the anatomical basis of prosopagnosia

Journal of Neurology, Neurosurgery, and Psychiatry, 1974, 37, 489-501

The anatomical basis of prosopagnosiaJ. C. MEADOWS'

From the Aphasia Unit, Boston V.A. Hospital, Boston, U.S.A.,and National Hospital, Queen Square, London

SYNOPSIS Evidence is presented that patients with prosopagnosia have right anterior inferioroccipital lesions in the region of the occipital temporal junction. Many if not all cases have anadditional lesion in the left hemisphere; this is often but apparently not always symmetrical withthe right hemisphere lesion. This evidence is discussed in relation to the anatomical connections ofthese regions and the results of experiments in animals.

Prosopagnosia is the specific inability to recog-nize familiar faces. Its clinical features have beenwell reviewed by previous authors-for example,Hecaen and Angelergues (1962). It will bepointed out here only that the severely prosopa-gnosic patient typically has no difficulty recogniz-ing everyday objects, although he may be quiteunable to recognize even members of his familyunless they speak, when he immediately recog-nizes their voices. Some severely prosopagnosicpatients may recognize certain people by utilizingsome specific attribute, such as a mole on theface, unusual height, spectacles of particulardesign, individual clothes, etc. However thesefeatures can be described in verbal terms, and thevisual percept of faces without such features doesnot lead to recognition.

Patients with this rare symptom have poster-iorly placed cerebral lesions with physical signssuggesting bilateral involvement in some casesbut in others there is clinical evidence only ofright hemisphere disease. Visual field defects arenearly always present, but these are not sufficientto account for the prosopagnosia, for othervisual material is usually perceived normally.Unfortunately, pathological studies have beenfew and have not led previous authors to anyfirm conclusions on the anatomical lesions in-volved. Rondot and Tzavaras (1969), who havereviewed the subject extensively, mostly fromthe neuropsychological point of view, haveI Present address: The National Hospital, Queen Square, LondonWCIN 3BG.

489

recently drawn attention to the occurrence ofbilateral lesions in published cases that havecome to necropsy, although they did not con-sider in detail the possible significance of thelocation of these lesions.

In the present author's view, there is alreadygood evidence concerning the localization oflesions in prosopagnosia. This conclusion isbased upon an analysis of clinical case reportsand a re-evaluation of pathological findings inreported cases. It is the purpose of this paper tomarshal this evidence and discuss possibleunderlying mechanisms.

NEUROPSYCHOLOGICAL ASPECTS

The possible significance of the anatomicallesions to be described will be better understoodif consideration is first given to the normal pro-cess of facial recognition and then to the neuro-psychological features of prosopagnosia.

FACES AS A CATEGORY It should be realized thatrecognition of faces is a most complex andsophisticated visual achievement which is par-ticularly gestalt-like in nature, for most faces areresistant to verbal interpretation. We can eachprobably recognize more than 1,000 faces, themajority of which differ in fine detail, but we dothis in a manner which it is quite impossible toconvey verbally. It depends on learning what areessentially visual pattern discriminations of greatcomplexity.

J. C. Meadows

There is no obvious counterpart to facialrecognition in other visual experience. Thus,there is no other visual category within which weare required to distinguish and remember suchminor differences in detail by predominantlynon-verbalizable means. It can be argued thatcertain individuals have special experience withparticular categories (the forester with trees, thehorticulturalist with flowers, etc.) and in thisrespect it is interesting that Bornstein (1963)described a birdwatcher with prosopagnosiawho could no longer differentiate similar birdsand a prosopagnosic farmer (Bornstein et al.,1969) who could no longer identify his individualcows. These however are special cases. Neverthe-less, the prosopagnosic patient of DeRenzi et al.(1968) had difficulty in distinguishing fruit (anapple from a peach) and there have been somecases with inability to distinguish particularchairs (Faust, 1955) or buildings (Pallis, 1955;Beyn and Knyazeva, 1962; Cole and Perez Cruet,1964). These are less striking than the prosopa-gnosia possibly because they are less difficultvisual tasks and also because they are less purelyvisual (since it is easier to describe the differencesin verbal terms than with faces).

Cases such as those just mentioned are relevantto the question of whether there are specialcharacteristics (apart from the complexity of thetask) about the brain's ability to distinguish faces.Certain experimental observations suggest thatthe brain processes information about faces in adifferent way from other visual material (see forexample, Yin, 1970). This is borne out by theeveryday observation that we are strikinglybetter at recognizing a face as familiar after asingle encounter than we are other material-forexample, an individual horse or a tree. However,this does not mean that skill in perception offaces is a fundamental and primitive capacity ofthe brain, as has been argued by some authors-for example, Bodamer (1947). It may be that welearn to distinguish faces to a degree not seenwith other categories because facial recognitionis from the very earliest age and throughout lifesuch an essential and determining aspect ofdaily living. Thus, it can be argued that we mighthave acquired the same perceptual skill in rela-tion to the configuration of one tree relative tothe next, if tree configuration were as major adeterminant of behaviour as is faces. The farmer

and bird watcher mentioned in the last para-graph who had acquired skills with othercategories, and subsequently lost them whenthey developed prosopagnosia, would tend tosupport this 'acquired' view, as does the well-known observation that 'all Chinamen lookalike' to the untravelled Westerner, but not tothe Westerner who has spent years in theOrient.The cautious conclusion is thus reached that

there are special characteristics about the brain'sability to discriminate and remember faces butthese may be acquired rather than inborn. Thesame skills may not be acquired with othercategories simply because there is no culturalpressure to acquire them. Accordingly, facialrecognition becomes by far the most complexand frequently encountered example of relativelypure visual discrimination learning that occurs ineveryday life. Considered in this way it becomesless surprising that it may be disturbed in rela-tive isolation.

PROSOPAGNOSIA AND THE AMNESIC SYNDROMEPerhaps the most important recent discovery inrelation to prosopagnosia is that patients withthis condition, who are unable to recognizefamiliar faces, perhaps even those of closefriends and family, may be able to discriminateand match faces normally. In other words thedisturbance appears to be one of memory forfaces. To understand this better, it is helpful toconsider some of the ways in which facialrecognition has been studied, together with theresults that have been obtained.A considerable literature has built up in

recent years showing that the right cerebralhemisphere plays a special role in visual percep-tion and visual memory, and indeed patientswith right hemisphere lesions who are notsuffering from clinical prosopagnosia performless well than those with left hemisphere lesionson scored tests involving facial discrimination(DeRenzi and Spinnler, 1966; Warrington andJames, 1967; Benton and Van Allen, 1968;Tzavaras et al., 1970). Such tests have been ofthree broad types:

1. Those testing visual memory for faces-suchas recognition of photographs of well-knownpublic figures (Warrington and James, 1967), or

490

The anatomical basis ofprosopagnosia

of previously unknown faces after an appropriatedelay (Milner, 1968). This type of test is the mostclosely related to the clinical symptom of pro-sopagnosia and was shown statistically by theseauthors in unselected series of patients to be mostimpaired in right temporal lesions. (In noneof these patients was clinical prosopagnosiapresent.)

2. Those testing immediate recall of pre-viously unfamiliar faces (DeRenzi and Spinnler,1966; Warrington and James, 1967; Milner,1968). The subject might be shown a facialphotograph which is then removed, and he isthen immediately told to pick out the face from adisplay of several faces.

3. Various matching tasks, in which a facialphotograph is matched with other photographsof the same face (Benton and Van Allen, 1968;Tzavaras et al., 1970). In some tests, matchinghas been with photographs taken from differentangles or in different ambient lighting, or modi-fied by grimaces, false beards, spectacles, etc.While performance in all these tests is im-

paired statistically in unselected groups ofpatients with right hemisphere lesions, they arenot necessarily impaired to the same extent.Thus, Warrington and James (1967) found nocorrelation in their series of patients (none ofwhom was clinically prosopagnosic) between im-paired recognition of well-known faces and im-paired immediate recall of unfamiliar faces. Theyspeculated that prosopagnosia might be alocalized form of the amnesic syndrome (a pointpreviously hinted at by Hecaen et al., 1957). Theamnesic syndrome is characterized by impair-ment of memory (both verbal and visuospatial)with preservation of adequate intellectual func-tions, and results from bilateral damage tocentral core limbic structures (see, for example,Brierley, 1966). Amnesic patients certainly per-form poorly in recognizing well-known faces (asthey do in other tests of memory) and they aretypically unaffected in matching tasks and inimmediate recall of faces.

This notion would hold, therefore, thatprosopagnosia is a disturbance of one aspect ofvisuospatial memory. Some support for thisview comes from Milner's (1968) finding thatdelayed recognition of faces was impaired(though not to the extent seen in prosopagnosia)after right temporal lobectomy, while immediate

recall was not, and from subsequent reports thatpatients with true prosopagnosia may performnormally on matching tasks. This has been so inthree patients so far reported (Tzavaras et al.,1970; Benton and Van Allen, 1972). However afourth patient reported by DeRenzi et al. (1968)performed badly on matching tasks and on bothimmediate and delayed recall, indicating that arestricted form of amnesia cannot account forall cases.There are objections to considering proso-

pagnosia as a restricted form of the amnesicsyndrome. If this were so, one would expectamnesic patients-for example, patients withalcoholic Korsakoff's psychosis-also to beprosopagnosic. In fact, it is very unusual for thedifficulty in recognizing faces that is experiencedby amnesic patients to compare in severity withtrue prosopagnosia. The severely amnesic patientis certainly unable to learn new faces but theauthor is aware of no convincing case where thisdifficulty extended retrogradely to faces learnt inthe remote past, as in prosopagnosia. Theseverest degree of deficit is exemplified by themuch studied subject H.M., who was amnesicafter bimedial temporal lobectomy. This subjecthas for many years since operation had severedifficulty in facial recognition, being unable evento recognize friends and neighbours (first metafter operation) who have been visiting himregularly for years (Milner, 1966). He has evenbeen known to invite total strangers into thehouse to await his mother's return, thinking thatthey must be friends he has not recognized. Bycontrast, the reports indicate that he recognizeshis parents normally. Such preservation of re-mote memory is a regular feature of the amnesicsyndrome which seems to be absent in prosopa-gnosia. Thus, the disturbance ofmemory for facesin prosopagnosia appears to be more severe anddifferent in character from that produced by bi-lateral lesions of central core limbic structures.

It is interesting to note, however, that patientswith prosopagnosia very frequently have disturb-ance of another memory function-topo-graphical memory (Hecaen et al., 1957;Bornstein, 1963; Rondot and Tzavaras, 1969).This is of some relevance, since patients with theamnesic syndrome itself have obvious clinicaldifficulty in two major visual functions-topo-graphical orientation and facial recognition. In

491

J. C. Meadows

fact, topographical orientation (memory) is lesspurely visual than is facial recognition, since ingeneral it involves more verbalizable material.Nevertheless, this relationship provides addi-tional support for those trying to relate proso-pagnosia and amnesia, in spite of the objectionsthat have already been voiced.There is some evidence that topographical

orientation and facial discrimination may be dis-sociable at cortical level: Newcombe andRussell's (1969) study of missile-injured patientsshowed that a maze-learning task was more im-paired with right parietal lesions, while defects offacial discrimination (Mooney's closure test)were most severe in patients with right temporallesions. We can therefore speculate on whetherprosopagnosia might depend upon some lesionintermediate between a posterior area requiredfor facial discrimination and more anterior tem-poral structures (with possible limbic connect-ions) required for learning. Such a lesion mightalso interrupt similar connections concernedwith topographical memory. The anatomicalevidence to be presented is compatible withthis hypothesis.

THE ANATOMICAL EVIDENCE

In the majority of published cases of prosopa-gnosia exact localization of the lesions is notdescribed. Investigations have been scanty andanalysis of the reported clinical details is fre-quently unrewarding from the viewpoint of pre-cise localization. It is clear that in many casesthere is evidence of bilateral disease (shownusually by bilateral field defects), although insome there is evidence only of right hemisphereinvolvement. Whether the overall clinical picturesuggests unilateral or bilateral disease, however,there is one clinical feature which is of goodlocalizing value. A left upper quadrantic visualfield defect is a remarkably consistent finding inreported cases. In a few cases, radiological oroperative findings have been described and in asmall number there have been necropsies. How-ever, to the author's knowledge, there are onlyseven detailed reports of necropsies in theliterature (but see Addendum).

VISUAL FIELD DEFECTS The visual field defects ina large series of patients with prosopagnosia areshown in Fig. 1. These comprise all reports

known to the author which have been publishedduring the last 40 years, together with five casesfrom the records of the National Hospital,Queen Square and Maida Vale. The cases brieflymentioned in Bornstein's review (1963) are not in-cluded because ofinadequate neurological details.

The present cases (1-5) will not be reported in detailin this paper. Case 1 had a large posteriorly placedright hemisphere glioma with radiological evidenceof invasion of the corpus callosum. Cases 2, 3, and 5had each had two or more strokes with historicalevidence of involvement of both hemispheres. Case4, a previously healthy 33 year old man, developedprosopagnosia after evacuation of a right posteriortemporal haematoma, arising from a minute angio-matous malformation. In this case, the prosopa-gnosia was not as severe as in the other cases; hecould, for example, recognize his wife withoutdifficulty, but failed to recognize people that he metonly once or twice a year (such as his cousins andcertain close friends) even though he had knownthem for many years. In all these cases there wasevidence of impaired topographical orientation.

Several points emerge from examination of Fig.1. Visual field defects are nearly always present,and bilateral defects occur in about half thecases. However, in the unilateral cases, thedefect is nearly always on the left. Most notableof all is the frequency of left upper quadrantinvolvement, suggesting a low, posteriorlyplaced lesion of the right cerebral hemisphere.Thus left upper quadrant involvement was notedin all the 19 cases with unilateral left-sideddefects and in all but one of the 15 cases withbilateral defects. There were however four caseswith unilateral right-sided defects and four caseswithout visual field defects. It is worth noting atthis point that two of the four cases with uni-lateral right-sided defects subsequently died andat necropsy had discrete right occipitotemporalpathology in addition to lesions in the left hemi-sphere (Arseni and Botez, 1958; Benson et al.,1974).Of the cases without field defects, two resulted

from carbon monoxide poisoning (Hecaen et al.,1957; Benton and Van Allen, 1972), while athird (Macrae and Trolle, 1956) followed closedhead trauma with coma for three weeks-con-ditions which usually cause bilateral braindamage. Bilateral damage in the case of Macrae

492


Milian (1932) 4 Meadows (1973)-case 1 Cole and Perez Cruet

He'caen et at. (1952)- Medw I93-ae2Gloning et at. (1966)case2 Meadows(1973)-case2 case 2

C Klein and Stack (1953) Meadows (1973)-case 3 Rondot et al. (1967)

C Alajouanine et al. (1953) Meadows (1973)-case 4 Gloning et al. (1970)

C) Faust (1955)-case I Meadows (1973)-case S

5 Faust (1955)-case 2 Holf and Potzl (1937)

t) de Busscher et al. (1956) Donini (1939) Potzl (1953)

j Hecaen et al. (1956)- Bodamer (1947)-case 1 Arseni and Botez (1958)

Hcase eta.(15)

E7j case22 Bodamer (1947)-case 2 Faust (1955)-case 4

Hecaen et at. (1957) Faust (1955)-case 5 h Rubens and Bensoncase 2 KF (1971)

Bornstein and Kidron Fut(95-cs5D (1959) Faust (1955)-case 6

5D case 1et at. (1966)- Faust (1955)-case 7 Faust (1955)-case 3

4>) De Renzi et al. (1968) H =caen et al. (1952)- Macrae and Trollecase 1 (1956)

jt') Bornstein et al. (1969) Pallis (1955) Wm= et at. (1957)case 1

Assal (1969) Beyn and Knyazeva Benton and Van Allen(1962) (1972)

FIG. 1. Visualfield defects in cases ofprosopagnosia published since 1930, including those offive casesfrom therecords of the National Hospital, Queen Square. The defects were homonymous but for convenience are hererepresented by a single field. The cases have been arranged into four groups depending upon whether there wereleft-sided defects alone, bilateral defects, right-sided defects alone, or no defects.

493

J. C. Meadows

and Trolle was supported by the finding oftachistoscopic tunnel vision for depth andcolour. Carbon monoxide poisoning in particu-lar may damage visual association cortex bi-laterally and cause cortical visual impairment(Garland and Pearce, 1967) or visual agnosia(Benson and Greenberg, 1969). The authorknows of an additional case of carbon monoxideinduced prosopagnosia without visual fielddefect in whom less troublesome visual agnosiawas also present (M. Espir, personal com-munication).The fourth published case without visual field

defect (Faust, 1955) had clear evidence of rightoccipitotemporal damage, even though no fielddefect was recorded. In this case, which resultedfrom a missile injury, there was a right occipito-basal bone defect associated with a piece of metal3 cm from the midline close to the petroustemporal bone, with dilatation of the trigone andposterior horn of the right lateral ventricle oncontrast radiography.

NECROPSY LOCALIZATION Pathological studiesconfirm the presence of right occipitotemporalpathology in all cases so far published and inmost cases there was an approximately sym-metrical lesion in the left hemisphere. In allcases there was evidence of bilateral disease(Table). In Wilbrand's case (1892) there wasright occipital softening extending inferiorlyfrom the calcarine fissure, with a corresponding

TABLELOCATION OF LESIONS IN CASES OF PROSOPAGNOSIA

THAT CAME TO NECROPSY

Right hemisphere Left hemisphere

Wilbrand(1892) Inferior occipital Inferior occipitotemporal

Heidenhain(1927) Inferior occipitotemporal Inferior occipitotemporal

Hecaen et al.(1957) Occipito-parietotemporal Invasion through

splenium toventricular wall

Arseni andBotez (1958) Occipito-parietotemporal Occipitotemporal

Pevsner et al.(1962) Inferior occipitotemporal Angular gyrus region

Gloning et al.(1970) Inferior occipitotemporal Inferior occipitotemporal

Benson et al.(1974) Inferior occipitotemporal Inferior occipitotemporal

small area of softening in the left hemisphere.Heidenhain's case (1927) showed infarctioninferiorly, in the region of both occipitotemporaljunctions, involving fusiform and lingual gyri.The case of Hecaen et al. (1957) had a right-sided glioblastoma which at operation wasinvading the occipital lobe, the posterior parts ofthe parietal and temporal lobes, and also theposterior part of the corpus callosum. Eventuallythe patient died and necropsy confirmed thepresence of a glioma which by now was infiltra-ting extensively throughout the right temporaland occipital regions. Involvement was howevermaximal in the medial and inferior aspects of theoccipital and posterior temporal regions, withextension into the corpus callosum as far as theopposite ventricular wall.The case of Arseni and Botez (1958) was of a

bipolar spongioblastoma which at necropsy wasinfiltrating extensively the left occipitotemporalregion. There was an additional deposit, about1-5 cm in diameter, described as involving theregion of the tapetum on the right.The case of Pevsner et al. (1962), reported in

life by Bornstein and Kidron (1959) showed thefollowing findings at necropsy:

'In the middle of the left inferior parietal lobule, inthe angular gyrus and in the upper part of thesuperior temporal sulcus, a yellow-brown, slightlydepressed lesion, 3 cm in diameter was found.Theanterior border of this lesion was formed by theangular gyrus and it extended on to the parieto-occipital sulcus which also formed its posteriorborder. These macroscopic findings at autopsyextended to a depth of 1 mm. On microscopy signsof mild atrophy, gliosis and rarefaction of cells wasseen.... On section of the brain, a narrow cysticcavity, 1 cm in length ... was found in the region ofthe inferior border of the medial surface of the righthemisphere, 3 cm anterior to the occipital pole,under the calcarine fissure which also formed theroof of the lesion. It was 4-6 mm below the cortex.'

The patient of Gloning et al. (1970) sustained aseries of cerebrovascular accidents in the fewmonths before death and at necropsy there wereareas of softening in various parts of the brain.Most striking in relation to his long-standingsymptom of prosopagnosia, however, were old,bilateral ischaemic infarcts, situated almostsymmetrically in the region of the occipito-temporal junction, destroying the central parts

494


of both fusiform gyri. They extended deeply andinvolved the basal parts of the optic radiationsbut spared the striate cortex.

Finally, the patient with associative visualagnosia reported in life by Rubens and Benson(1971) has recently died. This patient had severeprosopagnosia which persisted after ability toname common objects on visual presentation hadbecome almost flawless. Necropsy in this patient(Benson et al., 1974) showed infarction in leftposterior cerebral artery distribution withsplenial destruction and a second, discrete lesioninvolving the right fusiform gyrus. The latterinvolved predominantly the white matter, extend-ing from a point about 2 cm in front of the occi-pital lobe tip posteriorly to a point about 15 cmanterior to the junction of parieto-occipitalsulcus with calcarine sulcus.These are the only well-described necropsy

reports but there are two other cases that cameto necropsy mentioned briefly in the literature,both reported by Bornstein (1963, 1965). Hisfirst case is confusing; in the case report thelesion is described as a right occipital lobemeningioma causing left-sided hemianopia, yetlater in the paper he refers to the 'meningioma inthe left parieto-occipital lobe'. Necropsy is notmentioned in this paper. In his second publica-tion he refers again to this case and also mentionsan additional case. Without giving further detailshe states simply that postmortem examination'revealed a left parieto-occipital meningioma inone case and a glioblastoma in the same regionin the second'. It is difficult to comment oneither of these cases without further details.

RADIOLOGICAL OR OPERATIVE LOCALIZATIONThere are a few cases in which approximatelocalization can be deduced from radiologicaland operative findings. In these cases the findingsare in keeping with those already described.Hecaen and Angelergues (1962) mention threecases with difficulty in facial recognition aftersurgery to the posterior part of the right cerebralhemisphere: removal of right parieto-occipito-temporal region and right occipitotemporalregion in two cases with intractable epilepsy, andremoval of a large right parieto-occipitalmeningioma in the third. The case of Pallis (1955)showed defective filling of the right posteriorcerebral artery on angiography but there werebilateral visual field defects indicating bilateralinvolvement. Assal (1969) described a patientwith a proven right parietotemporal intracerebralhaemorrhage, and haemorrhage in the rightposterior temporal region was responsible in thepresent case 4. Finally, the present case 1 had alarge deeply placed glioma in the posterior partof the right hemisphere. Its localization is mosteasily seen from the brain scan (Fig. 2). Carotidand vertebral angiograms indicated a largemalignant tumour lying deep in the right, pos-terior temporoparietal region and involving thesplenium of the corpus callosum. At operation, alarge tumour was encountered at a depth of 2-5cm from the surface convexity of the brain. Fur-ther exploration showed that it had come throughthe medial surface of the hemisphere and wasextending under the free edge of the falx in theregion of the vein of Galen. It also occupied thebody of the right lateral ventricle, bulging into it

FIG. 2. Technetium brainscan of the present case 2,(a) anteroposterior and (b)right lateral views.

( a (b)

77-7

77

495

f hn)

J. C. Meadows

from its medial side. An incomplete removalwas performed. The patient was not followedup.

DISCUSSION

LOCATION OF LESIONS The evidence that hasbeen presented points clearly to the importanceof occipitotemporal lesions in prosopagnosia.The evidence for right-sided involvement is verystrong: lesions in this region were found in allthe seven cases that came to necropsy so farreported, and this correlates with the extremelyhigh incidence of left upper quadrantic visualfield defects in clinical case reports. On the otherhand, lesions at this site in the left hemispherewere found in only five of the seven necropsies,and right upper quadrantic field defects are notparticularly common clinically.The lesions in the right hemispheres involved

fusiform or lingual gyri in six of the seven cases.Immediately deep to these gyri lies the opticradiation, and this structure was obviouslyinvolved in most if not all these cases (a clearstatement is not made in every case). In theseventh case (Arseni and Botez, 1958) the lesionwas more dorsally placed, and was described asinvolving the tapetum (callosal fibres sweepingdown into the occipito-temporal region).

Left hemisphere lesions were present in allseven necropsies but only in five were theseoccipitotemporal as in the right hemisphere(Table). In the case of Pevzner et al. (1962) thatcame to necropsy the only left-sided lesiondescribed was mild gliosis in the angular gyrusregion. In the case of Hecaen et al. (1957) therewas invasion by tumour from the right hemi-sphere through the posterior part of the corpuscallosum, but only as far as the ventricular wall.As already stated, the purely clinical data alsofail to support the consistent presence of sym-metrical lesions, for right upper quadrantic fielddefects are not a regular finding in published casereports; in about half the cases there is no rightfield defect at all and in some of the caseswhere a right-sided defect is present it spares theright upper quadrant (Fig. 1). We may concluderather cautiously at this point that bilateraloccipitotemporal lesions may underlie manycases of prosopagnosia but that in some (andpossibly many) cases, the right occipitotemporal

lesion is not accompanied by a correspondinglysited lesion in the left hemisphere.What is the mechanism of the disturbance

produced by these occipitotemporal lesions?Their gross location has been discussed, but thepossible pathways affected have not yet beenconsidered. For the present, we will leave asidethe question of whether the lesion at this site isunilateral or bilateral and just consider thepossible structures involved.The lesions that have been described affect

mainly inferior occipitotemporal cortex (lingualand fusiform gyri) and do not usually extend for-ward far into the temporal lobe. They also in-volve the white matter deep to these gyri, com-prising incoming and outgoing fibres to thiscortex (including tapetal fibres descending fromthe posterior part of corpus callosum), inferiorlongitudinal fasciculus (ILF) and lower fibres ofoptic radiation.

In fact, the lower part of the optic radiationfuses and is intermingled with ILF, which is themain pathway from the classical visual associa-tion cortex (areas 18 and 19) into the temporallobe. These two structures are together known asthe external sagittal stratum. A lesion of thelower part of the right optic radiation in theoccipitotemporal region (accounting for the leftupper quadrantic field defect) will thus almostcertainly involve ILF and is likely in addition tointerrupt callosal (tapetal) connections betweeninferiorly placed occipitotemporal cortex on thetwo sides. Even the more dorsally placed lesionin the case of Arseni and Botez (1958) probablyinvolved the external sagittal stratum, for thisstructure is closely applied to the tapetum, uponwhich the tumour was centred, just lateral to theventricle.

These basal occipitotemporal lesions may thusinterrupt the main visual input to the temporallobe in addition to any effects arising fromdamage to the occipitotemporal cortex itself. Infact, some fibres of ILF probably interconnectvarious parts of inferior occipital and temporalcortex, and ILF may even form a chain ofconnections concerned with visual function, ex-tending from the occipital into the temporallobe (see below). Damage to ILF in thisregion may thus have some effects similar tolesions of occipitotemporal cortex itself, by in-volving the latter's incoming and outgoing con-

496


nections. Damage to the long fibres in ILF isprobably not important, for prosopagnosia doesnot appear to be associated with lesions more

anteriorly in the temporal lobe, further along thecourse of these same fibres.

COMPARISON WITH ANIMAL EXPERIMENTS Some

insight into the possible neuropsychologicalmechanism of lesions at this site may be gainedby considering the results of experiments inanimals. Prosopagnosia associated with bilateraloccipitotemporal lesions might be comparedwith the defect of visual discrimination learning(with sparing of both primary visuosensory func-tion and learning in other modalities) which fol-lows bilateral ablations of inferotemporal cortexin the macaque (see review by Gross, 1974). Thisanalogy may be relevant even to cases with in-volvement of only the right occipitotemporalregion, for this is the side of the brain which, inman, is most concerned with visuospatial func-tion and, more specifically to the present con-

text, with facial discrimination and recall (see thestatistical studies on unselected series of braindamaged patients already discussed).A simple explanation of prosopagnosia can

thus be suggested based on the occipitotemporallocation of the lesions and the fact that similarlyplaced lesions in monkeys impair visual dis-crimination learning. In man, facial recognitionis so much more complicated and more fre-quently undertaken than any other task involv-ing complex visual discrimination learning thatpart of the inferior occipitotemporal cortexappears to become functionally modified for thistask. It is suggested that destruction of thiscortex or its connections underlies prosopa-gnosia. Connections from this region to more

anterior temporal cortex and thence to mediallimbic structures are apparently still necessaryfor learning new faces, as has already been dis-cussed in relation to the amnesic syndrome.

THE PROBLEM OF THE LEFT HEMISPHERE LESION So

far we have avoided discussing the difficult prob-lem of the variable location of the left hemispherelesions and the question that follows from it-whether bilateral lesions are essential or not forthe development of prosopagnosia. It has alreadybeen noted that bilateral lesions have been presentin all cases that came to necropsy but that the

lesions in the left hemispheres have not all beenoccipitotemporal in location. Similarly, the evi-dence from clinical studies, in particular thevisual field data, leads one to question the role ofthe left hemisphere (at the very least, it must bedifferent from the role of the right hemisphere).This, of course, accords with the statisticalstudies in unselected brain damaged patientswhich show that the right hemisphere is of muchgreater importance in facial discrimination andrecall. It is natural to question whether the lefthemisphere lesion is necessary at all. The evi-dence, however, is overwhelming that bilaterallesions are important in the overall aetiology ofprosopagnosia. The high incidence of bilateraldisease clinically and the invariable presence ofbilateral lesions in the few necropsies so far per-formed can scarcely have occurred by chance.Moreover, bilateral lesions conveniently explainthe rarity of the condition. However, this stilldoes not explain the inconsistent location of theleft hemisphere lesion. It is not unreasonable tosuggest, for example, that the left occipito-temporal lesions are important (being locatedroughly symmetrically to those in the righthemisphere) but that in the two cases that cameto necropsy where the left hemisphere lesionswere not occipitotemporal (Hecaen et al., 1957;Pevzner et al., 1962) they might be incidental andirrelevant.Some alternative possibilities might be con-

sidered to account for some of the anomaliesthat have been discussed-in particular theclinical evidence implicating the right hemi-sphere more than the left. Some of those patientswith clinical evidence of only unilateral diseasemight indeed have only unilateral lesions and berare individuals in whom visual function, or atleast the function of facial recognition, is unusu-ally lateralized in the right hemisphere, whereasin most individuals bilateral lesions are requiredto cause prosopagnosia. Thus, some cases mighthave unilateral (right) occipitotemporal lesions,as in the case of Pevzner et al. (1962), whileother cases have bilateral occipitotemporallesions.Another possibility is that the combination of

lesions in bilateral cases has the same effect, sofar as physiological mechanism is concerned, as asingle strategically placed right-sided lesion. Forexample, recognition of familiar faces might

497

J. C. Meadows

depend on connections from both occipital cor-

tices to the right occipitotemporal cortex. Alesion causing destruction of the right primaryvisual cortex in combination with destruction atsome point along the pathway from the leftstriate cortex to the right occipitotemporalregion (probably via the left visual associationcortex, including the left occipitotemporal region,and callosal fibres passing in the splenium)might then have the same effect as a single rightoccipitotemporal lesion. Damage to the callosalfibres might be in the splenium itself, or morelaterally placed in either right or left hemisphere.It will be recalled that there was gliomatousinfiltration of the corpus callosum from the righthemisphere in both the case of Hecaen et al.(1957) that came to necropsy and case 1 of thepresent author. However, this explanation failsto account for the rarity of prosopagnosia,which would, for example, be expected to occur

regularly with right posterior cerebral arteryocclusion.Could the last explanation apply if an isolated

right hemisphere lesion were particularly dis-crete in location ? Could, for example, theneuronal networks in the chain of connectionsbeyond a particularly discrete right occipito-temporal lesion still have a decisive effect on thebrain's response? In other words, might thesenetworks actively 'inform the brain' that a

familiar face is unfamiliar simply because they donot receive the information necessary to identifyit? It is known that the left hemisphere can

identify faces when disconnected from the righthemisphere (Levy et al., 1972) and this is pre-

sumably how facial recognition is achieved, forexample, when right occipital and basal temporallobe is destroyed by right posterior cerebralartery occlusion. But, in a particularly discreteright occipitotemporal lesion of the type en-

visaged, is it possible that the 'negative' informa-tion from the right hemisphere in some manner

suppresses the 'positive' information from theleft hemisphere and leads to an actual denial ofrecognition ? With a larger right hemispherelesion, the right hemisphere mechanism wouldbe more deranged and identification by the lefthemisphere would take place.

This is an interesting speculation but not one

which can at the present time be supported; as

already discussed, there is so far no firm evi-

dence that a unilateral lesion can cause proso-pagnosia. It is to be hoped that further necropsystudies will explain some of the paradoxes con-cerning lesion location, at least at the grossanatomical level.

VARIETIES OF PROSOPAGNOSIA It has already beenpointed out that discrimination of faces maybe normal in prosopagnosia and that, at leastin some cases, the disorder can be consideredto be one of memory for faces. However, itis unlikely that a disturbance of visual memoryunderlies all cases of prosopagnosia. Thus thepatient of DeRenzi et al. (1968) was unableeven to discriminate faces, judging from hisperformance on matching tasks and tests ofimmediate recall. This is not too surprising, forfacial recognition is probably not a single-stepprocess but is likely to involve a series of stages.In fact, different types of abnormality can be pre-dicted. Discrimination of faces might be im-paired and the subject be unable to say whethertwo identical facial photographs are the same ornot. At a more complex level, he might succeedin this task but be unable to say whether twofaces are the same when photographed fromdifferent angles or in different ambient lighting.In turn, these processes of perceptual integrationmight be intact but the patient be unable to drawon memory function required for recognition.The last-mentioned seems to be the commonesttype of defect.The present study suggests that the anatomical

substrate for these functions includes the regionof the occipitotemporal junction, particularly onthe right. Gross et al. (1972) have provided someevidence in the monkey that the sort of hierarchi-cal perceptual processes demonstrated by Hubeland Wiesel (1962, 1965, 1968) in the geniculo-striate system of the cat and monkey may beextended into the inferotemporal cortex formore complex material, and it is reasonable toconsider whether a similar mechanism mightoccur with complex perceptual processes, likefacial recognition, in man. The known lateralizat-ion of visuospatial function in man is presumablyrelevant to the apparent greater importance ofthe right hemisphere but as already discussedcertain questions remain unanswered in relationto this.Such a hierarchical arrangement could under-

498


lie clinical differences between reported cases.

Against this background, it is easier to under-stand cases such as that of DeRenzi et al. (1968)with impaired facial discrimination, and cases

associated with metamorphopsia (altered or dis-torted vision), as discussed by Hecaen andAngelergues (1962), where there is a more pri-mary disturbance of perception. The cases ofBodamer (1947) fall into the latter category andhis first patient, for example, complained thatfaces appeared 'strangely flat; white with darkeyes, as if in one plane, like white oval plates ...

all the same' (see Critchley, 1953). Metamor-phopsia may also be very pronounced withoutactual prosopagnosia, as in Bodamer's thirdpatient who really had a specific metamorphop-sia involving faces, rather than prosopagnosia,since the patient could still recognize faces.

Finally, it must be remembered that difficultyin facial recognition may occur as part of lessspecific neurological disturbances. The questionof facial recognition in amnesic subjects hasalready been considered and it has been pointedout that the defect in recognition here does notappear to extend retrogradely to faces learned inthe remote past, but only as far retrogradely as

the amnesia itself extends; the difficulty is inlearning new faces during the period of retro-grade and anterograde amnesia.

Facial recognition may also be impaired inpatients with visual agnosias (for example,Benson and Greenberg, 1969; Taylor andWarrington, 1971). In these conditions objectscannot be identified when presented visuallyand it is therefore not too surprising that facesmay not be identified either. In 'apperceptive'visual agnosia there is bilateral disturbance ofvisual association cortex. On the other hand' associative' visual agnosia has been inter-preted as indicating disconnexion between thevisual areas and the posterior speech area

(Geschwind, 1965) and can result from lesionsconfined to the left hemisphere. It is theo-retically possible that a lesion confined to theleft hemisphere might similarly cause isolatedprosopagnosia. This would depend upon whetherfibres transmitting information about faces aresegregated sufficiently from other visual input tothe speech area. Of the four cases of prosopa-gnosia shown in Fig. 1, where the clinical datasuggested confinement of pathology to the left

hemisphere, two came to necropsy and in bothcases bilateral occipitotemporal lesions werefound (see previous section); rather surprisinglythe two right hemisphere lesions did not causedetectable left visual field defects. Thus there isno evidence at the present time that prosopagno-sia can result from an isolated left hemispherelesion.

In conclusion, however, it is worth consideringthe possible combined effects of partial interrup-tion of visual input to speech area together with aright occipitotemporal lesion, for such an ex-planation could explain the problematical case ofPevzner et al. (1962), that came to necropsywhere the only reported lesion in the left hemi-sphere was mild gliosis in the angular gyrusregion.

SUMMARY

Patients with prosopagnosia nearly always havea left upper quadrantic visual field defect,correlated clinically and pathologically with aright occipitotemporal lesion. The few necropsiesall show bilateral lesions but the right hemispherelesions have all involved the region of theoccipitotemporal junction.The lower part of the optic radiation fuses

with the inferior longitudinal fasciculus (ILF)and these two structures lie immediately dorsalto inferior occipitotemporal cortex. ILF is themain direct pathway by which visual informationreaches temporal lobe (and thence limbic struc-tures), but along its path its fibres also inter-connect the different areas of inferior occipito-temporal and posterior temporal cortex. Aslesions far forward in ILF are apparently notassociated with prosopagnosia, involvement ofinferior occipitotemporal cortex or its incomingand outgoing connections is probably the im-portant factor.

All cases of prosopagnosia that have come tonecropsy have also shown lesions in the lefthemisphere, mostly occipitotemporal, so thatapproximately symmetrical lesions were present.But in two out ofseven necropsies no left occipito-temporal lesion was described. Right-sided visualfield defects have been present in only about halfthe recorded cases. The problem of the inconsis-tent location of the left hemisphere lesion hasnot been resolved but various possibilities havebeen considered.

499

J. C. Meadows

Bilateral lesions of inferior temporal cortex inmonkeys impair visual discrimination learning.Facial recognition can in many respects beregarded as an extremely complex learning taskin visual discrimination.

ADDENDUM

Since this paper was written, three further cases havebeen drawn to the author's attention, in one ofwhich necropsy findings were published. A case isreported in Critchley's (1965) paper on anomalies ofcolour vision of cerebral origin; this patient had asuperior altitudinal hemianopia with prosopagnosia,loss of topographical sense, and impaired colourvision with xanthopsia. Lhermitte et al. (1972) reporta case that came to necropsy who, in addition todifficulty with faces was also unable to recognizefamiliar cars, a feature shown to some extent also bythe present case 5 and by the case of Macrae andTrolle (1956). At necropsy, Lhermitte's case showedbilateral lesions of fusiform gyrus with, on the right,involvement also of the adjacent lingual gyrus.Finally, Tzavaras, Merienne, and Masure (personalcommunication) describe a left-handed patient with aleft temporal lesion who presented with proso-pagnosia, amnesia, and language difficulty.

My thanks are due to Dr. D. Frank Benson for hiscomments on this manuscript.

REFERENCES

Alajouanine, T., Lhermitte, F., and Sabouraud, 0. (1953)Agnosie visuelle sans alexie. Revue Neurologique, 89, 158.

Arseni, C., and Botez, M. I. (1958). Consideraciones sobre uncaso de agnosia de las fisonomias. Revista Neuro-psiquiatria, 21, 583-593.

Assal, G. (1969). Regression des troubles de la reconnaissancedes physionomies et de la memoire topographique chez unmalade opere d'un hematome intracerebral parieto-temporal droit. Revue Neurologique, 121, 184-185.

Benson, D. F., and Greenberg, J. P. (1969). Visual formagnosia. Archives of Neurology (Chic.), 20, 82-89.

Benson, D. F., Segarra, J. M., and Albert, M. L. (1974).Archives of Neurology. (In press.)

Benton, A. L., and Van Allen, M. W. (1968). Impairment infacial recognition in patients with cerebral disease. Cortex,4, 344-358.

Benton, A. L., and Van Allen, M. W. (1972). Prosopagnosiaand facial discrimination. Journal of Neurological Sciences,15, 167-172.

Beyn, E. S., and Knyazeva, G. R. (1962). The problem ofprosopagnosia. Journal of Neurology, Neurosurgery, andPsychiatry, 25, 154-158.

Bodamer, J. (1947). Die Prosop-Agnosie. Archiv fur Psy-chiatrie und Nervenkrankheiten, 179, 6-53.

Bornstein, B. (1963). Prosopagnosia. In Problems of DynamicNeurology, pp. 283-318. Edited by L. Halpern. HadassahMedical Organization: Jerusalem.

Bornstein, B. (1965). Prosopagnosia. In 8th InternationalCongress of Neurology, Vienna, 1965. Proceedings, Vol. 3,pp. 157-160. Wiener Medizinischen Akademie: Vienna.

Bornstein, B., and Kidron, D. P. (1959). Prosopagnosia.Journal of Nelurology, Neutrosurgery, and Psychiatry, 22,124-131.

Bornstein, B., Sroka, H., and Munitz, H. (1969). Proso-pagnosia with animal face agnosia. Cortex, 5, 164-169.

Brierley, J. B. (1966). The neuropathology of amnesic states.In Amnesia, pp. 150-180. Edited by C. W. M. Whitty and0. L. Zangwill. Butterworths: London.

Cole, M., and Perez-Cruet, J. (1964). Prosopagnosia. Neulro-psychologia, 2, 237-246.

Critchley, M. (1953). The Parietal Lobes. Arnold: London.Critchley, M. (1965). Acquired anomalies of colour percep-

tion of central origin. Brain, 88, 711-724.De Busscher, J., Hoffman, G., and Kluyskens, J. (1956).Agnosia visuelle temporaire pour les personnes et optico-spatiale pour les objets a la suite d'un ictus unique. ActaNeurologica Belgica, 56, 162-176.

De Renzi, E., and Spinnler, H. (1966). Facial recognition inbrain-damaged patients. An experimental approach.Neurology (Minneap.), 16, 145-152.

De Renzi, E., Faglioni, P., and Spinnler, H. (1968). Theperformance of patients with unilateral brain damage onface recognition tasks. Cortex, 4, 17-34.

Donini, F. (1939). Su di un caso di aprasia costruttiva congrave disorientamento esospaziale e perdita della facoltadel viconoscimento della fisonomia della personne. NotePsichiatre, 68, 469-485.

Faust, C. (1955). Die zerebralen Herdstorungen bei Hinter-hauptsverletzungen und ihre Beurteilung. Thieme: Stuttgart.

Garland, H., and Pearce, J. (1967). Neurological complica-tions of carbon monoxide poisoning. Quarterly Journal ofMedicine, 36, 445-455.

Geschwind, N. (1965). Disconnexion syndromes in animalsand man. Part 2. Brain, 88, 585-644.

Gloning, I., Gloning, K., Hoff, H., and Tschabitscher, H.(1966). Zur prosopagnosie. Neuropsychologia, 4, 113-131.

Gloning, I., Gloning, K., Jellinger, K., and Quatember, R.(1970). A case of 'prosopagnosia' with necropsy findings.Neuropsychologia, 8, 199-204.

Gross, C. G. (1974). In press, quoted by Gross et al., 1972.Gross, C. G., Rocha-Miranda, C. E., and Bender, D. B.

(1972). Visual properties of neurons in inferotemporalcortex of the macaque. Journal ofNeurophysiology, 35, 96-111.

Hecaen, H., and Angelergues, R. (1962). Agnosia for faces(prosopagnosia). Archives of Neurology (Chlic.), 7, 92-100.

Hecaen, H., Ajuriaguerra, J. de, Magis, C., and Angelergues,R. (1952). Le probl&me de l'agnosie des physionomies.Ence~phale, 41, 322-355.

Hecaen, H., Angelergues, R., Bernhardt, C., and Chiarelli, J.(1957). Essai de distinction des modalites cliniques del'agnosie des physionomies. Reviue Neurologique, 96, 125-144.

Hecaen, H., Penfield, W., Bertrand, W., and Malmo, R.(1956). The syndrome of apractognosia due to lesions ofthe minor cerebral hemisphere. Archives of Neurology(Chic.), 75, 400-434.

Heidenhain, A. (1927). Beitrag zur Kenntnis der Seelen-blindheit. Monatsschriftfur Psychiatrie und Neuirologie, 66,61-116.

Hoff, H., and Potzl, 0. (1937). Ueber eine optisch-agnostischeStorung des 'Physiognomie-Gedachtnisses' (Bezeihungenzur Ruckbildung eine Wortblindheit). Zeitschrift fur diegesamte Neurologie und Psychiatrie, 159, 367-395.

Hubel, D. H., and Wiesel, T. N. (1962). Receptive fields,binocular interaction and functional architecture in thecat's visual cortex. Journal of Physiology, 160, 106-154.

500


Hubel, D. H., and Wiesel, T. N. (1965). Receptive fields andfunctional architecture in two nonstriate visual areas (18and 19) of the cat. Journal ofNeurophysiology, 28, 229-289.

Hubel, D. H., and Wiesel, T. N. (1968). Receptive fields andfunctional architecture of monkey striate cortex. Journal ofPhysiology, 195, 215-243.

Klein, R., and Stack, J. J. (1953). Visual agnosia andalternating dominance; analysis of a case. Journal ofMentalScience, 99, 749-762.

Levy, J., Trevarthen, C., and Sperry, R. W. (1972). Percep-tion of bilateral chimeric figures following hemisphericdeconnexion. Brain, 95, 61-78.

Lhermitte, F., Chain, F., Escourolle, R., Ducarne, B., andPillon, B. (1972). Etude anatomoclinique d'un cas deprosopagnosie. Revue Neurologique, 126, 329-346.

Macrae, D., and Trolle, E. (1956). The defect of function invisual agnosia. Brain, 79, 94-110.

Milian, G. (1932). Cecite morphologique. Bulletin del'Academie de Medecine, 107, 664-666.

Milner, B. (1966). Amnesia following operation on thetemporal lobes. In Amnesia, pp. 109-133. Edited byC. W. M. Whitty and 0. L. Zangwill. Butterworths:London.

Milner, B. (1968). Visual recognition and recall after righttemporal-lobe excision in man. Neuropsychologia, 6, 191-209.

Newcombe, F., and Russell, W. R. (1969). Dissociated visualperceptual and spatial deficits in focal lesions of the righthemisphere. Journal of Neurology, Neurosurgery, andPsychiatry, 32, 73-81.

1'allis, C. A. (1955). Impaired identification of faces andplaces with agnosia for colours. Journal of Neurology,Neurosurgery, and Psychiatry, 18, 218-224.

I'evzner, S., Bornstein, B., and Loewenthal, M. (1962).Prosopagnosia. Journal of Neurology, Neurosurgery, andPsychiatry, 25, 336-338.

Potzl, 0. (1953). Zur Agnosie des Physiognomiegedachtnisses.JWiener Zeitschrift fir Nervenheilkunde, 6, 335-354.

Rondot, P. Tzavaras, A., and Garcia, R. (1967). Sur un casde prosopagnosie persistant depuis quinze ans. ReviueNeurologique, 117, 424-428.

Rondot, P., and Tzavaras, A. (1969). La prosopagnosie apresvingt annees d'etudes cliniques et neuropsychologiques.Journal de Psychologie Normale et Pathologiqie, 66, 133-166.

Rubens, A. B., and Benson, D. F. (1971). Associative visualagnosia. Archives of Neurology (Chic.), 24, 305-316.

Taylor, A., and Warrington, E. K. (1971). Visual agnosia: asingle case report. Cortex, 7, 152-161.

Tzavaras, A., Hecaen, H., and Le Bras, H. (1970). Le prob-leme de la specificite du deficit de la reconnaissance duvisage humain lors des lesions hemispheriques unilaterales.Neuiropsychologia, 8, 403-416.

Warrington, E. K., and James, M. (1967). An experimentalinvestigation of facial recognition in patients with uni-lateral cerebral lesions. Cortex, 3, 317-326.

Wilbrand, H. (1892). Ein Fall von Seelenblindheit undHemianopsie mit Sectionsbefund. Deutsche Zeitschrift fiirNervenheilkunde, 2, 361-387.

Yin, R. K. (1970). Face recognition in brain injured patientsa dissociable ability? Neuropsychologia, 8, 395-402.

501

the anatomical basis of prosopagnosia

Documents