1998) resulted in a highlyreliable living effect when familiarity, name frequency, typi-
cality and age of acquisition were jointly controlled for.
Caramazza and Shelton (1998) do not take into account the
frequent, although non-constant association between
impaired category and defective type of knowledge.
Contrary to other reports (for example, Sartori and Job,
1988; Silveri and Gainotti, 1988) the impairment for visual
properties in MA was not specific for living items, i.e. it failedto show an interaction with category. It is noteworthy that the
presence of interaction is not deemed to be necessary by some
computational models, such as Farah and McClelland (1991),
in which damage to visual properties results in a category-
specific disorder because of the differential weight of visual to
functional information between living and non living entities.
Our original idea was to replicate the testing procedures
developed by Caramazza and Shelton (1998) to assess whetherthe domain-specificity theory could be applied to our patient.
This was not the case. The presence of a more severe loss of
specific visual rather than functional knowledge could support
an interpretation according to the SFT. It must be however
underlined that there is evidence that a deficit in perceptual
knowledge is neither necessary (Caramazza and Shelton, 1998)
nor sufficient (Coltheart et al., 1998; Lambon Ralph et al.,
1998) to bring about a selective disorder for the living category.At the moment no theory can account for our case com-
pletely, and more importantly, no theory can account for all
the cases described in the literature.
What can be concluded from the insufficient explanatory
power of current theories about category-specific disorders? A
possible interpretation is that different mechanisms may be
responsible for category-specific deficits in individual
patients. Patients with category-specificity are very hetero-geneous, and each patient represents a ‘‘nature experiment’’
that permits verification of a specific hypothesis about normal
cognitive processing structures. As observed by Gainotti and
Silveri (1996), although patients with a category specific
deficit for living things have some characteristics in common
from the clinical, anatomical and pathophysiological point of
view, this does not necessarily mean that they constitute a
homogeneous entity.Heterogeneity is an interesting aspect to consider, because
it may reflect, in addition to different patterns of impairments
of distributed conceptual representations, different mecha-
nisms of disruption of information processing likely to be
related to lesion site.
It must be underlined that most of the theories of knowledge
organisation do not include any treatment of processing
aspects. As has already been mentioned, it is clear that
retrieving the name of a specific feature is different from
verifying a sentence or comparing two statements. Further,
the role of specific features may differ among tasks: for
example, there is evidence that visual features are central
in the definition of conceptual representations of living enti-
ties, while functional features may be more important for
name representations (Marques, 2002). In the absence of a
careful consideration of the requirement of any individualtask, the variability of patients’ performance may in the end
represent a confounding factor from the point of view of the
contribution of category-specific disorders to the understand-
ing of how semantic memory is organised. Finally, different
aetiologies and different lesion locations may affect not only
conceptual representations, but also the mechanisms of access
and retrieval of semantic knowledge (consider, for example,
the contrast between the involvement of the semantic orlexical level – see Gainotti 2000 for a review). What is clearly
lacking is a comprehensive theory which can take all these
different factors into account.
References
Basso A, Capitani E, Laiacona M. Progressive language impairment withoutdementia: A case with isolated category specific semantic deficit. Journal ofNeurology, Neurosurgery and Psychiatry 1988; 51: 1201–7.
Benton AL, Silvan AB, Hamsher K de S, Varney NR, Spreen O. Facialrecognition. Firenze Organizzazioni Speciali 1992.
Borgo F, Shallice T. When living things and other ‘‘sensory-quality’’categories behave in the same fashion: A novel category-specificity effect.Neurocase 2001; 7: 201–20.
Borrini G, Dall’Ora P, Della Sala S, Marinelli L, Spinnler H. AutobiographicalMemory. Sensitivity to age and education of a standardized enquiry.Psychological Medicine 1989; 19: 215–24.
Cappa SF, Frugoni M, Pasquali P, Perani D, Zorat F. Category-specific namingimpairment for artefacts: A new case. Neurocase 1998; 4: 391–7.
Caramazza A, Shelton JR. Domain-specific knowledge systems in the brain:The animate-inanimate distinction. Journal of Cognitive Neuroscience1998; 10: 1–34.
Chao LL, Haxby JV, Martin A. Attribute-based neural substrates in temporalcortex for preceiving and knowing about objects. Nature Neuroscience1999; 2: 913–9.
Coltheart M, Inglis L, Cupples L, Michie P, Bates A, Budd B. A semanticsubsystem of visual attributes. Neurocase 1998; 4: 353–70.
De Renzi E, Faglioni P. Normative data and screening power of a shortenedversion of the Token Test. Cortex 1978; 14: 41–9.
De Renzi E, Faglioni P, Ruggerini C. Prove di memoria verbale di impiegoclinico per la diagnosi di amnesia. Archivio di Psicologia, Neurologia ePsichiatria 1997; 38: 303–18.
De Renzi E, Lucchelli F. Are semantic systems separately represented in thebrain? The case of living category impairment. Cortex 1994; 30: 3–25.
Durrant-Peatfield MR, Tyler LK, Moss HE, Levy J. The distinctiveness ofform and function in category structure: A connectionist model. In:Proceedings of the Nineteenth Annual Cognitive Science Conference,University of Stanford. Mahwah, NJ: Erlbaum.
Farah MJ, McClelland JL. A computational model of semantic memoryimpairment: Modality specificity and emergent category specificity. Journalof Experimental Psychology: General 1991; 120: 339–57.
Folstein MF, Folstein SE, McHugh PR. ‘Mini-Mental state’ a practical methodfor grading the cognitive state of patients for the clinician. Journal ofPsychiatric Research 1975; 12: 189–98.
Funnell E, Sheridan J. Categories of knowledge? Unfamiliar aspects ofliving and nonliving things. Cognitive Neuropsychology 1992; 9:135–53.
Gainotti G. What the locus of brain lesion tells us about the nature of thecognitive defect underlying category-specific disorders: A review. Cortex2000; 36: 539–50.
200 C. Rosazza et al.
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
Gainotti G, Silveri MC. Cognitive and anatomical locus of lesion in a patientwith category-specific semantic impairment for living beings. CognitiveNeuropsychology 1996; 13: 357–89.
Gainotti G, Silveri MC, Daniele A, Giustolisi L. Neuroanatomical correlates ofcategory-specific semantic disorders: A critical survey. Memory 1995; 3:247–64.
Garrard P, Lambon Ralph MA, Hodges JR, Patterson K. Prototypicality,distinctiveness and intercorrelation: Analyses of the semantic attributes ofliving and nonliving concepts. Cognitive Neuropsychology 2001; 18: 125–74.
Grant DA, Berg EA. A behavioral analysis of damage of reinforcement andease of shifting to new responses in a Weigl-type card-sorting problem.Journal of Experimental Psychology 1948; 38: 404–11.
Humphreys GW, Forde EM. Hierarchies, similarity and interactivity in objectrecognition: ‘‘Category-specific’’ neuropsychological deficits. Behav BrainSci. 2001; 24(3): 453–76.
Ishai A, Ungerleider LG, Martin A, Schouten JL, Haxby JV. Distributedrepresentation of objects in the ventral visual pathway. Proceedings ofthe National Academic Sciences of United States of America 1999; 96:9379–84.
Laiacona M, Barbarotto R, Trivelli C, Capitani E. Dissociazioni semanticheintercategoriali: Descrizione di una batteria standardizzata e dati normativi.Archivio di Psicologia, Neurologia e Psichiatria 1993; 54: 209–48.
Laiacona M, Capitani E, Barbarotto R. Semantic category dissociations: Alongitudinal study of two cases. Cortex 1997; 33: 441–61.
Lambon Ralph MA, Howard D, Niglitingale G, Ellis AW. Are living and non-living category-specific deficits causally linked to impaired perceptual orassociative knowledge? Evidence from a category-specific double dissocia-tion. Neurocase 1998; 4: 311–38.
Lotto L, Dell’Acqua R, Job R. Le figure di PD/DPSS. Misure di accordo sulnome, tipicita, familiarita, eta di acquisizione e tempi di denominazione per266 figure. Giornale Italiano di Psicologia 2001; XXVIII: 193–207.
Marques FJ. Names, concepts, features and the living/non-living thingsdissociation. Cognition 2002; 85: 251–75.
Martin A, Chao LL. Semantic memory and the brain: Structure and processes.Current Opinion Neurobiology 2001; 11: 194–201.
McCarthy R, Warrington EK. Evidence for modality-specific meaning systemsin the brain. Nature 1988; 334: 428–30.
Morrison CM, Ellis AW, Quinlan PT. Age of acquisition, not frequency affectsobject naming, not object recognition. Memory and Cognition 1992; 20:705–14.
Moss HE, Tyler LK. A progressive category-specific semantic deficit for non-living things. Neuropsychologia 2000; 38: 60–82.
Osterrieth PA. Le test de copie d’une figure complexe. Archives dePsychologie 1994; 30: 206–356.
Perani D, Schnur T, Tettamanti M, Gorno-Tempini M-L, Matarrese M, Fazio F.Word and picture matching: A PET study of semantic category effects.Neuropsychologia 1999; 37: 293–306.
Raven JC. Coloured progressive matrices: Sets A, Ab, B. London: H.K. Lewis,1962.
Riddoch ML, Humphreys GW. Birmingham object recognition battery. Hove:Lawrence Erlbaurn Associates 1993.
Samson D, Pillon A, De Wilde V. Impaired knowledge of visual andnon-visual attributes in a patient with a semantic impairment for livingentities: A case of a true category-specific deficit. Neurocase 1998; 4:273–90.
Sartori G, Job R. The oyster with four legs: A neuropsychological study on theinteraction of visual and semantic information. Cognitive Neuropsychology1988; 5: 105–132.
Sartori G, Job R, Miozzo M, Zago S, Marchiori G. Category-specific form-knowledge deficit in a patient with herpes simplex virus encephalitis.Journal of Clinical and Experimental Neuropsychology 1993; 15: 280–99.
Silveri MC, Gainotti G. Interaction between vision and language in category-specific impairment. Cognitive Neuropsychology 1988; 5: 677–709.
Siri S, Kensinger EA, Cappa SF, Hood KL, Corkin S. Questioning the living/non living dichotomy: Evidence from a patient with a restricted semanticdeficit, submitted.
Snodgrass J, Vanderwart M. A standardized set of 260 pictures: Norms forname agreement, familiarity and visual complexity. Journal of ExperimentalPsychology: General 1980; 6: 174–215.
Tranel D, Damasio H, Damasio AR. A neural basis for the retrieval ofconceptual knowledge. Neuropsychologia 1997; 35: 1319–27.
Warrington EK. The selective impairment of semantic memory. QuarterlyJournal of Experimental Psychology 1975; 27: 635–57.
Warrington EK, James M. The visual object and space perception battery. BurySt. Edmunds: Thames Valley Test Company.
Warrington EK, McCarthy RA. Categories of knowledge: Further fractiona-tions and an attempted integration. Brain 1987; 110: 1273–96.
Warrington EK, Shallice T. Category specific semantic impairments. Brain1984; 107: 829–54.
Wilson B, Cockburn J, Baddeley AD. The Rivermead Behavioural MemoryTask. Bury St. Edmunds, Suffolk, UK: Thames Valley Test Company.
Received on 30 July, 2002; resubmitted on 26 November, 2002;accepted on 29 November, 2002
Category-specific semantic disorders 201
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14
The heterogeneity of category-specificsemantic disorders: evidencefrom a new case
C. Rosazza, E. Imbornone, M. Zorzi, E. Farina,L. Chiavari and S. F. Cappa
AbstractWe report a new case of category-specific semantic impairment, affecting livingentities, in a patient with traumatic brain damage. In the present investigationwe attempted to replicate as closely as possible the testing procedures whichhave been developed by Caramazza and Shelton (1998) to evaluate EW, apatient with a selective semantic disorder for the animal category. The results inour patient indicated a different performance profile, characterised by a moreextensive semantic disorder for living entities, and by a more severe loss ofspecific visual rather than functional knowledge. These findings concur withother evidence indicating that category-specific semantic disorders are hetero-geneous, reflecting different mechanisms of impairment, most likely associatedwith different neurobiological underpinnings.
JournalNeurocase 2003; 9: 189–202
Neurocase Reference Number:542/02
Primary diagnosis of interestCategory-specific semantic disorder
Author’s designation of caseMA
Key theoretical issue* Status of visual and functional knowledge in a patient with semantic disorderfor living entities
Key words: category-specific semantic impairments; reductionist and non-reductionist theories; heterogeneity; task repairments; different mechanismsof disruption of information processing
Scan, EEG and related measuresComputerised tomography, single photon emission tomography
Standardized assessmentMMSE, Raven PMC, Token test, digit span, Corsi span, Rey figure copy andrecall, Logical Memory, Rivermead Behavioural Memory Test, Naming Testand Questionnaire, Wisconsin Card Sorting, Weigl, Benton Face Recognition,BORB
Other assessmentPicture naming, object and part decision tasks, questionnaires semanticknowledge
Lesion location* Hypodense right frontal area, multiple small frontal bilateral hypodensities,
reduction of perfusion in the left mesial temporal region
Lesion typeClosed head injury
LanguageEnglish
202 C. Rosazza et al.
Dow
nloa
ded
by [
The
Uni
vers
ity o
f M
anch
este
r L
ibra
ry]
at 1
7:30
09
Dec
embe
r 20
14