Neurocase (2006) 12, 307–316
Copyright © Taylor & Francis Group, LLC
ISSN: 1355-4795 print / 1465-3656 online
DOI: 10.1080/13554790601125940

A Case of Impaired Naming and Knowledge of Body Parts.
Are Limbs a Separate Sub-category?
NNCS

MARCELLA LAIACONA1, NADIA ALLAMANO1, LORENA LORENZI1 and ERMINIO CAPITANI2
Impaired Naming and Knowledge of Body Parts

1

Division of Neurology, Aphasiology Unit, “S. Maugeri” Foundation, IRCCS, Rehabilitation Institute of Veruno, Italy
Milan University, Clinic for Nervous Diseases, S. Paolo Hospital, Milan, Italy

2

A left hemisphere stroke patient presented a disproportionate difficulty for body parts knowledge without autotopagnosia. The deficit
concerned the lexical-semantic representation of body parts and was most severe for limbs. The ability to gesture was spared and action
naming was not more impaired than object naming.
On the basis of normal naming latencies, we conclude that limbs are the most vulnerable component of the overall category of body
parts. This vulnerability is not explained by unbalanced nuisance variables. More cognitive effort is probably required for the appropriate
differentiation of limbs during semantic processing and lexical access.

Introduction
Some neuropsychological abilities related to body parts can
be impaired with disproportionate severity after brain lesions.
Broadly speaking, these tasks can be grouped into two
classes: (a) tasks where the patient is asked to point to body
parts in the frame of the whole body (on his/her own body, on
a third person’s body, or on a picture), and (b) tasks where
the patient is asked to name body parts, to describe their function or to identify them on verbal request. Early clinical
descriptions introduced the term autotopagnosia with reference to patients who failed in the first class of tasks but were
proficient when pointing was not requested. Since then, the
nature of autotopagnosia has been widely discussed. De
Renzi and Faglioni (1963) and De Renzi and Scotti (1970)
hypothesized that autotopagnosia was part of a more general
impairment in any task that requires the patient to divide a
whole into its parts, but this account was not confirmed by
successive studies. At variance, Semenza and Goodglass
(1985) considered that the success in identifying body parts
under all conditions depended on a common basic factor,
identified as the strength and precision of the conceptual representation of individual body parts, whether indicated by
name, visual pointing, or touch. This notion was based on the
observation that, also in non-verbal tasks, the word frequency
Received 30 March 2006; accepted 16 November 2006.
This study was partly supported by a MIUR grant to E.C. and by
a Milan University FIRST grant. Fabio Comazzi assisted in setting
the naming latencies records. Rosemary Allpress revised the
English text.
Address correspondence to Marcella Laiacona, S. Maugeri Foundation, Scientific Intitute of Veruno, Division of Neurology,
Aphasiology Unit, Veruno Medical Centre, via per Revislate, 28010
Veruno (Novara), Italy. E-mail: mlaiacona@fsm.it

of individual words strongly correlated with success in pointing to body parts; moreover, also in non-verbal tasks patients
produced a preponderance of conceptually-related errors as
functional and vicinity errors. The latter authors defined as
“functional errors” eye-ear substitution, toe-thumb substitution, and joint-joint substitutions, and defined as “vicinity
errors” failures within the same limb of the target or interchanged face-head constituents.
More recent studies have hypothesized that distinct cognitive representations of the body are called into play in different tasks. Sirigu et al. (1991) proposed that autotopagnosia
affects visuospatial representations of the body schema in
presence of an intact semantic and lexical representation of
body knowledge. The patient they examined could not localise body parts on herself or on another person when
requested verbally or otherwise, and was unable to describe
their location verbally; however, she could name the same
body parts and define their function. Consistent with this evidence, Denes et al. (2000) suggested that in autotopagnosia
the spatial representation of the body (“body schema” or
“conscious awareness of the body”) is lost, but the meaning
of words related to the body is spared. The latter authors supported the view of Reed and Farah (1995) that a supramodal
cognitive system is necessary for coding spatial relations
among body parts, while other systems cater for the spatial
encoding of extrapersonal items. More recently, Buxbaum
and Coslett (2001) have interpreted autotopoagnosia as an
impairment in processing information about the shape and
contours of the surface of the human body and, hence, an
impairment in coding the local relationship between body
parts (denoted with the term “body structural description”).
Interestingly, also the opposite phenomenon—preserved
ability to localize body parts but defective recognition or
naming of them—has been reported, although less frequently.
A delimited body parts anomia was described by Dennis (1976)

308
in a patient who correctly identified body parts on herself or
on a person in front of her and who was highly successful on
non-verbal topographical judgements, but was impaired in
the production and, to a lesser extent, comprehension of body
part names. Errors involved the substitution of one body part
for another. Dennis interpreted the disorder as a deficit of lexical selection within the semantic category of body parts
when there were insufficient distinctive features. More
recently, Suzuki et al. (1997) reported a case of comprehension deficit restricted to body parts. The patient’s difficulty
was not consistent with autotopagnosia or with a general
part-whole deficit: It was the name itself of the body part that
the patient found difficult to comprehend. These authors
claimed that lexical knowledge of body parts has a categoryspecific organisation, and related its anatomical substrate to
the left parietal lobe. This type of representation that includes
lexical-semantic, prepositional information about the human
body has been denoted as “body image” or “body semantics”
(Buxbaum et al. 2001).
There are also descriptions of cases of selective sparing of
the body parts category. The patient reported by Shelton et al.
(1998) presented such sparing on several tests of verbal comprehension and production, and the authors suggested that body
parts are a category with a distinct semantic representation that
can fractionate as the result of brain damage. Coslett et al.
(2002) reported another patient, affected by semantic dementia,
showing a selective sparing of human body knowledge.
The comprehensive definition of what representations are
necessary for the different cognitive activities centred on
body parts, already attempted by Sirigu and colleagues, was
recently reappraised by Schwoebel et al. (2004) and
Schwoebel and Coslett (2005), who stressed again that there
are at least 3 distinct types of representation of the human
body. One representation is a topological map that codes
body part location and defines body part boundaries and
proximity relationships (“body structural description”),

M. Laiacona et al.
another is a lexical-semantic representation that includes
body part names and codes knowledge of body parts as their
functions and their relations with artefacts (“body semantics” or “body image”), and the third provides a dynamic
mapping of the current position of body parts relative to one
another (“body schema”). This model provides an extensive
frame for explaining the separate occurrence of autotopagnosia (as a disorder of the body structural description), and
of the isolated sparing or impairment of body parts knowledge (as a disorder of the body semantics). The hypothesis
of a dynamic mapping of body position during action, moreover, suggests a theoretical link between body parts deficits
and apraxia.
In this study we describe a patient affected by a disproportionate naming difficulty for body parts on the basis of various tasks designed to check category-specificity of the
impairment of naming, comprehension, and semantic knowledge. We analyzed if any error type was prevalent, and if any
particular subset of body parts was more affected than others.
We also checked if the motor skills related to the use of tools
were affected, and if there was a relationship between body
parts knowledge and action knowledge.

Case report
FC, a 42-year-old right-handed male with a high school education, suffered in January 1996 an extensive left hemisphere
infarction as a consequence of left carotid artery dissection. A
CT scan performed in March 1996 disclosed a large softening
covering a large cortical area and deep structures supplied by
the left middle cerebral artery (Figure 1). FC presented right
hemiplegia and right homonymous hemianopia. Early after
stroke, language examination showed severe aphasia, verbal
output was limited to single syllables, and auditory verbal
comprehension was limited to single words.

Fig. 1. CT-scan showing a large cortical and sub-cortical hypodense area in the cerebral regions supplied by the left middle cerebral artery.

309

Impaired Naming and Knowledge of Body Parts
At the beginning of this study (Autumn 1999–Spring
2000), FC underwent a language examination.

(i) Aachener Aphasie Test (AAT)
On AAT (Luzzatti et al., 1994) FC’s deficit could not be
univocally classified, and was compatible with mild Broca’s
aphasia (63% probability) and with mild amnestic aphasia
(27% probability). Spontaneous speech was reduced and
slightly dysarthric, without overt articulatory impairment.
Verbal communication was possible with minimal assistance from the examiner. The syntactic structure was poor
as sentences were formed by only two or three words with
few function words. Some verbal substitutions and word
finding difficulties were present, but no phonemic substitutions were produced. On conversation he produced the following utterances: “.. I was drinking a coffee with …, I felt
unwell, .. the doctor .. one hour after, no a day after .. I went
to … Verbania and I do not recover … .. (consciousness)”.
Written language (reading aloud and spelling) was quite
good, showing only mild syllabisations (“staccato”) in reading and few letter omissions or substitutions on writing to
dictation test.

(ii) BADA Aphasia examination
On the more extensive Batteria per l’Analisi dei Deficit
Afasici (BADA) examination (Miceli et al., 1994) the output
speech was reduced, with simple sentences, word finding
difficulties, and few semantic substitutions. Telling the story
of Little Red Riding-hood he said: “ ….. the bad wolf go to
the (dal-dalla) grandmother ….. the grandmother eat the ….,
the wolf eat the grandmother and then they …. disguise
themselves ……. as grandmother …. Little Red Riding-hood
eat …..no! the other, the beast, the wolf, wolf eat Little Red
Riding-hood …..”. On phonological tasks, FC made few
errors (9/232: 3.9%) at sublexical level. Similarly, on lexical
decision, word transcoding, and word-picture matching few
errors were present (6/160: 3.8%, 6/193: 3.1%, and 2/120:
1.7%, respectively). At the grammatical level FC made few
errors (13/203: 6.4%) that in comprehension were confined
to reversible sentences on auditory presentation.
On the BADA, most errors affected naming tasks (overall
35/118: 30%) where semantic substitutions and word finding
difficulties were evident.

et al., 2006)1 are available. These 60 stimuli include 10 pictures for each of the following categories: animals, fruit, vegetables, tools, vehicles, and furniture. FC was able to give the
correct name to 46/60 stimuli (77%). Lexical frequency and
the precocity of word acquisition were significant predictors
of naming success. FC was impaired in naming stimuli from
different categories, including both biological and man-made
objects.

The disproportionate impairment of body parts
FC underwent long-term, uninterrupted language rehabilitation. During rehabilitation, the naming problems of the
patient, often consisting of semantic errors, were more
marked with low frequency words but were noticeable even
for body part names notwithstanding their high frequency.
With the latter category, FC often made semantic errors both
during rehabilitation and in everyday conversation. For
instance, instead of arm he would say “elbow”, instead of
elbow “knee”, and instead of foot “hand”. Sometimes the
patient made semantic errors also with verbs that denote body
actions: Asked to name the picture of an onion, FC could not
retrieve the correct name and pantomimed the action of crying, but after a short delay he said “… they make me cough”.
To confirm our clinical impression of a relatively more
severe impairment with body parts we submitted FC to a supplementary naming examination using the 10 Snodgrass and
Vanderwart stimuli representing body parts. In order to contrast body parts with really comparable stimuli of different
categories, we selected 20 stimuli from the subset of Laiacona
et al. (1993)2 matched to the 10 body parts items for frequency, precocity of acquisition (Barbarotto et al., 2005), and
item difficulty, as these proved to be significant predictors for
naming success. In addition to picture naming, FC was given
a pointing-to-pictures task and verbal semantic questionnaire
based on the 10 body parts from the Snodgrass and Vanderwart set and 20 other stimuli from the subset of Laiacona et al.
Table 1 shows that body parts stimuli still had a higher lexical
frequency but presented a slightly later age of acquisition.
In each task, the difficulty estimated from a control group
was almost identical for body parts stimuli and for the other
20 stimuli.
On picture naming, the patient was asked to name the 30
pictures, presented individually in random order. We considered
1

(iii) Naming pictures of different categories
In order to better study the influence of lexical and semantic
variables on naming, FC was given a battery based on strictly
controlled material (Laiacona et al., 1993). This includes 60
pictures from the Snodgrass and Vanderwart (1980) set for
which the Italian lexical frequency (Bortolini et al., 1972)
and an objective index of precocity of acquisition (Barbarotto

The precocity of acquisition index considered here for each
stimulus was gender specific and was the percentage of 97 boys
aged 3–6 years that correctly named a coloured picture representing
the stimulus. Therefore, unlike a true age of acquisition measure,
high values of our index correspond to words that are acquired
early.
2
The body parts stimuli were lips, ear, nose, finger, hair, leg, arm,
foot, eye, and hand. The other stimuli were bed, train, truck, table,
grapes, apple, vase, bicycle, motorbike, chair, sofa, airplane, car,
lemon, tomato, lamp, nail, butterfly, rooster, and hammer.

310

M. Laiacona et al.

Table 1. Variables potentially influential on the success rate for the different groups of stimuli

Lexical Frequency
Precocity of acquisition index (0 to 100, low value indicates late acquisition)
Difficulty index for naming (0 to 60, with top difficulty=0)
Difficulty index for the questions (0% to 100%, with top difficulty=0%)
Judgement about the difficulty of each question (0 to 5, with top difficulty=0)

as a correct response: the stimulus name, a synonymous
name, or a non-synonymous name accepted as correct by at
least 3 of 5 judges. We distinguished correct responses given
immediately from correct responses given after a delay or self
correction.
On the word-picture forced-choice matching task, the
patient was asked to point to the picture representing the
stimulus name given by the examiner. The target picture and
four foils from the same semantic category were arranged
vertically on a card.
On the standardized verbal semantic questionnaire we
investigated the semantic knowledge of the same 30 items
used for the study of naming and comprehension. For each
stimulus, superordinate category and subordinate features
were probed with six questions. For example, for the stimulus
EAR, 1) a general superordinate information question was: Is
it an object, a vegetable, or a body part? 2) superordinate
information within the same category: Is it part of the bowels,
a sense organ, or a limb? 3) subordinate information about
perceptual attributes: Do we have one ear, two ears, or more?
4) subordinate information about a comparison of perceptual
attributes: Is it smaller than a hand? 5) subordinate information about associative functional aspects: Is it used for looking, for tasting, or for hearing? 6) subordinate information
about associative contextual aspects: Is it on the upper part of
the trunk, on the lower part, or on the head?
The probes were presented orally in standardised random
order. One point was given to every correct response. The 16
errors observed with body parts concerned 9 of the 10 stimuli: leg (4), foot (3), finger (3), and one error each for ear,
hand, nose, eye, arm, and lip. The more general superordinate question was always correct; the errors concerned 5
superordinate questions within the same category, 6 perceptual questions, and 5 associative questions. The 13 errors
observed with the remaining stimuli concerned 1 general
superordinate question, 7 perceptual questions and 5 associative questions.

Body Parts (n = 10)

Other stimuli (n = 20)

Mean=126.0 (SD=119.4)
Mean=68.90 (SD=27.67)
Mean=56.10 (SD=8.25)
Mean=98.03 (SD=3.74)
Mean=4.83 (SD=0.20)

Mean=33.8 (SD=38.1)
Mean=84.50 (SD=16.14)
Mean=58.20 (SD=3.20)
Mean=98.76 (SD=3.20)
Mean=4.78 (SD=0.27)

Table 2 reports the percentage of correct responses, separately for the two groups of stimuli.
On statistical analysis of immediate naming success, FC
named body parts (40%) significantly worse than the other
stimuli (80%), (chi-square = 4.800, df = 1, exact two-sided
p = .045). Repeating the analysis with a logistic regression
model and introducing the concomitant variables, the disadvantage of body parts was confirmed (chi-square = 6.246,
df = 1, p = .012). A similar trend was found when responses
given after long latencies or self-correction were included,
yielding a chi-square of 3.400, df = 1, exact p = .065 (twosided) and .035 (one-sided).
The analysis of the other tests had a similar outcome. On
pointing to pictures on verbal command, FC showed errors
with body parts only, and again the difference was significant
(chi-square = 6.667,df = 1, exact two-sided p = .029). A
significant impairment of body parts was observed also
with the questionnaire (chi-square = 7.420, df = 1, two-sided
exact p = .009).
In summary, the deficit that FC showed with body parts
was of moderate severity but all tests confirmed its relative
impairment with respect to other stimuli.
A preliminary question at this point was whether FC’s relative impairment with body parts was peculiar to our patient
or whether it might be common to other aphasic patients who
were given the same material. To this end, we scrutinized our
records of aphasic patients presenting a naming deficit of
comparable severity. As FC was 46/60 correct on naming
Snodgrass and Vanderwart stimuli, we selected all the male
patients on record who were affected by vascular aphasia and
presented a comparable immediate naming performance with
the same 60 stimuli (46 ± 2 correct responses, i.e. from 44 to
48/60 correct). The suitable scores were 17. As these control
examinations were strictly severity-matched for stimuli other
than body parts, we contrasted the body parts naming performance of FC (7/10 correct even allowing long latencies and
self corrections) with the body parts scores of the control

Table 2. Percentage of correct responses given by FC on the two groups of stimuli for the different tasks

Picture naming (short latency without self-corrections)
Picture naming (allowing long latency and self-corrections)
Pointing to a picture on verbal command
Verbal Semantic Questionnaire

Body Parts (n = 10)

Other stimuli (n = 20)

(4/10) 40%
(7/10) 70%
(7/10) 70%
(44/60) 73%

(16/20) 80%
(19/20) 95%
(20/20) 100%
(107/120) 89%

311

Impaired Naming and Knowledge of Body Parts
patients. For the control group, we found 8 flawless performances, 8 performances with 1 error, and 1 performance with
2 errors, with a mean of 0.59 body parts errors for each examination (against 3 errors committed by FC). This finding
clearly shows that, even admitting latencies and self-corrections,
the body parts naming of FC was peculiar with respect to the
control aphasic patients with the same general severity.
Given the contrast between body parts naming and other
stimuli naming, and taking into account the unusual quality
of FC’s errors, we planned further tasks to examine this category. We now considered an extended set of 20 body parts
items: lip, eyebrow, nose, teeth, neck, breast, back, belly, buttocks, eye, ear, fingers, arm, elbow, wrist, hand, leg, knee,
ankle, and foot. The inquiry may be viewed under 2 complementary perspectives. The first was whether the body parts
deficit is confined to verbal tasks or whether it includes non
verbal tasks and is associated to autotopagnosia. The second
question concerned the type of cognitive representation of
body parts, in particular if body parts form a homogeneous
set of items or need to be further distinguished.

Was the body parts deficit confined to verbal tasks?
(a) Verbal tasks
FC was given three naming tasks based on 20 body parts
stimuli:

Then FC was given 4 comprehension tests on verbal command with the same 20 body parts. We asked him to point to
body parts on verbal command. In addition, he was asked to
judge the correctness of pointing to body parts performed by
an actor on a third person after verbal command given by the
examiner. For each stimulus the actor showed 1 time the
correct body part, and another time a wrong body part, in
randomised sequence. The tasks were:
- Pointing to body parts on patient’s own body
- Pointing to body parts of a third person
- Pointing to pictures representing isolated body parts among
four alternatives
- Judgements about the correctness of pointing to body parts
after verbal command performed by a third person
Table 4 shows the percentages of correct responses given by
FC. Five matched normal controls were flawless.
Overall, FC correctly pointed to the requested body part or
correctly judged the pointing on 75/100 stimuli (75% correct). Regarding the judgment task, FC made 10 errors: On 2
occasions he judged that a right performance was wrong, and
on 8 occasions he judged that a wrong response was right. In
these comprehension tests no significant difference emerged
whether the target was himself, another person, an isolated
picture, or when the subject did not have to point to anything
as in the last task (chi-square = 2.133, df = 3, ns).

- Naming of body parts singled out by the examiner on the
patient’s body
- Naming of body parts singled out by the examiner on a third
person’s body
- Naming of body parts represented as isolated pictures

(b) Non verbal tasks

Table 3 shows the percentages of correct responses, allowing
long latency and self correction. Five matched normal controls were flawless.
Overall, in these tasks FC correctly named 35/60 (58%) body
parts stimuli. Here FC was slightly more impaired when the
body part was indicated on a third person (chi-square = 4.149,
df = 1, p = .042), suggesting that he was in some way facilitated by the perception of the part in relation to his own body
or by isolation of the stimulus pictorially.

- The examiner touched a part on the patient’s body while the
patient kept his eyes closed, and the patient was asked to
point to the corresponding picture (among 4 alternatives).
- The subject had to point on his own body to the single part
shown by the examiner in an isolated drawing.
- The subject pointed on a whole body drawing to the part which
the examiner had presented him as a single part drawing.

Table 3. Percentages of correct responses on body parts naming
task
Task

Success rate

Naming of body parts indicated by the examiner
on the patient’s body
Naming of body parts indicated by the examiner
on a third person’s body
Naming of body parts represented as isolated
pictures

(12/20) 60%
(8/20) 40%
(15/20) 75%

FC was asked to identify body parts by pointing to each of
the same 20 stimuli after non-verbal commands consisting of
pictorial or tactile stimuli. Three tasks were devised:

Table 4. Percentages of correct responses on pointing to body parts
after a verbal command
Task

Success rate

Pointing to own body part on verbal command
Pointing to body part of a third person on
verbal command
Pointing to pictures representing isolated body
parts (among 4 alternatives) on verbal
command
Judgements about the correctness of pointing
to body parts performed by a third person
after verbal command given by the
examiner

(13/20) 65%
(15/20) 75%
(17/20) 85%

(30/40) 75%

312

M. Laiacona et al.

Table 5. Percentages of correct responses on pointing to body parts
after a non-verbal command
Task

Success rate

The examiner touches a patient’s body part,
with the patient’s eyes closed, and the
patient then points to the corresponding
picture (among 4 alternatives)
The patient points on his own body to
single parts shown by the examiner as
isolated drawings
The patient points on the drawing of a
whole body to the part that the examiner
has shown as a single part drawing

(18/20) 90%

(20/20) 100%

(20/20) 100%

Table 5 shows the percentages of correct responses of FC.
Five matched normal controls were flawless.
Overall FC showed a very good performance (58/
60 = 96.7% correct) on these non verbal task: this finding is
consistent with the verbal nature of the deficit.
(c) Conclusions
At this point we can contrast verbal and non-verbal tasks performed with the same stimuli. Table 6 summarises the results
according to the mode of assessment used for identifying
body parts.
Table 6 shows that verbal tasks were performed worse than
non-verbal tasks (68.8% and 96.7% correct, respectively;
chi-square = 18.841, df = 1, p < .001). Also, a comparison
restricted to pointing tasks (verbal versus non-verbal) confirmed the same pattern: the verbal condition was significantly worse (chi-square = 12.550, df = 1, p < .001). Hence,
it was the verbal nature of the task that determined the deficit:
FC’s deficit appeared when asked to retrieve the name of a
body part or access its meaning starting from a verbal input.
The results of these tests permitted us to decide whether
our patient presented autotopagnosia. FC was given verbal
and non-verbal tasks on which he had to point to his own
body parts. In the former type of task, he had to point on his
own body to the part named by the examiner; in the latter
task, he had to point on his own body to the single part shown
by the examiner as an isolated drawing. As such, the output

of these tasks was identical, i.e. pointing to a patient’s body
part; what distinguished the tasks was the input. In the case of
autotopagnosia we would expect the patient to be impaired
with both types of input, whereas in the case of a purely verbal deficit we would expect only verbal commands to impair
performance. The performance of FC was consistent with the
latter hypothesis. On verbal input the patient gave 13/20 correct responses (65%, see Table 4), whereas on pictorial input
he was 20/20 correct (100%, see Table 5). This was a significant difference (chi-square = 8.760, df = 1, p = .003). This
excludes the presence of autotopagnosia and shows that the
deficit regarding body parts was linked to the verbal content
of the task.

Contrasting different body parts
In the set of tests reported above, FC was by inspection more
impaired with limbs than with other body parts. The 20 stimuli included 9 limbs (arm, fingers, elbow, wrist, hand, leg,
knee, ankle, and foot) and 11 non-limbs (lip, eyebrow, nose,
teeth, neck, breast, back, belly, buttocks, eye, and ear).
Considering the above 11 tasks (3 verbal pointing and a
double judgement on verbal pointing performed by an actor
on a third person, 3 non-verbal pointing, and 3 naming tasks)
limb stimuli were 60/99 (60.6%) correct, against 108/121
non-limb stimuli (89.3%). This difference was highly significant (chi-square = 24.760, df = 1, p < .0001). Splitting the
tasks according to the verbal/non-verbal nature, on verbal
tasks limbs were 34/72 correct (47.2%) and non-limbs 75/88
correct (85.2%). At the comparison, chi-square was 26.340,
(df = 1, p < .0001). On non-verbal tasks, limbs were 26/27
correct (96.3%) and non-limbs 32/33 correct (97%). The latter comparison yielded a chi-square < 1, ns. We concluded
that the deficit of our patient mainly concerned verbal tasks
of limb stimuli.
Why were limbs more impaired than non-limbs on verbal
tasks? Limbs might be semantically more related to one
another, and in this case we should expect two empirically verifiable facts: (i) limbs should be more frequently interchanged
with respect to non-limbs, and (ii) limb naming with the same
material should prove more difficult also for normal subjects.

(a) Error types
Table 6. Verbal and non-verbal tasks in the assessment of body
parts identification
Task
Naming of body parts
Pointing to body parts on verbal
command
Pointing to own body or to a drawing
after pictorial or tactile stimulus

Type of
demand
Verbal
Verbal

Success
rate
35/60 (58.3%)
75/100 (75%)

Non-verbal 58/60 (96.7%)

Following Semenza and Goodglass (1985), we grouped verbal task errors into 4 categories, separately for limbs and nonlimbs. We excluded the judgment task, (where the patient
made 8 errors with limbs and 2 errors with non-limbs stimuli)
as only yes-no responses were allowed and in the test construction the wrong alternatives (foils) did not systematically
include all the possible error types. Consequently, the error
counts reported in Table 7 refer to 6 tasks: 3 pointing on verbal command and 3 naming tasks. Errors were classified as
follows:

313

Impaired Naming and Knowledge of Body Parts
Table 7. Number of wrong responses classified by error type
Functional Vicinity Circum- Random
No
errors
errors locutions errors
response
Limbs
Non-limbs

15
0

8
8

0
0

6
2

1
1

- Functional errors: joint substitution, limb substitution, eye-ear
substitution
- Vicinity: a different part of the same limb, interchanged
face-head constituents
- Random: neither of the above
- No Response
Obviously, functional errors are expected more frequently
with limbs, but within limbs it is still interesting to see
whether the patient showed more functional or vicinity
errors, i.e. conceptually-related errors than random errors or
no responses. Table 7 shows the overall distribution of error
types on the verbal tasks.
Conceptually related errors, i.e. functional and vicinity
errors (31/41=75.6%), clearly represented the majority of
wrong responses. In particular, with limb stimuli the patient
showed an impressive number of functional errors.
(b) Naming latencies
To decide if limbs stimuli were basically more difficult also
for normal subjects, we examined a group of normal subjects
on a naming task based on the same stimuli and contrasted
the naming latencies of limbs and non-limbs stimuli.
The normal control sample included 16 adult males
matched to FC for education and age (42 ± 5 years). Controls
were asked to name the above 20 body parts pictures (9 limbs
and 11 non-limbs). Full-screen pictures were presented one at
a time on a 19” colour TV screen, and the time of naming
onset was recorded by a microphone connected to a PC and
automatically evaluated. After the naming of each stimulus
the picture disappeared, and the following stimulus appeared
after 2000 msec. The median naming latencies for each stimulus, together with the lexical frequency of each stimulus, are
displayed in Table 8.
The mean of the median values was 1305.4 msec for nonlimbs (sd = 531.8) and 1282.7 msec for limbs stimuli
(sd = 302.2). The difference was not significant (t < 1, df =
18, ns). However, this could be due to the fact that different
Italian terms exist for some body parts (e.g. “back” can be
termed “dorso” or “schiena”). There were 5 such stimuli in
our set (belly, buttock, back, breast and lips) and all of them
represented non-limbs. Restricting the analysis to the remaining 15 stimuli, limbs showed a trend toward longer latencies
(1287.7 vs. 1037.83) that yielded a t-point of 1.837 (df = 13,
p = .089). To control for the effect of lexical frequency, we
introduced the two lexical frequency values as covariates in

the comparison between limbs and non-limbs. The mean frequency values did not significantly differ between the two
subgroups of body parts with a single name, but frequency
was always higher for the limbs stimuli (95.2 vs. 77.7 for
written frequency, and 613.9 vs. 363.3 for spoken frequency). After introducing the frequency measures as covariates, the contrast between the naming latency of limbs and
non-limbs yielded an F value of 4.620 for the analysis
adjusted for written frequency (df 1, 12, p = .053) and of
5.027 for the analysis adjusted for spoken frequency (df =
1,12, p = .045)3.
It is not easy to explain why limbs stimuli, that present a
higher lexical frequency, showed longer naming latencies. A
possible reason could be “operativity”. According to Gardner
(1973), objects which can be readily operated upon (“operative elements”) can be named more easily than other objects
(“figurative elements”) which are continuous with the surrounding context and thus known in a figurative or limited way.
The former objects, defined as operative, are known through
a variety of actions and sensory modalities, are relatively
discrete and separate entities, have definite contours, and are
easy to grasp, manipulate and operate upon. Body parts can be
classified following this criterion, and we checked whether a
possibly higher incidence of operativity among non-limbs
items could explain their better performance.
The 20 stimuli employed in this study were classified as 11
operative (arm, fingers, hand, leg, foot, lip, nose, teeth, eye,
ear, breast) and 9 figurative elements (elbow, wrist, knee,
ankle, eyebrow, neck, back, belly, buttocks) based on the rating of 3 independent judges. Considering the 3 naming tasks
above, operative stimuli were 22/33 (67%) correct, against
13/27 figurative stimuli (48%). Hit scores were not significantly different: chi-square = 2.095, df = 1, p = .191, ns.
With respect to the limbs/non-limbs contrast, 5/9 limbs
(55%) and 6/11 non-limbs (54.4%) were endowed with operativity. As these figures are substantially similar, the principle of operativity cannot explain why limbs were more
difficult for our patient.
In order to further check the relevance of operativity, we
reanalysed FC’s naming performance after introducing the
operativity classification in the statistical model. Considering
the three naming tasks referred to above, we carried out a
logistic regression analysis where the number of correct
responses (from 0 to 3) was the dependent variable, and the
model variables were the classification of each stimulus as (i)
limb/non-limb and (ii) operative/figurative, thus allowing to
evaluate each factor considering the covariance with the
other. This analysis confirmed the significance of the limb/
non limb factor (chi-square = 13.757, p < .001) and showed
only a trend toward a better performance of operative stimuli
3

Interestingly, a trend toward higher latencies in the naming of
limbs with respect to non-limbs was observed in normal subjects by
Barbarotto, Laiacona and Capitani (unpublished) using different
pictures.

314

M. Laiacona et al.

Table 8. Median naming latencies (msec) of the control group (upper cell). In the lower cell we report two measures of the lexical
frequency of the stimulus in the Italian language: The written lexical frequency according to Bortolini, Tagliavini, and Zampolli (1972)
from a corpus of 500.000 words, and, separated by a slash, the spoken lexical frequency from a different corpus of 500,000 words (Di
Mauro, Mancini, Vedovelli, and Voghera, 1993)
Non-limbs
Latency
Frequency
Limbs
Latency
Frequency

Belly
2655
11/29
Knee
1164
44/91

Lips
904
34/374

Buttock
1221
0/0

Finger
1091
45/217

Ear
933
40/196

Eyebrow
1257
4/0

Elbow
1423
0/41

Hand
819
375/3250

(chi-square = 2.832, p = .093), without interaction between
these factors. This outcome indicates that operativity is not
relevant for explaining our findings.

(c) Conclusions
The error analysis confirmed that limbs were semantically
more inter-related than non-limbs. The former class of stimuli, in fact, showed an impressively large number of functional errors. The fact that limbs were more difficult to name
also for normal controls is in line with their closer relatedness. This relative difficulty of limbs stimuli cannot be traced
back either to an effect of lexical frequency (that was higher
for limbs) or to an effect of operativity of the stimuli, which
was balanced between limbs and non-limbs and was not a
significant predictor of naming success.

Nose
954
41/202

Eye
906
291/1491

Ankle
1788
7/0

Teeth
1012
45/194

Wrist
1490
16/89

Neck
1165
45/97

Arm
1542
112/573

Back
1806
24/67
Leg
1219
84/278

Breast
1547
14/230
Foot
1008
174/986

et al., 1994) originally included 52 pictured objects and 50
pictures representing specific motor actions. Action naming
was examined by asking the patient to utter the corresponding verb. From the objects set we excluded 23 stimuli that
represented body parts, clothes and manipulable objects, and
from the actions set we excluded 24 stimuli that were not
clearly agentive verbs, i.e. verbs whose meaning implies the
presence of an agent subject. The stimuli were presented one
at a time.
FC named correctly 26/29 (89.7%) objects and 20/26
(77%) actions, presenting word finding difficulties and
semantic substitutions with both classes of stimuli. No significant difference between the two set of stimuli was detected
(Chi-square = 1.624, df = 1, ns). The same outcome was
observed repeating the analysis by means of a logistic regression where lexical frequency was introduced as a covariate.
These results do not corroborate the hypothesis that limb processing impairment implies a greater semantic deficit of
actions with respect to non-manipulable objects.

Other relevant aspects: apraxia and action naming
Was FC’s semantic knowledge of limbs linked to his ability
to use limbs for performing gestures? Was it and the ability to
perform gestures linked to his semantic knowledge of
actions? If any relationship existed, a blurred limbs representation at semantic level could imply either object use apraxia
or, alternatively, an action naming impairment (reflecting a
defective semantic knowledge of actions).
Regarding apraxia, our patient was not impaired in performing gestures. FC was 14/14 correct on the object use test
by De Renzi et al. (1968). He also scored within the norm
(64/72 correct) on the ideomotor apraxia test by De Renzi
et al. (1980) that calls for imitation of 12 meaningful and 12
meaningless gestures not related to the use of objects: a maximum of 3 points is assigned for each gesture. Our patient
was 32/36 correct on both the meaningful and meaningless
sections. This argues against the hypothesis that a semantic
impairment of limbs knowledge is sufficient to impair actual
gestures performance, and this conclusion applied to all types
of gesture investigated.
Concerning the influence of semantic representation of
limbs on semantic representation of actions, we gained
insight from an action naming test. The BADA test (Miceli

Discussion
In verbal tasks tapping naming and comprehension skills,
FC was more severely impaired with body parts stimuli
than with other categories. The disproportionate naming
deficit was detected using different sets of stimuli, and a
severity-matched group of aphasic controls showed that
the relative deficit with body parts was not a general finding. FC was impaired with body parts on verbal tasks, but
not on tasks devoid of verbal content: the normal performance on pointing tasks without verbal orders excluded
autotopagnosia.
The dissociation between autotopagnosia and the verbal or
semantic knowledge of body parts is well represented in the
literature and forms the basis of the current theoretical models of how body parts are represented in our cognitive system.
Buxbaum and Coslett (2001), Schwoebel et al. (2001, 2004)
and Schwoebel and Coslett (2005) have suggested that there
are three separate cognitive representations of the human
body, namely “body schema”, “body semantics” and “body
structural descriptions”. According to this model, the impairment
of body structural description is associated to autotopagnosia,

Impaired Naming and Knowledge of Body Parts
while a deficit of the body schema impairs the use of body
parts for actions, and the deficit of body parts semantics
affects naming, verbal comprehension and other verbal tasks
regarding body parts. Body semantics may also contribute to
the performance of meaningful gestures. In this frame, our
patient suffered from a selective impairment of body parts
lexical-semantic knowledge, which Coslett et al. (2002) and
Schwoebel et al. (2004, 2005) have defined as “body semantics” or “body image”.
The literature reports several cases of selective impairment
or selective sparing of body parts semantics (see Introduction). Suzuki et al. (1997) traced back this selective impairment to a lesion of the left parietal lobe. Shelton et al. (1998)
reported a selective sparing of body parts for case I.O.C: this
patient presented a left occipital infarct with atrophy of the
left posterior horn of the lateral ventricle and some involvement of the temporal lobe; no parietal lesion was reported.
Coslett et al. (2002) reported another patient presenting body
parts sparing, who was affected by mild generalised atrophy,
most pronounced in the left anterior temporal lobe. In agreement with the latter reports is the fact that body parts as a category is often relatively spared among patients affected by
basal temporal or occipital lesion associated to biological categories impairment (Capitani et al., 2003). Our patient presented an extensive infarct in the areas supplied by the left
middle cerebral artery, and a restricted localization was therefore not feasible. Moreover, his deficit with body parts was
disproportionate but not selective, therefore a precise localization is perhaps not so crucial.
Moving to the cognitive aspects of this study, we found on
verbal tasks that FC was more severely impaired with limbs
than with other body parts. The contrast between limbs and
non-limbs is intriguing. Why were limbs more impaired? It
was neither due to lexical frequency, which was higher for
limbs than non-limbs, nor to an operativity factor. Other variables, such as age of acquisition and concept familiarity, were
not available for the extended set or their use was not plausible (as in the case of familiarity, because everybody is intrinsically familiar with his/her own body parts). However, we
would guess that the limbs/non-limbs discrepancy would not
change even if age of acquisition were taken into account,
because the latter parameter is generally well correlated to
lexical frequency.
We think that the higher difficulty of limbs stimuli likely
arose at the level of internal representations that need to be
activated during verbal tasks tapping body parts knowledge.
The picture of a body part, after accessing the corresponding image in the structural description system, probably
activates a chunk of knowledge which stores all the properties of that stimulus. It would be important to decide at this
point whether FC’s deficit was of a lexical or semantic
nature.
Admittedly, adjudicating between a lexical or semantic
origin is not an easy task. Some aspects of our patient, however, argue in favour of a semantic nature of the deficit, or at
least would suggest that a purely lexical impairment is not

315
sufficient to explain our data. First, the impairment of FC was
not confined to naming but also concerned pointing tasks on
verbal command and, at the preliminary examination,
extended to the semantic questionnaire. Therefore, his deficits were not confined to a single lexicon, but also affected
semantic representations or, at least, the connections between
the semantic system and the input- and output lexical stores.
Only a multiple lexical impairment could account for this
pattern, unless we postulate a disconnection between the
semantic and lexical systems. Second, the body parts categorical effect was significant also taking into account the influence of word frequency, a variable generally associated to the
lexical structure, and the body parts impairment was
observed even using frequency-matched stimuli. Third, the
error quality (vicinity and functional errors) seems more consistent with the presence of at least a semantic component of
the deficit. Finally, on the semantic questionnaire on body
parts, FC was invariably correct when he was requested to
retrieve general superordinate information, and this demonstrates that the lexical entries were not impaired in themselves. If FC’s deficit can be located at the semantic level, the
next relevant question is whether the internal representation
of limbs is different from that of non-limbs.
A first possible answer is that FC presented an apparent
disproportionate impairment of body parts overall, but was
affected by a very disproportionate impairment of limbs only.
This is equivalent to supposing that limbs are a separate category from the remainder of body parts, and that in some
patients (although no reports exist as yet) limbs could be even
selectively spared. The alternative explanation is that body
parts are a unique category, but within this category limbs are
more vulnerable, for a reason different from the effect of nuisance variables.
How can we explore the possible dependence/independence of limbs semantics from that of other body parts? In
theory, a special status for limbs could be hypothesized if
sound limbs semantics were necessary for a wider realm of
cognitive performances concerned with limbs use. To this
end, we checked (i) whether FC was affected by apraxia and
(ii) if he presented a greater impairment on action verb naming than object naming. If the limbs semantics impairment
were associated with the latter deficit(s), limbs semantics
could be pivotal for different aspects of cognition and it could
be a latent variable well suited to endowing limbs with a distinct cognitive role.
Concerning apraxia, FC was flawless on the test of
apraxia of use and was normal on a test of ideomotor
apraxia, independently of the meaningfulness of the gestures that the patient was requested to imitate. This means
that the resources necessary for performing gestures were
distinct from the internal representations necessary for the
verbal processing of information about limbs. This is partially at variance with the conclusions by Schwoebel et al.
(2004): in terms of their model, the significant predictors
of the performance of meaningful gestures were not only
the scores of “body schema” but even “body semantics”

316
tasks. In our case, FC presented a “body semantics” impairment but was not impaired while performing meaningful
gestures.
The second class of cognitive representations possibly
associated to limbs semantics is actions semantics, since
actions are mostly performed using limbs. If there were a
common cognitive representation underlying limbs semantics
and action semantics, limbs naming and action naming
should be intertwined, and for FC the latter class of stimuli
would be more defective than other categories, e.g. nonmanipulable objects. Also in this case, however, action
verbs were not defective with respect to nouns representing
non-manipulable objects.
We can come back now to the question whether the
semantic representation of limbs is clearly distinct from that
of other body parts, or if there is a unique body parts category with an internal difficulty gradient. We are inclined to
support the latter account. The first reason is that naming
latency data of normal subjects indicate that limbs stimuli
are more difficult than non-limbs stimuli, and this depends
neither on lexical frequency nor on the peculiar set of pictures used. The second invokes a parsimony principle. The
assumption that limbs are a distinct category could be supported if their semantic knowledge had a wider selective
and distinctive role in the cognitive scenario. This could be
the case, for instance, if sound limbs semantics were necessary for efficient gestures or a sound semantic representation of actions. The latter associations were not confirmed
by our data. The reason why limbs are more difficult than
non-limbs might be that proximate-limb pairs (hand-arm) or
correspondent limbs (hand-foot) are more similar than any
pairs of non-limb body parts (ear-neck, or eye-nose), and a
greater cognitive effort is required for their appropriate differentiation during semantic processing and lexical access.
This is witnessed by the frequent occurrence of semantic or
vicinity errors with limbs stimuli, and not with non-limbs
stimuli. Interestingly, this class of errors was reported in
some detailed accounts of the error types of patients
described in the literature (Semenza and Goodglass, 1985;
Semenza, 1988).
In conclusion, our case is an example of a disproportionate impairment of semantic representation of the category of
body parts. The fact that body parts can be disproportionately affected, or even selectively impaired or spared, after
left hemisphere damage is well known from the literature.
This study adds to the existing data the evidence that, within
body parts, limbs can be impaired with particular severity,
and that the impairment of body parts semantics, especially
severe for limbs, does not affect the ability to gesture and
does not impair action naming more than object naming.
Since the published reports so far have generally neglected
fine-grained distinctions within the body parts category, we
hope that future studies and a formal comparison between
limbs and non-limbs will provide further clarification on
this topic.

M. Laiacona et al.
References
Barbarotto R, Laiacona M, Capitani E. Objective versus estimated age of
word acquisition: A study on 202 Italian children. Behavior Research
Methods, Instruments, & Computers 2005; 37: 644–650.
Bortolini V, Tagliavini C, Zampolli A. Lessico di frequenza della lingua
italiana contemporanea. Milano, Italy: Garzanti, 1972.
Buxbaum LJ, Coslett HB. Specialised structural descriptions for human
body parts: Evidence from autotopagnosia. Cognitive Neuropsychology 2001; 18: 289–306.
Capitani E, Laiacona M, Mahon B, Caramazza A. What are the facts of
semantic category-specific deficits? A critical review of the available
evidence. Cognitive Neuropsychology 2003; 20:213–61.
Coslett HB, Saffran EM, Schwoebel J. Knowledhe of the human body: A
distinct semantic domain. Neurology 2002; 59: 357–63.
De Renzi E, Faglioni P. L’autotopagnosia. Archivio di Neurologia, Psicologia e Psichiatria 1963; 24: 1–34.
De Renzi E, Motti F, Nichelli P. Imitating gestures: A quantitative approach
to ideomotor apraxia. Archives of Neurology 1980; 37: 6–10.
De Renzi E, Pieczuro A, Vignolo LA. Ideational apraxia: a quantitative
study. Neuropsychologia 1968; 6: 41–52.
De Renzi E, Scotti G. Autotopagnosia: Fiction or reality? Archives of Neurology 1970; 23: 221–227.
Denes G, Cappelletti JY, Zilli T, Dalla Porta F, Gallana A.. A category-specific
deficit of spatial representation: The case of autotopagnosia. Neuropsychologia 2000; 38: 345–50.
Dennis M. Dissociated naming and locating of body parts after left anterior
temporal lobe resection: An experimental case study. Brain and Language 1976; 3: 147–63.
Di Mauro T, Mancini F, Vedovelli M, Voghera M. Lessico di frequenza dell’
italiano parlato. Milano, Italy: ETAS Libri, Fondazione IBM, 1993.
Gardner H. The contribution of operativity to naming capacity in aphasic
patients. Neuropsychologia 1973; 11: 213–20.
Laiacona M, Barbarotto R, Trivelli C, Capitani E. Dissociazioni semantiche
intercategoriali: Descrizione di una batteria standardizzata e dati normativi. Archivio di Neurologia, Psicologia e Psichiatria 1993; 54: 209–48.
Luzzatti C, Willmes K, De Bleser R. Aachener Aphasie Test (AAT). Versione italiana. Firenze, Italy: Organizzazioni Speciali, 1994.
Miceli G, Laudanna A, Burani C, Capasso, R. Batteria per l’analisi dei deficit afasici. Roma, Italy: CEPSAG, 1994.
Reed R, Farah MJ. The psychological reality of the body schema. A test
with normal participants. Journal of Experimental Psychology: Human
Perception and Performance 1995; 2:334–43.
Schwoebel J, Coslett HB. Evidence for multiple, distinct representations of
the human body. Journal of Cognitive Neuroscience 2005; 17: 543–53.
Schwoebel J, Buxbaum LJ, Coslett HB. Representations of the human body
in the production and imitation of complex movements. Cognitive
Neuropsychology 2004; 21: 285–98.
Schwoebel J, Coslett HB, Buxbaum LJ. Compensatory coding of body-part
location in autotopagnosia: Evidence from extrinsic egocentric coding.
Cognitive Neuropsychology 2001; 18: 363–81.
Semenza C. (1988) Impairment in localization of body parts following brain
damage. Cortex 1988; 24: 443–449.
Semenza C, Goodglass H. Localization of body parts in brain injured subjects. Neuropsychologia 1985; 23: 161–75.
Shelton JR, Fouch E, Caramazza A. The selective sparing of body part
knowledge: A case study. Neurocase 1998; 4: 339–51.
Sirigu A, Grafman J, Bresser K, Sunderland T. Multiple representations
contribute to body knowledge processing. Brain 1991; 114: 629–42.
Snodgrass JG, Vanderwart M. Standardized set of 260 pictures: Norms for
name agreement, image agreement, familiarity and visual complexity.
Journal of Experimental Psychology: Learning, Memory & Cognition
1980; 6: 174–215.
Suzuki K, Yamadori A, Fujii T. (1997) Category-specific comprehension
deficit restricted to body parts. Neurocase 1997; 3: 193–200.