~ Pergamon PIl: S0028-3932(97)00083 3

Neuropsychologia, Vol. 35, No. 12, pp. 1555 1563, 1997 % 1997 Elsevier Science Ltd. All rights reserved

Printed in Great Britain 0028 3932/97 $17.00+0.00

Face recognition and postero-inferior hemispheric lesions

STEPHANIE CLARKE,*t ASTRIGH LINDEMANN,* PHILIPPE MAEDER,§ FRAN(~OIS-XAVIER BORRUAT¶ and GIL ASSAL*

*Division de Neuropsychologie, Centre Hospitalier Universitaire Vaudois, 1011 Lausanne, Switzerland; $lnstitut de Physiologie, Centre Hospitalier Universitaire Vaudois, 1011 Lausanne, Switzerland; §Service de Radiodiagnostique, Centre Hospitalier

Universitaire Vaudois, 1011 Lausanne, Switzerland; ~lH6pital Ophtalmique, Centre Hospitalier Universitaire Vaudois, 1011 Lausanne, Switzerland

(Received 26 November 1996; accepted 21 May 1997)

Abstract--Two patients with severe and lasting prosopagnosia were studied for visuo-cognitive functions and anatomo-clinical correlations. Both patients were deficient in recognizing familiar faces and items of previously well known categories (plants and fish for patient 1, mountains for patient 2). Patient 2, but not patient 1, was also deficient for matching of unknown faces. Patient 1, but not patient 2, was achromatopsic. Both patients had bilateral symmetrical lesions. Patient 1 had a lesion of the inferior occipito- temporal cortex, including inferior parts of early stage visual areas. Patient 2 had a lesion of the inferior temporal and fusiform gyri anterior to the early stage visual areas. When compared in Talairach space, the lesions of both patients had minimal overlap. Thus, severe and lasting prosopagnosia was associated with two almost exclusive lesion sites in the postero-inferior part of the hemispheres. Comparison between activation studies of face processing (by others) and our lesion study uncovered several paradoxes. Lesions of regions involved in a given task in normal subjects do not produce a deficit in this task, as shown here for gender discrimination and partially for face matching. Conversely, lesions of a region not specifically involved in a given task in normal subjects can produce a deficit in the task, as shown here for face identification. ~) 1997 Elsevier Science Ltd. All rights reserved

Key Words: prosopagnosia; visual agnosia; extrastriate visual areas; man.

Introduction

Prosopagnosia, defined as failure to recognize previously known faces, has often been described in association with inferior temporo-occipital lesions [10, 13, 19]. In most published cases, prosopagnosia was accompanied by agnosia for other visual categories; achromatopsia, agnosia for colours; pure alexia; visuo-spatial or visuo- constructive disorders; memory deficits and, in most cases, visual field deficits. Cases ofprosopagnosia without visual field deficit [17] and without achromatopsia [3, 18] are relatively rare. In most reported cases, prosopagnosia was associated with bilateral or right unilateral inferior temporo-occipital lesions [9, 10, 13].

Work of the past 20 years has shown that the visually- related cortex of non-human primates consists of a large

t Address for correspondence: Division de Neuro- psychologie, CHUV, 1011 Lausanne, Switzerland; tel.: +41 21 314 10 92; fax: +41 21 314 13 19; e-mail: Stephanie. Clarke@chuv.hospvd.ch.

number of functionally defined visual areas. In macaque monkeys, over 30 areas have been identified in the extra- striate visual cortex by electrophysiological and ana- tomical criteria [16, 28]. Some of these areas contain a high proportion of functionally selective neurons, such as the colour-specific neurons in V4 and the motion- specific neurons in V5 [28]. Other, less precisely defined areas have been described more rostrally in the parietal and temporal cortices [16]. In particular, neurons selec- tively responsive for faces have been found in the infero- temporal cortex [12], the superior temporal sulcus [20] and the amygdala [22].

The human occipital cortex is classically subdivided into three cytoarchitectonically defined areas [4]. Area 17 is known to contain a single (the primary) visual area. In contrast, areas 18 and 19 are believed to contain multiple visual areas, defined as individual representations of the contralateral visual hemifield; this is known to be the case in monkeys, where several boundaries between visual areas are characterized by bands of callosal afferents and/or by changes in myeloarchitecture (for review see,

1555

1556 S. Clarke et al./Prosopagnosia and visual areas

e.g., [16]). Trac ing o f h u m a n visual cal losal connect ions [6] and re la ted cyto- and myeloarch i tec ture led to the ident i f icat ion of the boundar i e s o f several ear ly stage ext ras t r ia te visual areas. As in macaque monkeys , these areas tend to be referred to as V2, V3, V4 and V5 MT; they are re t ino topica l ly organized and are believed to sustain ear ly stage visual process ing [11, 23, 27].

We invest igated two cases o f severe and last ing pro- sopagnos ia and agnos ia for previously well known cat- egories. The aim of this work was three-fold: (i) to assess the role o f ear ly stage visual areas in p rosopagnos ia ; (ii) to de te rmine types o f face recogni t ion deficits result ing f rom ana tomica l ly dist inct lesions; and (iii) to compa re the sites o f foci ac t iva ted selectively by face st imuli in no rma l subjects ( f rom studies by others) with the sites of lesions p roduc ing face recogni t ion deficits. Pre l iminary results were publ i shed in abs t rac t form [8].

Case histories and methods

Case 1 is that of a 57-year-old right-handed man, who was trained as a florist but worked at the onset of the disease as a salesman in fish and game. At the age of 38 years he received an artificial aortic valve. Its replacement 3years later was accompanied by left motor hemisyndrome, which recovered completely within 6 months. The present illness started abruptly with blindness, disorientation and memory deficits. Within a few hours the patient recovered parts of the visual field, first the right and then the left lower quadrant. He described the partially recovered visual field as devoid of colours. The com- puter tomography (CT) scan performed on the ninth day after the onset of the symptoms showed hypodense lesions in the territory of the posterior cerebral artery on both sides, com- patible with recent infarctions (Fig. 1), and another hypodense lesion on the right frontal convexity compatible with an old infarction. Goldman perimetry was performed 6weeks after the onset of the symptoms and showed superior altitudinal hemianopsia. Within the first weeks, the main complaint was the inability to recognize faces, the absence of colour and the loss of familiarity. Later on, the patient described this initial period as "walking in a world which was no longer familiar".

Neuropsychological examinations were performed every 2 4 days during the acute phase and every 2 5 weeks later. During the first 3weeks the patient presented severe letter-by-letter alexia, and deficits in mental arithmetic, visuo-constructive praxias and in an alternating graphic task (Luria); in contrast, speech production, comprehension and repetition, naming of drawings of objects (French adaptation of the Boston naming test), ideomotor praxias, tactile recognition and naming of objects, verbal fluency and progressive matrices were within normal limits. At a later stage, the patient improved in reading and was able to read a text slowly, but remained deficient in reading non-words. His performance at mental arithmetic and alternating tasks (Luria) remained poor. Memory testing revealed sparing of short-term memory and preserved learning of new skills, but major deficits in long-term acquisition of new material. Visuo-cognitive and visuo-spatial testing revealed the presence of severe achromatopsia, prosopagnosia, visual agnosia for specific categories of items, and topographical dis- orientation (also see below). In the acute stage, the patient was severely deficient in disentangling spatially overlapping figures, but he improved his performance later. At both the acute and chronic stages, he was deficient in the Hooper Visual Organ- ization and in the Columbia Visual Association tests. Texture

discrimination, judgement of line orientation and orientation on a map were within normal limits. Colour discrimination was impossible in the acute stage and the patient described the world around him as devoid of colour. Later, colour and hue discrimination improved, but the patient still remained deficient on the Farnsworth 15 hue test (desaturated).

Case 2 is that of a 50-year-old right-handed woman, a nurse and, in her free time, a keen mountaineer. At the age of 39 years, she suffered a closed head injury in a mountaineering accident. A CT scan showed bilateral inferior temporal haematomas which were evacuated surgically. During the first postoperative week, the patient was comatose and was then in a confusional state for 4 weeks. Neurological examination revealed an absence of motor and sensory deficits, and normal eye movements, visual field, visual acuity and colour discrimination. A C T scan performed after the evacuation showed bilateral posterior baso- temporal hypodensity. The patient was regularly examined in our Division between 2 and 11 years after the accident. During this period her performances changed little. We report here in detail the last evaluation performed 11 years after the injury. Magnetic resonance imaging (MRI) performed at this time showed bilateral symmetrical hypointense lesions of the middle parts of the medial and inferior temporal and the fusiform gyri (Fig. 2). The hippocampal formation and the parahippocampal gyrus were preserved. Kinetic Goldmann visual fields and col- our vision were normal. The main complaint was the inability to recognize faces and previously known parts of a nearby town, and a loss of pleasure in seeing mountains. The villages and the town of the region she lived in have lost their familiarity and have become "strange and sometimes mysterious".

Neuropsychological assessment showed preserved language functions except for a moderately deficient naming of drawings of objects, preserved tactile recognition of objects and their naming, and preserved visuo-constructive and ideomotor prax- ias. Different tests of verbal and non-verbal fluencies were mod- erately deficient or at the lower range of normal performance. The performance in several tests of executive functions was within normal limits. Intelligence tests revealed a low average intellectual ability. Memory testing revealed sparing of short- term memory, but major deficits in long-term acquisition of new material. Visuo-cognitive and visuo-spatial testing revealed severe prosopagnosia, visual agnosia for specific categories of items and topographical disorientation, but fully normal colour perception. Recognition of spatially overlapping figures and orientation on a map were within normal limits.

Anatomical analysis and interindividual comparisons

Individual human brains vary greatly in size and shape. This represents a serious problem for comparison of data collected in different subjects. One solution currently adopted in brain imaging studies consists of comparing locations within different brains by means of normalized coordinate systems. Talairach and Tournoux [26] introduced a proportional grid and a coor- dinate system anchored in the forebrain commissures. Three planes are placed within the brain, corresponding to a hori- zontal plane through the anterior and posterior commissures and two coronal planes through the anterior and posterior commissures respectively. The hemispheres are subdivided into 12 horizontal (four below and eight above the bicommissural plane), 11 coronal (four anterior to the anterior commissure, three between the anterior and posterior commissures, four posterior to the posterior commissure) and eight sagittal slices (four for each hemisphere). The subdivision and the cor- responding stereotaxic coordinates in millimetres have been published for a reference brain [26]. We used Talairach coor- dinates to describe the lesions of our two patients. The site and extent of each lesion as apparent on the MRI or CT scan were

S. Clarke et al./Prosopagnosia and visual areas 1557

Fig. 1. CT scan of patient 1. Horizontal sections through the lower (A, B) and upper (C, D) parts of the visual cortex are shown; left is to the right. The occipito-temporal lesion was compatible with a recent infarction and was slightly larger on the left side.

determined within the corresponding horizontal, coronal and parasagittal planes of the proportional grid and represented on the standard brain.

Results

Recognition of previously known.faces

Patient 1 was severely deficient in the recognition of previously known faces. For the first days after the lesion he was unable to recognize the face of his wife by visual cues and learned only slowly to rely on extrafacial indi- cations, such as hair, ears and clothing. Identification of

faces of famous people was tested at different times after the onset of the symptoms. The patient was shown photo- graphs of famous people and asked to name them or to give biographical details. On day 17 the patient identified correctly none of six photographs presented, at 6 weeks none of six, at 10weeks four of six, at 3months two of seven, at 5 months none of eight, at 8 months three of 14, at 11 months none of six, and at 17.5 months one of six. When the patient was given the name of a famous person, which he said he knew, and asked to indicate the photo- graph of the person among several other photographs, he was severely impaired. However, he was accurate in describing the face on the photograph, indicating the sex, age, facial expression and often the profession of the

1558 S. Clarke et al./Prosopagnosia and visual areas

Fig. 2. MRI of patient 2 performed 11 years after the accident. Tl-weighted images (A-E) of parasagittal sections from the right (A, C, E) and left (B, D) hemispheres, and T2-weighted image from a coronal section (F). The lesions were symetrical and comprised the middle parts of the medial and inferior temporal gyri (A-D, F), as well as the fusiform gyrus (F); the hippocampal formation

and the parahippocampal gyrus were spared on both sides (F).

person. In particular, he was able to classify correctly photographs of famous people according to their pro- fession (actors, politicians). When given the names of famous people he was able to give their biographical details.

Patient 2 was severely deficient in the recognition of previously known faces. She complained about being unable to recognize her family and friends. She noticed that her difficulty was greater for male and children's than for female faces. She was severely deficient in identifying famous people from photographs of faces. When pre- sented with 15 photographs she identified none of them, although all 15 names were familiar to her and she could give biographical details of these people. Like patient 1, she was accurate in describing the face on the photo-

graph, indicating the sex, age and facial expression. She was unable to classify photographs of famous people into categories according to their profession. Conversely, indications as to the profession of a given celebrity did not facilitate the recognition of the photograph. Covert recognition of famous faces was tested in a learning task of congruent and non-congruent face name associations; no advantage for congruent associations was observed.

Matching of unfamiliar and familiar faces

The matching of unfamiliar faces was tested with the Benton test of facial recognition [2]. Patient 1 was tested 6 weeks after the onset of the symptoms with a shortened

S. Clarke et al./Prosopagnosia and visual areas 1559

version of the test and was found to be at the lower limit of normal performance (17 right choices out of 27). At 8 months he was tested with a long version and was found to be within normal limits (42 right choices out of 54). The matching of different photographs of famous people (different standard views, taken on different occasions) was preserved throughout the follow-up.

Patient 2 was moderately deficient on the long version (37 right choices out of 54). An additional test of mat- ching of unknown faces was constructed with con- temporary Swiss faces; the patient had a very poor performance (four right choices out of 18). She was severely impaired in matching photographs of famous people (different standard views, taken at different occasions).

Visual recognition within specific categories

Patient 1 was severely deficient in recognizing pre- t~iously known plants andfish, in both of which categories he had professional knowledge. Recognition of plants was tested by presenting colour photographs or coloured drawings of common house and garden plants; he had to name the plant or, if unable to do so, to describe it and then to choose the most likely name from three to four possible ones. In the first session, 4 weeks after the onset of his disease, he was unable to name, or to choose the right name, for flowers as common as a daffodil, fuchsia or hyacinth, while he gave relatively accurate descriptions of the pictures. He gradually performed better, mainly by describing verbally the picture and then comparing this description with his knowledge of plants. However, even a year later he was still deficient in recognizing photographs of common plants (Fig. 3). Recognition of fish was tested by presenting colour photographs of freshwater and sea fish and asking him to name them or, if unable, to describe them and to choose the right name from two to three possible ones. The patient performed very poorly; he was unable to propose a name for most items, and when he did he was wrong. In most instances he was accurate in some details of visual description, such as the stripes of a perch or the dots on a trout, but often these details lead to misidentifications, e.g., the dots on a pike lead to the (wrong) identification as a trout. Fur- thermore, patient 1 was deficient in recognizingpre~,iously known handwritin9 of members of his family. He was unable both to identify samples as belonging to his wife, his brother etc., and to pick them out by a sense of familiarity. He was also unable to match two samples of handwriting of the same person from a selection of others. Patient 1 also had difficulties in recognizing build- ings. When he returned home from the hospital he felt that his house was not as he remembered, e.g., the ceilings were much lower. Eater, he reported difficulties in ident- ifying previously known houses in his village. In a test of known buildings of his region his performance was severely deficient when compared to that of his wife.

100 %-

50 %

X X

X

X X X

0

×x#x 0

Recogn. Recogn. Classif. plants mountains mountains

• case 1 0 case 2

X controls

Fig. 3. Recognition of items from previously well known cat- egories. Recognition of common house and garden plants ('Recogn. plants' in the figure) was tested by presenting the subjects colour photographs or coloured drawings. Controls were florists matched for age to patient 1. Recognition of moun- tains and Alpine landscapes ('Recogn. mountains') was tested by presenting coloured photographs of views from the region which patient 2 knew well. Controls were members of the Swiss Alpine Club matched for age to patient 1, but living in Lausanne and thus having only limited mountaineering experience in the region represented in the photographs. Classification of moun- tain landscapes into Swiss Alpine and extra-European ones ('Classif. mountains') was tested by presenting coloured photo-

graphs. Same controls as for previous test.

Patient 2 was severely deficient in recognizing pre- t, iously known mountains and Alpine landscapes. She said that since her accident previously known views seemed unfamiliar to her and that she did not experience pleasure in looking at them. Identification of mountains and Alpine landscapes was tested by presenting the patient with colour photographs of views from regions she said she knew well and asking her to name or to locate them (Fig. 3). Although she could name a few typical views from the Swiss Alps, she was unable to identify several views from the valley she lived in. Another test was con- ceived to investigate whether the patient was able to distinguish different types of mountain landscapes. Col- our photographs from either the Swiss Alps or from non- European mountain regions (Himalayas, Mount Kenya, Fudjiyama, etc.) had to be classified as ~Swiss' or ~not Swiss'. The patient performed in this test less well than all controls (Fig. 3). Patient 2 was able to recognize her own handwritin 9 among several others. However, she was unable to pick two samples of the handwriting of the same person from among others, even if the written text was the same in all samples. The patient complained of being unable to recognize previously known buildings in her village and in a nearby town, and she was deficient in a test of recognition of famous European monuments.

1560 S. Clarke et al./Prosopagnosia and visual areas

Localization of the lesions in relation to early stage visual 61reas

the fusiform gyrus, between - 2 6 and - 3 7 r a m of the anteroposterior axis (Fig. 4).

Both patients had bilateral lesions in the inferior parts of the hemispheres. The lesion of patient 1 comprised the inferior part of the occipital lobe and the infero-posterior part of the temporal lobe; it was larger on the left than on the right (Fig. 1). We compared the Talairach and Tournoux coordinates [26] of the lesion to that of pre- viously determined visual areas [6]. On the right side, the lesion comprised the postero-inferior part of V1, the inferior part of V2, and the whole of VP and V4; the anterior limit of the lesion was ca 1 cm anteromedial to the anterior boundary of V4 (Fig. 4). On the left side, the lesion comprised the postero-inferior part of V1, the inferior part of V2, and the whole of VP and V4, extend- ing over 3 cm medially and anteriorly beyond V4 (Fig. 4). The lesion of patient 2 was symmetrical and was limited on both sides to inferior temporal and fusiform gyri. On the right side, the most anterior point of the lesion was at - 5 mm and the most posterior point at - 39 m m of the anteroposterior axis. On the left side, the most anterior point was at - 7 mm and the most posterior at - 3 7 m m . In both hemispheres, the lesion did not encroach on early stage visual areas and was at least 2 cm anterior to the anterior boundary of V4. In the right hemisphere, the lesions of patients 1 and 2 did not over- lap; in the left hemisphere, there was a small overlap on

D i s c u s s i o n

Both our patients were severely prosopagnosic, having difficulties in recognizing known people in everyday life and photographs of famous people. Both were accurate in describing the face and indicating the sex, age and facial expression. Patient 1, but not patient 2, was able to classify correctly photographs of famous people according to their profession and to match correctly photographs of known and unknown faces. In both cases the recognition deficit was not limited to faces, but extended to other categories well known by the patient. It is a surprising feature that a purely associative type of prosopagnosia was associated with a posterior lesion (patient 1), whereas an associative plus apperceptive type of prosopagnosia was found with a more anterior lesion (patient 2).

A large part of the lesion of patient 1 was within early stage visual areas and the underlying white matter. Only a small part of it was estimated to be anterior to visual area V4 (Fig. 4). Visual areas V1, V2, V3 and V4 are retinotopically organized [11, 23], and V4 appears to be involved in colour processing [27]. Activation studies indicate that in normal subjects these areas are not

, , ~ , , , h I ' i I I J I I b

NV1 NV2 []]]]V3 [~VP ~]V4 [~V5/MT

I I I I I I I i i [ i i i i i i

I I I I I I ~ I I [ t I I I I I i , i , , , I i i i i , i i i i

Fig. 4. Comparison of lesions in cases 1 and 2 and the positions of putative visual areas as determined in a previous study [6]. The CT scan of patient 1 and the MRI of patient 2 were assessed with the proportional grid technique [26]. The location and the extent of the lesions are reported here on the standard brain. The primary visual area (VI) and five early stage extrastriate visual areas (V2, V3, VP, V4 and V5) as determined anatomically are indicated [5, 6]. Note that lesions of patient 1, but not those of patient 2, include

several early stage visual areas. Talairach coordinates are indicated; long marks indicate 0 points, small marks 1 cm distances.

S. Clarke et al./Prosopagnosia and visual areas 1561

directly involved in identification of previously known faces [24]. The fact that patient 1 was severely and last- ingly impaired in identification of previously known faces can be interpreted in two ways. Either the damaged early stage areas and the immediately adjacent cortex provide critical input to the more anterior temporal cortex involved in face identification, or the white matter below these areas conveys such input (e.g., from upper V1 and V3, or V3A) to the anterior part of the inferior temporal cortex. It is striking that the lesion of the ventral part of early stage visual areas did not lead to apperceptive defi- cits in face recognition. The lesion of patient 2 was entirely anterior to the early stage visual areas. The pati- ent was severely impaired in identification of previously known faces and moderately impaired on face matching. It is very likely that her lesion was within a region that is more specifically involved in face processing than the early stage visual areas.

Activation (positron emission tomography) and elec- trophysiological studies indicate that the perception of faces is associated with neural activity in the inferior part of the occipito-temporal cortex. Sergent et al. [24] compared the activation patterns during face identi- fication and gender categorization. Regions activated selectively during a gender categorization task (control task =grating categorization) were found posteriorly in the inferior part of the occipital cortex. Regions activated selectively during face identification task (control task = gender categorization) were found more anteriorly in the inferior part of the temporal lobe (Fig. 5). Haxby et al. [14] showed that matching of unknown faces was associated with activation of the occipital and temporal parts of the fusiform gyrus. Allison et al. [1] recorded a surface-negative potential (N200) that was selective to faces in the posterior temporal part of the fusiform gyrus; their region was co-extensive with the region activated by passive viewing of faces [21] and by matching of unknown faces [14] in the right hemisphere, but was anterior and lateral to it in the left hemisphere (Fig. 5). In another activation study, Haxby et al. [15] investigated the acti- vation patterns during a short-term memory task for (unknown) faces (Fig. 5). They identified a region on the occipital and posterior temporal parts of the fusiform gyrus, the activation of which decreased with increasing memory delay; they proposed that this region is primarily involved in face perception. A small region on the left inferior temporal gyrus and other regions outside the inferior part of the temporal and occipital lobes (not discussed here) were increasingly activated with longer delays; the authors concluded that these regions were involved in maintaining a face representation in working memory.

The four activation studies [14, 15, 21, 24] and the electrophysiological study [1] are compatible with a two-

stage model o f face processing [25]. The first stage is identi- fication independent and includes discrimination of faces versus other objects, gender discrimination and com- parison between faces such as that involved in matching

of unknown faces. It was proposed to be associated with neural activity in the inferior occipital and the infero- posterior temporal cortex. The second stage is identi- fication linked and includes the recognition of known faces; it was proposed to be associated with neural activity in the inferior part of the temporal cortex, includ- ing its very anterior parts.

The comparison of the above-described face activation studies and our cases uncovers a double paradox. First, bilateral destruction of the postero-inferior regions acti- vated by face matching and gender discrimination did not lead to the impairment of these functions (Fig. 5); the deficit observed in patient 1 was an associative type of prosopagnosia. Second, bilateral destruction anterior to the regions activated by face matching lead to an impairment in face matching. Thus, the participation of a given region in a given function in normal subjects does not imply that this area is essential for the realization of that function after brain damage.

The significance of the discrepancy between activation and lesion studies is not immediately clear. Several points need to be considered.

1. The main comparison we are proposing is between the study of Sergent et al. [24], who analysed mean state- dependent changes in blood flow by averaging across seven subjects, and our two patients taken indi- vidually. De Renzi et al. [10] have argued that there is great individual variability in the hemispheric spe- cialization for face recognition. Similar variability may exist for the intrahemispheric distribution of face processing units. More observations with accurate anatomoclinical correlations are needed.

2. A lesion within a small brain region may affect the functional integrity of a distant region, e.g., a small lesion limited to a part of the left anterior thalamic nuclei caused a chronic hypometabolism in the left posterior cingulate cortex which is believed to receive afferents from the anterior thalamic nuclei [7]. De Renzi et al. [10] studied the cerebral glucose metab- olism by means of the [~SF]deoxyglucose method in three prosopagnosic patients with postero-inferior lesions. They have shown that apparently ana- tomically intact regions within the ipsilateral occipital and temporal lobes were hypometabolic.

3. Regions involved in a given function in normal sub- jects under normal circumstances may not be essential to this function. Other circuits may be involved in cases of brain damage, perhaps sustaining different strategies that can be used for this function. Our results indicate that this could be the case for face matching or gender discrimination. Patient 1 had a lesion that included foci activated by these tasks in normal subjects (Fig. 5) and yet he did not present the corresponding deficits. However, compared to normal subjects he performed more slowly on these tasks, scanning carefully in a detailled fashion the faces pre- sented to him.

1562 S. Clarke et al./Prosopagnosia and visual areas

right left

4

case 1 case 2

right left

J

~gender discrimination vs gratings [ ] face identification vs gender discrimination

Sergent et al. 1992

right left . . . . .

'~';~: matching of unknown faces Haxby et al. 1994

~ _ ~ N200 to unknown faces All ison et al. 1994

right left : T

Working memory for faces: , - ' , - ' , . " , .

.2.2..".. activation decrease with delay

~,','~%~; activated with long delay (21s)

Haxby et al. 1995

Fig. 5. Comparison of lesion sites in our cases and activation studies of face processing. Top left: sites of lesions as compared to area V4; top right and bottom left and right: lesions (outlined) compared to foci of activation as indicated in quoted studies. Note that (i) lesion of regions involved in gender discrimination did not produce gender discrimination deficit (compare [24] and case 1); (ii) lesion of large parts of regions involved in matching of unknown faces did not produce deficit in matching of unknown faces (compare [1, 14] and case 1); (iii) lesion outside regions involved in matching of unknown faces produced deficit in matching of unknown faces (compare [1, 14] and case 2); (iv) lesion comprising some of the loci specifically activated during face identification produced prosopagnosia (compare [24] and case 2); and (v) lesion amidst the foci specifically activated during face identification

produced prosopagnosia (compare [24] and case 1).

4. The correct functioning of early stage areas, and hence performance they normally sustain in normal subjects, may depend on feedback f rom higher order areas. Such an interpretation is consistent with the case O.R. [10]; he had a right temporo-parietal lesion involving inferior and middle temporal, as well as supramarginal and angular, gyri. This lesion was associated with hypometabol ism in the ipsilateral medial, inferior and lateral occipital cortex. This type o f feedback influence could explain some rather puzzling deficits observed in patient 2. She had a lesion that was outside the

region activated by face matching in normal subjects and yet she was deficient in this task.

Acknowledgements We are very grateful to Prof. F. Regli for referring to us for neuropsychological diagnosis and rehabili- tation patient 1, and to Dr J. de Preux patient 2. We thank Dr P. G. H. Clarke for comments on the manuscript. This work was supported by Swiss National Science Foundation grants 31-362987.92 and 3231-41607 to S. Clarke.

S. Clarke et al./Prosopagnosia and visual areas 1563

References

1. Allison, T., Ginter, H., McCarthy, G., Nobre, A. C., Puce, A., Luby, M. and Spencer, D. D., Face recognition in human extrastriate cortex. Journal of Neurophysiology, 1994, 71,821 825.

2. Benton, A. L., Van Allen, M. W., Hamsher, K. deS. and Levin, H. S., Test of Facial Recognition, Form SL. Stimulus and Multiple Choice Pictures. Depart- ment of Neurology, University of Iowa Hospitals, Iowa City, IA, 1975.

3. Boudouresques, J., Poncet, M., Alicherif, A. and Bal- zamo, M., Agnosie des visages. Bull. Acad. Nat. Mbd., 1979, 163, 693 702.

4. Brodmann, K., Vergleichende Localisationslehre der Grosshirnrinde in ihren Prinzipien dargestellt auf Grund des Zellenbaues. Johann Ambrosius Barth, Leipzig, 1909.

5. Clarke, S., Modular organisation of human extra- striate visual cortex: evidence from cytochrome oxi- dase pattern in normal and macular degeneration cases. European Journal of Neuroscience, 1994, 6, 725-736.

6. Clarke, S. and Miklossy, J., Occipital cortex in man: organization of callosal connections, related myelo- and cytoarchitecture, and putative boundaries of functional visual areas. Journal of Comparative Neur- ology, 1990, 298, 188 214.

7. Clarke, S., Assal, G., Bogousslavsky, J., Regli, F., Townsend, D. W., Leenders, K. L. and Blecic, S., Pure amnesia after unilateral left polar thalamic infact: topographical and sequential neuro- psychological and metabolic (PET) correlations. Journal of Neurology, Neurosurgery and Psychiatry, 1994, 57, 27 34.

8. Clarke, S., Lindemann, A. and Assal, G., Pro- sopagnosia with and without achromatopsia: relationship with early stage visual areas. European Journal of Neuroscience, Supplement, 1995, 8, 33.

9. Damasio, A. R., Damasio, H., Van Hoesen, G. W., Prosopagnosia: anatomic basis and behavioral mech- anisms. Neurology, 1982, 32, 331 341.

10. De Renzi, E., Perani, D., Carlesimo, G. A., Silveri, M. C. and Fazio, F., Prosopagnosia can be associ- ated with damage confined to ther right hemi- sphere-an MRI and PET study and a review of the literature. Neuropsychologia, 1994, 8, 893 902.

11. De Yoe, E. A., Carman, G., Bandettini, P., Glick- man, S., Wieser, J., Cox, R., Miller, D. and Neitz, J., Mapping striate and extrastriate visual areas in human cerebral cortex. Proceedings of the National Academy of Science U.S.A., 1996, 93, 2382 2386.

12. Gross, C. G., Rocha-Miranda, C. E. and Bender, D. B., Visual properties of neurons in inferotemporal cortex of the macaque. Journal of Neurophysiology, 1972, 35, 96-111.

13. Gr~sser, O. J. and Landis, T., Visual Agnosias and

Other Disturbances of Visual Perception and Cognition. Macmillan, London, 1991.

14. Haxby, J. V., Horwitz, B., Ungerleider, L. G., Maisog, J. M., Pietrini, P. and Grady, C. L., The functional organization of human extrastriate cortex: a PET-rCBF study of selective attention to faces and locations. Journal of Neuroscience, 1994, 14, 6336- 6353.

15. Haxby, J. V., Ungerleider, L. G., Horwitz, B., Rapo- port, S. I. and Grady, C. L., Hemispheric differences in neural systems for face working memory: a PET- rCBF study. Human Brain Mapping, 1995, 3, 68 82.

16. Knierim, J. J. and Van Essen, D. C., Visual cortex: cartography, connectivity, and concurrent process- ing. Current Opinion in Neurobiologt)', 1992, 2, 150 155.

17. Levin, H. S. and Peters, B. H., Neuropsychological testing following head injuries: prosopagnosia with- out visual field defect. Diseases of the Nervous System, 1976, 37, 68 71.

18. Michel, F., Perenin, M. T. and Sieroff, E., Pro- sopagnosie sans hemianopsie apr6s 16sion unilaterale occipito-temporale droite. Rev. Neurol., 1986, 142, 545 549.

19. Michel, F., Poncet, M. and Signoret, J. L., Les lesions responsables de la prosopagnosie sont-elles toujours bilaterales. Rev. Neurol., 1989, 145, 764 770.

20. Perrett, D. I., Rolls, E. T. and Caan, W., Visual neurons responsive to faces in the monkey temporal cortex. Experimental Brain Research, 1982, 47, 329- 342.

21. Puce, A., Allison, T., Gore, J. C. and McCarthy, G., Face-sensitive regions in human extrastriate cortex studied by functional MRI. Journal of Neuro- physiology, 1995, 74, 1192-1199.

22. Rolls, E. T., Neurons in the cortex of the temporal lobe and in the amygdala of the monkey with responses selective for faces. Human Neurobiology, 1984, 3, 209 222.

23. Sereno, M. 1. A., Dale, M., Reppas, J. B., Kwong, K. K., Belliveau, J. W., Brady, T. J., Rosen, B. R., Tootell, R. B. H., Borders of multiple visual areas in human revealed by functional MRI. Science, 1995, 268, 889 893.

24. Sergent, J., Ohta, S. and MacDonald, B., Functional neuroanatomy of face and object processing. A posi- tron emission tomography study. Brain, 1992, llfi, 15 36.

25. Sergent, J. and Poncet, M., From covert to overt recognition of faces in a prosopagnosic patient. Brain, 1990, 113, 989-1004.

26. Talairach, J. and Tournoux, P., Co-Planar Stereo- taxic Atlas of the Human Brain. Thieme Medical, New York, 1988.

27. Zeki, S., A century of cerebral achromatopsia. Brain, 1990, 113, 1721-1777.

28. Zeki, S. M., Functional specialization in the visual cortex of the rhesus monkey. Nature, 1978,274, 423- 428.