Impaired face and word recognition without object agnosiafn2

Impaired face and word recognition without object agnosiafn2

\ PERGAMON Neuropsychologia 26 "0888# 30Ð49 Impaired face and word recognition without object agnosia0 Laurel J[ Buxbauma\b\\ Guila Glosserc\ H[ Br...

251KB Sizes 0 Downloads 116 Views

\ PERGAMON

Neuropsychologia 26 "0888# 30Ð49

Impaired face and word recognition without object agnosia0 Laurel J[ Buxbauma\b\\ Guila Glosserc\ H[ Branch Cosletta\b a

Moss Rehabilitation Research Institute\ Philadelphia\ PA 08030\ USA b Temple University Hospital\ Philadelphia\ PA\ USA c University of Pennsylvania Medical Center\ Philadelphia\ PA\ USA Received 1 December 0886^ accepted 4 May 0887

Abstract A leading account of high!level visual recognition proposes that the recognition of faces\ objects\ and words is mediated by two processing capacities[ Words are assumed to require the capacity to represent numerous parts\ whereas faces are processed whol! istically\ and hence require the representation of complex units[ Object recognition requires the capacity to represent both numerous and complex parts[ As object recognition depends upon the same processing capacities underlying face and word recognition\ this account predicts that patients with severe alexia and prosopagnosia should be de_cient in tests of object recognition[ We report a patient who is unable to recognize words or faces\ yet performs relatively well on tests of object recognition[ The two!capacity theory cannot accommodate this pattern of performance without additional assumptions[ Þ 0887 Elsevier Science Ltd[ All rights reserved[

0[ Introduction Visual agnosia is a modality!speci_c de_cit in rec! ognition which is not attributable to impairments in {low level| visual processes or semantic memory[ Patients with visual agnosia are frequently de_cient in face\ common object\ and word processing to unequal degrees[ Some agnosics\ for example\ are impaired in recognizing faces\ but not other classes of visually!similar objects^ that is\ they exhibit isolated prosopagnosia[ Other patients exhi! bit prosopagnosia as well as associated de_cits in reco! gnizing objects ð6\ 04Ł[ There are numerous reports of patients with isolated de_cits in reading[ Finally\ there are many agnosics who are de_cient with words and objects\ but not faces[ The fact that face\ common object\ and word pro! cessing capacities are neither invariably impaired tog! ether nor spared together suggests that the processes underlying face\ object\ and word recognition di}er in important ways[ Although the speci_cs vary\ many accounts of visual agnosia propose that the disorder results from some form of high level visual impairment1 which a}ects the capacities required to form or access  Corresponding author[ Tel[] 990 104 345 8486^ fax] 990 104 345 8403^ e!mail] LBuxbaumÝaehn1[einstein[edu 0 Supported by NIH Grant R18DC92068!90 to the _rst author[ 1 While we recognize that the distinction between {apperceptive| and {associative| agnosia is blurred\ we refer here to agnosias usually charac! terized as {associative|[

structural descriptions ð7Ð00\ 19Ł\ disrupts the structural descriptions themselves ð24Ł\ or prevents transmission of information from structural descriptions to semantics ð16\ 21Ł[ Given the dissociability of face\ object\ and word impairments in agnosia\ it seems plausible that each of these classes of stimuli are recognized by means of a distinct module or procedure in one or more of these processing stages[ For example\ one might postulate three di}erent structural description systems\ each underlying recognition of a major class of visual stimuli[ On the basis of a recent analysis of 88 cases of visual associative agnosia\ Farah ð8Ð00Ł has questioned the assumption of class!speci_c processes[ She noted that there were reports of patients with isolated impairments in face recognition and others with de_cits in face and object recognition[ There were also patients with de_cient object and word recognition[2 Finally\ there were patients with impaired recognition of all three stimulus classes[ In contrast\ there was a dearth of clearly described patients with isolated de_cits in object recognition\ or with intact object recognition in the context of impaired word and face processing[ Thus\ common object recognition always appeared to {pattern| with word or face processing[ Farah suggested that the observed patterns would be predicted if there are just two types of shape rep! resentation capacity underlying the formation of struc!

2 Patients with isolated word recognition impairments "acquired alexia# are quite common\ and were not included in the search[

9917Ð2821:87 ,*see front matter Þ 0887 Elsevier Science Ltd[ All rights reserved PII] S 9 9 1 7 Ð 2 8 2 1 " 8 7 # 9 9 9 3 7 Ð 6

31

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

tural descriptions of all classes and types of objects ð09Ł "also see ð17Ł#[ She proposed that one system generates structural descriptions that are highly decomposed into features or parts\ such that the parts are explicitly rep! resented as structures in their own right[ These structural descriptions require the capacity to represent numerous parts[ A second system generates structural descriptions for objects that undergo little or no shape decomposition[ This type of structural description requires the capacity to represent complex parts[ These two main types of representational capacity are posited to underlie processing of words\ objects\ and faces[ On this account\ faces are processed as non!decom! posed perceptual wholes\ and hence face recognition requires the integrity of a structural description system able to represent complex parts[ "In the case of faces\ the whole object may be a single {part|#[ Words\ in contrast\ undergo extensive decomposition\ and require a system capable of representing numerous parts[ Finally\ com! mon object recognition requires both of these capacities\ but taxes neither of them as much as face and word recognition[ Thus\ an impairment in processing per! ceptual wholes will cause prosopagnosia but not alexia^ conversely\ a de_cit in processing multiple parts will result in alexia but not prosopagnosia[ We will refer to this as the two!process account[ Two predictions emerge from this account[ First\ pat! ients with intact wholistic and part!based processing should be unimpaired with all three classes of stimuli[ Thus\ a patient with intact face and word processing should never be de_cient in recognizing common objects[ Second\ severe de_cits in both wholistic and part!based processing should cause de_ciencies in recognizing all object classes^ thus\ patients with severe alexia and pro! sopagnosia should always be substantially impaired in common object recognition[2 These predictions are\ in fact\ consistent with the patterns observed in the 88 cases reviewed ð8\ 09Ł] there were no patients who clearly exhi! bited isolated object agnosia\ nor were there clear cases of isolated sparing of object recognition[ Recently\ however\ Rumiati\ Humpheys\ Riddoch and Bateman ð23Ł reported a case study which presents an apparent challenge to Farah|s _rst prediction[ The patient\ Mr W\ was selectively impaired in recognizing objects and pictures of objects but performed normally on tests of word and face recognition[ Thus\ Mr W[ is an agnosic without either prosopagnosia or alexia[ Inves! tigation revealed impaired access to semantic knowledge of objects from pictures[ These was also evidence of imp! aired visual knowledge on such tests as {object decision|\ which require judgment of whether a depicted object is real or arti_cial[ Rumiati et al[ ð23Ł suggested that Mr 2 Farah ð09Ł has noted that very mild impairments in both types of structural description might result in mild prosopagnosia and alexia without observable object agnosia[

W|s performance was consistent with a de_cit at the level of the structural description system[ Interpretation of Mr W|s case is clouded by the fact that he was also de_cient in accessing semantic knowledge about objects and words from auditory input\ suggesting that semantic loss may have contributed to the overall pattern of his de_cit with visual input[ Citing the fact that Mr W performed nor! mally on at least one object decision task\ Farah ð7Ł has suggested that Mr W|s de_cits with objects arose from a processing state beyond that at which the structural description is contacted[ On this account\ data from Mr W do not seriously challenge her hypothesis[ Humphreys and Rumiati ð11Ł have responded by reporting a second patient with isolated object agnosia[ MH had impaired visual access to stored knowledge of objects\ and made predominantly visual naming errors[ Her performance on tests assessing semantic memory access from words was normal[ She was severely impaired in object decision\ suggesting a de_cit at the level of the structural description[ In contrast\ she performed nor! mally in reading and face recognition tasks[ Thus\ MH appears to present a convincing challenge to Farah|s _rst prediction[ To this point\ however\ there have been no reported cases relevant to the second prediction derived from the two!process account\ namely\ that one should never see a patient with relative sparing of object recognition in the context of severe alexia and prosopagnosia[ We now present data from just such a patient[ WB exhibits severe\ debilitating de_cits in reading and face recognition\ and relatively milder de_cits in recognizing some objects^ in fact\ object recognition for several object categories is perfect[ We _rst document WB|s performance with objects\ faces\ and words\ and then provide evidence that his object processing\ though not unimpaired\ is sub! stantially better than his face processing on a test that directly pits one against the other[ We will argue that the two!process account can not accommodate these data without additional assumptions[ 0[0[ Subject WB was a 36 year!old\ right!handed man with a history of focal motor seizures involving the left upper extremity\ complex partial seizures\ and secondarily generalized tonic!clonic convulsions beginning at age 09[ The seizure disorder was attributed to an accident at age 6 in which he was struck on the right side of the head with a golf club[ WB describes himself as having been an {average| student who enjoyed reading novels by Hemingway and Salinger early in high!school[ WB|s Seizures increased in duration and intensity during his teen years[ At age 06 he underwent right frontal lobectomy and at age 07 right temporal lobectomy for seizure relief[ He continued his studies with a home tutor\ and graduated from high! school at age 19[ At age 11 he underwent right cry!

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

32

opulvinectomy[ The next year seizure control was attempted by electrical stimulation with an electrode grid placed over the right paleocerebellum[ Ten years later\ in 0870\ he su}ered a large bilateral posterior intra! parenchymal hemorrhage during surgery to remove the electrode grid\ resulting in transient cortical blindness[ Neurologic examination at the time of the present investigations revealed a spastic hemiparesis and hyper! re~exia on the left\ and a left central facial weakness[ Pupils were equal and reactive to light and accom! modation[ Ocular motility was full with nystagmus on right lateral gaze[ Visual _elds were full[ WB resides in an independent!living complex\ where he is assisted by a personal aide[ He continues to experience multiple focal motor seizures daily[ He is wheelchair! bound[ In everyday life\ he denies di.culties with object recognition[ However\ he notes that he has been unable to read since 0870 because he {can|t recognize letters or words|[ He also acknowledges severe\ debilitating prob! lems with face recognition]

mal[ On the line bisection and line cancellation subtasks of the Rivermead Behavioral Inattention Test ð18Ł\ there was mild left neglect[ For example\ on a subtask requiring bisection of three 19[4 cm lines\ WB erred an average of 2[1 cm to the right^ on a line cancellation subtask\ he failed to cancel 3:19 lines\ all in the lower left quadrant of the page[ There was no evidence of neglect on _gure! copying tasks[ On the {cookie theft| picture from the Boston Diagnostic Aphasia Examination "in which a boy attempts to steal cookies while his sister watches^ the mother washes dishes while the water over~ows#\ WB gave a detailed inventory of objects in the picture "e[g[\ plate\ curtain\ sink\ apron\ stool# and noted that there were {three people|\ but provided no information about the relationship among them[

{{I have trouble recognizing people from just faces alone[ I look at their hair color\ listen to their voices [ [ [ I use clothing\ voice\ and hair[ I try to associate something with a person one way or ano! ther [ [ [ what they wear\ how their hair is worn[||

Preliminary investigations suggested that WB exhi! bited severe impairments in reading[ In this section we describe the results of additional reading assessment[ Unless otherwise noted\ stimuli were taken from the PALPA battery ð13Ł[ Although WB|s visual acuity of 19:39 is adequate for reading average!size print\ he noted on preliminary reading trials that he thought letter rec! ognition to be less di.cult with large stimuli[ Conse! quently\ all stimuli were presented in 13 point Helvetica font[

0[1[ Background testing Clinical neuropsychological testing performed three years prior to the investigations reported here revealed a Verbal IQ in the Low!Average range^ vocabulary and verbal reasoning were well within Average range[ Lan! guage comprehension and ~uency were normal[ Verbal memory\ as assessed by the Wechsler Memory Scale\ Revised\ was in Low!Average range[ Most importantly from the present perspective\ an unusual visual recognition pattern was noticed[ Per! formance was normal on the Benton Visual Form Dis! crimination Test ð2Ł\ and object recognition\ as assessed by the Boston Naming Test ð05Ł\ was in the low normal range relative to age peers "36:59^ 67) correct#[ In contrast\ there was severe impairment on the Benton Face Recognition Test ð2Ł\ a measure of unfamiliar face recognition[ On an informal measure of recognition of 09 familiar\ {high frequency| faces "e[g[\ Hitler#\ WB failed to recognize a single face[ Finally\ WB|s reading was profoundly impaired^ most words were read in a letter! by!letter manner[ At the time of the current evaluation\ WB|s visual acuity was 19:39 by neuro!ophthalmologic exam\ and contrast sensitivity was normal for spatial frequencies ranging from 9[56Ð04 cycles per degree[ WB also per! formed normally on the Farnsworth D!04 Panel Test of color perception[ Detection of degraded shapes on the Visual Object and Space Perception Battery ð25Ł was nor!

1[ Experimental investigations 1[0[ Reading

1[0[0[ Single letter recognition On a letter reversal discrimination task\ WB performed with 64) accuracy "16:25#[ He failed to reject 5:07 reversed letters\ and rejected 2:07 correctly!oriented letters[ On a task requiring him to name upper case letters reported singly\ performance was imperfect "13:15#[ The two errors appeared to be visual in nature "Q : G^ C : G#[ A second letter!recognition task required WB to match a probe letter to a target letter printed in a di}erent case given a choice of two[ He performed with 85) accuracy "12:13#[ In summary\ performance was con! sistent with mild impairment of letter recognition[ 1[0[1[ Word recognition WB was asked to read words varying in grammatical class at a limited\ but long "59ý# exposure[ Performance was severely impaired "06:79^ 10)#\ and was char! acterized by a letter!by!letter strategy[ There was no evi! dence of a word!class e}ect "nouns 4:19^ adjectives 2:19^ verbs 5:19^ functors 2:19#[ All errors were visual "e[g[\ appear : a[p[p [ [ [ apple^ speak : s[o[e[ [ [ [ sock^ client : c[l[i[n[t[ [ [ [ count^ maybe : m[a[y[d[o#[ Left neglect appeared to contribute to only 7) "4:52# of the errors "e[g[\ amount : c[o[u[n[t[ [ [ [ n[o[ [ [ [ not^ ancient :

33

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

c[i[o[n[t[ [ [ [ cent^ else : s[o[ [ [ [ e[l[ [ [ [ also#[ In contrast\ letter recognition di.culties appeared to contribute to almost all errors[ 1[0[2[ Word length effects To brie~y assess whether word length a}ected naming latency\ as would be expected in the case of letter!by! letter reading\ we asked WB to perform the {letter length| subtask of the PALPA ð13Ł\ which contains six words each of three\ four\ _ve and six letters in length[ Reading latency for correctly read words was measured by hand! held stopwatch[ WB correctly read 4:5 three!letter words\ 5:5 four!letter and 5:5 _ve!letter words\ and 2:5 six!letter words[ The four errors were visual in nature "cup : u[p[ [ [ [ up^ cheese : c[h[r[ [ [ [ c[h[e[ [ [ [ chess< chase<^ square : s[g[u[ [ [ [ s[p[u[a[r[e[ [ [ [ spry< spread< spare<^ priest : p[s[t[ [ [ [ p[r[y[ [ [ [ p[r[i[ [ [ [ present<#[ Note that while letter misidenti_cation is evident\ left neglect appears to contribute to only one of the errors[ As Fig[ 0 shows\ data from the correctly read words are suggestive of a marked e}ect of word length on reading latency[ Thus\ consistent with preliminary investigations\ WB exhibits severe letter!by!letter reading\ a pattern of performance associated with pure alexia[ 1[1[ Naming of orally spelled words Pure alexia has frequently been attributed to de_cits in prelexical processes\ such as visual attention or letter recognition\ that preclude parallel access to the visual word!form and semantic systems "see\ e[g[\ ð0\ 03\ 12\ 14\ 29Ł#[ One source of evidence that the locus of reading impairment is indeed prelexical is the ability to recognize orally!spelled words[ There are several lines of evidence

Fig[ 0[ WB|s latency to read words as a function of word length[

that orally!spelled words are recognized by transforming the auditory input into a visual image\ which is {read| via the same visual!word form system used with printed words ð06\ 18\ 20Ł[ Thus\ relatively unimpaired rec! ognition of orally!spelled words would suggest that the word form system is accessible through auditory input\ and that WB|s word recognition problem is visual in nature[ WB was asked to name the same words from various grammatical classes presented in Study 0b above^ in the present Study the examiner spelled each word aloud at a rate of one letter per s[ WB|s performance was superior relative to his de_cits with written words "57:79^ 74) correct^ chi sq[  54[2\ P ³ 9[9990#[ There was no word!class e}ect "nouns 06:19^ adjectives 06:19^ verbs 07:19^ functors 05:19#[ The twelve errors were all close approximations of the target words "e[g[\ severe : savior^ describe : decide#[ WB was also asked to name irregular words from the PALPA battery spelled aloud by the examiner[ His per! formance here\ too\ was relatively unimpaired "14:29^ 72) correct#[ Only one of the _ve errors was attributable to regularization "colonel : colonial#^ the remainder were of indeterminate origin "e[g[\ cough : coach^ rou! tine : runt#[ In summary\ WB is mildly impaired in recognizing letters\ and severely de_cient on tests of word recognition[ His relatively unimpaired performance on naming of words spelled aloud suggests that he is able to access the visual word form system from phonologic input[ This\ in turn\ suggests that the locus of his impairment in word recognition is prelexical[ 1[2[ Face recognition 1[2[0[ Recognition of famous faces Recall that WB notes that he is unable to recognize faces in his daily life\ relying instead on such peripheral cues as hair and clothing[ We _rst assessed WB|s face recognition with the Benton Famous Faces task ð2Ł\ in which a photograph of a person named by the examiner must be selected from an array of three[ His performance was severely impaired "21:64^ 32) correct#[ Control sub! jects aged 39Ð49 performed with 81) accuracy ð2Ł[ WB|s errors could not be attributed to left neglect] more than a third "18# of his responses were to the left!most face in the array[ WB|s attempt to rely upon visual feature cues was evident in many of his spontaneous verbalizations during the task "e[g[\ Rosalyn Carter : she had shortish hair^ Adlai Stevenson : he was bald^ Emperor Hirohito : sounds oriental\ so I can check for the eyes#[ These ver! balizations are suggestive of the integrity of at least some semantic knowledge of the subject|s visual features[ In may cases\ infact\ WB volunteered a wealth of semantic information about the subject "e[g[\ Marilyn Monroe :

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

sexy blonde\ starred with Tony Curtis in Some Like it Hot^ Count Basie : Big Band leader^ Joe Namath : Broadway Joe;# but nevertheless failed to select the target photograph[ Although the evidence is indirect\ WB|s ver! balizations are consistent with the possibility that his prosopagnosia is not attributable to loss of semantic knowledge about the subjects of the photographs[ WB also performed a relatively less di.cult forced! choice famous faces test from our laboratory\ which requires subjects to select a photograph of a person named by the examiner from an array of two[ Photo! graphs of non!famous people serve as foils[ In addition to recent political and sports _gures\ the 37 item test contains numerous photographs of contemporary {icons| of popular culture "e[g[\ Mel Brooks\ Alfred Hitchcock\ Walther Matthau\ Goldie Hawn\ Mick Jagger#[ WB|s performance of 20:37 "54)# correct does not di}er sig! ni_cantly from chance "z  0[77\ P × 9[94 by binomial test#[ 1[2[1[ Recognition of visually degraded faces WB was also given the short form of the Benton Facial Recognition Test\ which assesses whether the subject|s ability to recognize faces is susceptible to visual degra! dation of the stimuli[ It requires a probe photograph to be matched to a target photograph in an array of six^ the probe and target di}er in viewing angle or lighting conditions[ WB performed correctly on only 09:16 "26)# items[ His score\ converted to a long form score of 13 according to the correction procedure described by Benton\ Van Allen\ Hamsher + Levin ð3Ł\ is in the range of severe impairment "normal range 39Ð43#[ Consistent with our speculations based on WB|s verbalizations\ above\ his severe impairment in matching faces when target and probe di}er in viewing angle or lighting sug! gests that at least a portion of his de_cit in facial rec! ognition arises at pre!semantic levels[ 1[3[ Discussion Assessment of WB|s word and face recognition abilities con_rms that he is severely de_cient in recognizing both of these classes of objects[ In both cases\ there is evidence suggesting that the de_cits are attributable to some aspect of pre!semantic processing[ In the context of intact visual acuity\ visual _elds\ contrast sensitivity\ color dis! crimination\ and _gure!ground segmentation\ the data suggest that his de_cit is likely in some aspect of high level visual processing which precludes access to the vis! ual word form system[ In the case of faces\ the level at which the de_cit arises is less clear^ however\ WB|s inability to match faces across di}erent views or in degraded conditions\ as well as the semantic information he spontaneously provides about subjects\ is at least con! sistent with a de_cit prior to the level of semantics[ Below\ we will provide further evidence for the integrity of WB|s

34

access to semantic information about the visual features of common objects and animals "Section 1[6#[ On most accounts\ WB would almost certainly be char! acterized as a severe associative agnosic[ The two!process hypothesis would predict that WB should also be sub! stantially impaired in common object recognition[ We tested this possibility next[ 1[4[ Object recognition 1[4[0[ Matching across view shifts WB was asked to perform our laboratory|s version of Warrington|s Unusual Views Task ð28Ł[ In this version\ subjects are shown a probe photograph depicting an object in a standard view "e[g[\ a side!view of a shoe#\ familiar view "e[g[\ a shoe from a superior {bird|s eye view|#\ or odd view "e[g[\ the sole of an upside!down shoe#\ and are asked to match it to a photograph in an array of three that depicts the same object in a standard view[ On odd view trials\ the probe and target photo! graphs share few "or no# visual features[ There are 22 trials each in the standard\ familiar\ and odd view con! ditions[ WB performed errorlessly in the standard view condition\ and relatively accurately in the familiar view "29:22^ 89)# and odd view "18:22^ 77)# conditions[ In contrast to his severely impaired performance when required to match photographed faces across view shifts "see Section 1[2[1#\ data from the odd view condition of the present study suggest that WB is able to derive a structural description of photographed objects from visual input[ 1[4[1[ Real object naming We next assessed WB|s ability to recognize and name common household objects[ Twenty!_ve objects from the categories tools\ kitchen utensils\ and o.ce supplies were presented to him singly for naming[ Appendix 0 provides a list of the stimuli[ Note that many of these objects were quite similar in color\ size\ and form "e[g[\ tin snips\ wire cutters\ scissors\ hedge trimmers\ pliers\ wrench#\ and that only the precise object name was accepted as correct[ WB|s performance was unhesitating and errorless[ 1[4[2[ Picture naming WB was also asked to name line drawings of objects\ which lack the cues of 2!dimensionality and color present in real objects[ He performed the 064!item version of the Philadelphia Naming Test ð22Ł\ which contains pictured items from a variety of categories[ Stimuli were exposed for 59 s\ and items not correctly named within this time were scored as incorrect[ Latencies to name items cor! rectly were recorded by the examiner from a hand!held stopwatch[ WB performed with an overall accuracy of 67)

35

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

"026:064#[ Seventy!two percent of correct items were named within _ve s[ Test items were divided on a post!hoc basis into eleven categories] food\ plants\ animals\ tools\ miscellaneous man!made objects\ architectural items\ clothing\ natural phenomena\ body parts\ and people[ Although the cat! egories are not balanced for the frequency or familiarity of the pictured objects they contain\ these data are pro! vided as a rough index of WB|s performance with objects of di}erent types[ WB|s accuracy as a function of category is presented in Table 0[ Overall\ WB performed better with non!living things "090:011\ 72)# than with living things "25:42\ 57)#[ He rapidly named objects likely to have numerous parts "e[g[\ horse\ typewriter\ bus\ tractor# as well as those that appear likely to be processed as con_gurational wholes "e[g[\ door\ hat\ candle\ glass\ pillow#[ Errors in all categories appeared visual in nature "e[g[\ elephant : something with a small tail [ [ [ monkey<^ clown "with ru/ed collar# : a lion^ camera : washing machine^ pipe : vacuum cleaner [ [ [ lawn mower [ [ [ something with a spout#[ In some instances\ WB|s erroneous responses appeared to result from failure to integrate the parts of the percept into a unitary object "e[g[\ bin! oculars : medicine bottles^ toilet : a bowl and a table#[ 1[5[ Discussion WB is able to recognize most objects across large shifts in perspective which preclude matching on the basis of low!level visual features[ His ability to recognize a variety of real objects from visually similar categories "tools\ household items\ utensils# is normal[ His recognition of line drawings of objects\ while imperfect\ appears less impaired than his severely de_cient performance on tests of reading and face recognition[ In particular\ per! formance appears relatively accurate for non!living things[ Moreover\ for some categories of objects "e[g[\

Table 0 WB|s picture naming performance Item category

Food Transportation Tools Natural phenomena Plants Architecture Misc[ man!made "non!tool# Humans Body parts Animals Clothing

Items correct N

")#

03:03 6:6 00:00 5:5 3:3 09:01 31:44 09:02 6:09 05:15 8:04

099 099 099 099 099 72 65 66 69 51 59

foods\ transportation\ tools\ natural phenomena\ and plants#\ recognition is perfect[ The fact that WB|s errors were all visual mis! identi_cations suggests a de_cit in visual processing\ rather than in semantic knowledge[ Nevertheless\ the apparent relative impairment with living things raises the possibility that a semantic de_cit\ relatively more severe for living things "or for visual features of both living and non!living things\ see ð02Ł# may contribute to object recognition failure[ We assessed this possibility next[ 1[6[ Semantic knowledge 1[6[0[ Semantic probes test WB was given the a word version of the semantic probes test of Chertkow and Bub ð5Ł\ in which knowledge of each of 19 animals and 19 man!made objects is probed with 4Ð7 forced!choice questions[ These queries probe visual attributes "e[g[\ does a rhinoceros have horns or antlers<# and functional:semantic attributes "e[g[\ does a rhinoceros live in the desert or the plains<# approximately equally[ WB performed perfectly on all 19 man!made items[ He made a single error on an item assessing his knowledge of the shape of donkeys| ears "pointy or ~oppy#[ This performance is well within the normal range[ 1[6[1[ Pyramids and palm trees test WB also performed the Pyramids and Palm Trees Test ð07Ł[ In the version of the task administered to WB\ access to semantic knowledge from auditory input was assessed[ On each trial\ the subject is shown a page containing three written words[ The {probe| word is read aloud "and pointed to# by the examiner\ followed by the two additional words[ Only one of these additional words is associated with the probe "e[g[\ hand!gloves "not slip! pers#^ battery!~ashlight "not lamp#\ and the subject must select the associated word[ WB performed with 83) accuracy "38:41#\ a score just one point below the normal cut!o} "49:41#[ On the assumption that auditory input accesses the same semantic system used in picture recognition ð4Ł "but see ð27Ł#\ these data\ in combination with visual mis! identi_cation errors\ indicate that WB|s failures to reco! gnize pictured objects may be attributed to de_cits in visual\ and not semantic processing[ This\ in turn\suggests that the apparent category speci_city of the pattern of impaired and spared object recognition is likely to be an artifact of some visual characteristic of the stimuli "see ð15Ł#[ The contrast between picture recognition and sem! antic knowledge of objects is similar to the disparities observed in reading and face recognition[ Thus\ it appears possible that a de_cit in the same late visual processing stages a}ects processing of words\ faces\ and objects\ but to unequal degrees[

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

1[7[ Face and object processing] a direct comparison To this point\ we have presented data suggesting that WB exhibits severe impairments in recognizing faces and words\ in the context of much milder impairments in recognizing objects which is manifest only when line drawing!recognition is assessed[ It might be argued that the observed disparities in performance occur because the processing of faces and words rely on the same perceptual resources as do objects\ but tax them more heavily[ In fact\ the di.culty argument has been advanced to explain the apparent selectivity of face processing impairment in prosopagnosia[ One version of this argument suggests that recognizing a speci_c face is di.cult because faces are members of a relatively homogeneous category of objects ð6Ł[ To assess this possibility\ Farah\ Levinson\ and Klein ð01Ł developed an elegant test that directly pits face rec! ognition against recognition of exemplars from another homogeneous group of objects[ Control subjects were asked _rst to study 39 black and white pictures of faces of male students for 2 s each\ and then to perform old:new judgments on a set of 79 face pictures\ half of which are {old|[ The same procedure was used to assess recognition of a set of 39 eyeglass frames which were selected to be as visually similar as possible "e[g[\ there were numerous versions of {aviator| frames and numerous horn!rimmed frames#[ The subjects tested by Farah et al[ ð01Ł performed signi_cantly better with faces than with eyeglasses[ A modi_ed version of the task "three exposures of each photograph\ at 5 s each# was then given to LH\ a prosopagnosic subject[ LH failed to show a face superi! ority e}ect\ performing about equally on the face and eyeglass recognition tasks[ The investigators argued that LH|s problem with face recognition could not be attri! buted to a mild version of a general object recognition impairment[ They concluded that face recognition must rely upon processors which are at least partially distinct from those supporting object recognition[ The faces and eyeglasses test of Farah et al[ ð01Ł allows us to test whether WB|s object recognition is indeed spared relative to his face recognition[ If so\ it would provide support for the assertion that WB|s relatively intact object recognition is not an artifact of the relative di.culty of face recognition\ but re~ects sparing of pro! cesses speci_c to objects[ WB was given the same version of the faces and eye! glasses test administered to LH\ the prosopagnosic sub! ject of Farah et al[ ð01Ł[ WB studied 39 photographed faces three times each for 5 s[ He then viewed a set of photographs which comprised the 39 {old| faces and 39 {new| faces in random order[ The same procedure was used to assess WB|s recognition of eyeglasses[ As shown in Fig[ 1\ the control subjects of Farah et al[ ð01Ł performed signi_cantly more reliably on the face discrimination task[ LH performed approximately equ!

36

ally poorly on both tasks\ suggesting that he exhibited a relative impairment in face recognition[ The disparity between WB|s face and object processing was even more pronounced than that of LH[ He was unable to reliably discriminate {old| from {new| faces "d?  9[25\ z  0[2#[ In contrast\ his discrimination of {old| and {new| eye! glasses was reliable "d?  0[4\ z  2[3\ P ³ 9[94#[ These data indicate that WB|s recognition of objects from a visually homogeneous category is superior to his face recognition[ In addition\ the data suggest that the relative sparing of WB|s object recognition is unlikely to be attributable to the relative ease of the object rec! ognition tasks administered[ Finally\ in accord with the conclusion of Farah et al[ ð01Ł\ it suggests that the pro! cesses underlying object recognition are at least partially distinct from those supporting face recognition[

2[ General discussion Farah|s ð8Ð00Ł account of the two types of structural description underlying recognition of words\ objects\ and faces suggests that patients with severe alexia and pro! sopagnosia are de_cient in representing structural descriptions composed of numerous parts\ required to read\ as well as in representing complex\ wholistic struc! tural descriptions\ necessary for face recognition[ These patients\ then\ should also be severely impaired in com! mon object recognition\ which requires the rep! resentation of both complex and numerous parts[ WB|s performance presents a challenge to this prediction[ WB is a patient with severe alexia and prosopagnosia[ He is unable to read or recognize people in his daily life\ yet he performs perfectly in recognizing common 2! dimensional objects of similar sizes\ shapes\ and colors[ Although his performance on picture naming tasks indi! cates that his object recognition is not perfectly normal\ we have provided evidence that it is certainly spared relative to reading and face recognition[ On a test explicitly designed to render object recognition more di.cult than face recognition in control subjects\ WB|s performance was superior on the object recognition task[ How can this pattern of performance be explained< One possible explanation for the data is that the two! process account is essentially correct\ but requires modi! _cation[ On the modi_ed two!process account\ residual capacities used to represent complex parts and numerous parts could be combined to permit object recognition[ With even minimal sparing of each of these abilities\ it could be argued\ there might be su.cient abstract shape processing to recognize many objects[ While this is a reasonable hypothesis\ the account leads to a prediction that is opposite to that derived from the original theory[ It suggests that there should be numerous patients with prosopagnosia and alexia for whom minimal residual processing permits relatively good object recognition[ As

37

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

Fig[ 1[ Comparison of faces and glasses test performance of control subjects and prosopagnosic LH\ reported by Farah\ Levinson\ + Klein ð01Ł\ with that of patient WB[

noted\ a careful review of the literature ð8Ð00Ł has revealed an absence of such cases[ Other possible accounts of WB|s performance suggest that his spared object recognition is attributable to di}erences between objects and faces at semantic "as well as structural# levels of representation[ On one such account\ WB|s spared recognition of objects "and par! ticularly inanimate objects# relative to faces might result from dissimilarity at both structural and semantic levels within the category of manmade as compared to bio! logical objects "see ð1\ 10Ł#[ Another related account pro! poses that biological objects are distinguished largely by their visual features at structural levels\ whereas man! made objects are distinguished predominantly by func! tional:associative elements in the semantic system[ Thus\ man!made objects may be relatively robust to visual loss ð26Ł[ These accounts\ however\ have di.culty accom! modating WB|s excellent performance on the eyeglass task[ When the requirements of the task are to distinguish highly similar exemplars of a single object "i[e[\ di}erent versions of eyeglasses#\ it seems implausible that semantic and:or functional elements could provide assistance[ In addition\ the semantic accounts are mute with respect to the reasons for WB|s impaired word recognition[ A third possible explanation for the pattern of data obtained is that WB|s de_cits arise at a processing stage {earlier| than that invoked by the two!process account[ It is di.cult to envision\ however\ how the relative superi! ority of WB|s object recognition could be attributed solely to de_cits prior to the level of the structural description[ On such an account either "0# all classes of objects might be expected to be a}ected approximately equally "cf ð08Ł#\ and:or "1# factors relevant to pre!structural processing stages\ such as surface texture\ shading\ color\ 2!dimen!

sionality\ and inter!class similarity might be predicted to a}ect recognition[ Thus\ de_cits limited to pre!structural levels would not explain the resilience of WB|s processing of 1D black!and!white photographs of highly similar eye! glasses\ in the context of his inability to recognize 2D familiar faces even in everyday life[ The present data do not permit us to o}er a de_nitive account of WB|s performance without undue specu! lation[ Indeed\ he may su}er de_cits at several levels of processing[ While we have noted that WB|s relatively spared object recognition appears inconsistent with de_! cits arising solely at pre!structural levels\ it may yet be the case that he su}ers pre!structural de_cits as well as impairment at the level of the structural description "e[g[\ within the structural description system used for face recognition#[ Thus\ the data do not entirely refute the two!process account\ which focuses on processes occur! ring solely at the level of the structural description[ The data suggest\ however\ that the predictions of the account require modi_cation[ Because patterns of high level visual agnosia may derive from interactions of de_cits at several processing levels\ it is indeed possible to observe relative sparing of object recognition in patients with severe pro! sopagnosia and alexia[

Appendix 0 Items of the Real!object naming task "Study 3b# Can opener Candlestick Cheese grater Corkscrew

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

Flashlight Fork Glue Hammer Hand mixer Hedge trimmer Metal spatula Monkey wrench Paintbrush Pepper grinder Pliers Rubber spatula Scissors Scotchtape Screwdriver Sponge Spoon Tin snips Vegetable peeler Wire cutters Wooden spoon

ð03Ł ð04Ł ð05Ł ð06Ł ð07Ł ð08Ł

ð19Ł

ð10Ł ð11Ł

ð12Ł

ð13Ł

References

ð14Ł

ð0Ł Arguin M\ Bub DN[ Single!character processing in a case of pure alexia[ Neuropsychologia 0882^20]324Ð47[ ð1Ł Arguin M\ Bub DN\ Dudek G[ Shape integration for visual object recognition and its implication in category speci_c visual agnosia[ Visual Cognition 0885^2]110Ð64[ ð2Ł Benton AL\ Hamsher K de S\ Varney NR\ Spreen O[ Contributions to neuropsychological assessment[ New York] Oxford University Press\ 0872[ ð3Ł Benton AL\ Van Allen MW\ Hamsher K de S\ Levin HS[ Facial Recognition Test] Form SL[ Unpublished Test Materials\ 0867[ ð4Ł Caramazza A\ Hillis A\ Rapp BC\ Romani C[ The multiple sem! antics hypothesis] Multiple confusions< Cognitive Neu! ropsychology 0889^6]050Ð78[ ð5Ł Chertkow H\ Bub D[ Semantic memory loss in Alzheimer!type dementia[ In] Schwartz MF\ editor[ Modular De_cits in Alzheimer! Type Dementia[ MIT Press\ Cambridge\ MA\ 0889[ ð6Ł Damasio AR\ Damasio H\ Van Hoesen GW[ Prosopagnosia] Ana! tomic basis and behavioral mechanisms[ Neurology 0871^21]220Ð 30 ð7Ł Farah MJ[ Distinguishing perceptual and semantic impairments a}ecting visual object recognition[ Visual Cognition 0886^3]088Ð 195[ ð8Ł Farah MJ[ Specialization within visual object recognition] Clues from prosopagnosia and alexia[ In] Farah MJ\ Ratcli} G\ editors[ The neuropsychology of high!level vision[ Lawrence Erlbaum\ Hillsdale\ NJ\ 0883[ ð09Ł Farah MJ[ Patterns of co!occurrence among the associative agnosias] implications for visual object representation[ Cognitive Neuropsychology 0880^7]0Ð08[ ð00Ł Farah MJ[ Visual agnosia[ MIT Press\ Cambridge\ M[A[ 0889[ð01Ł Farah MJ\ Levinson KL\ Klein KL[ Face perception and within!category discrimination in prosopagnosia[ Neu! ropsychologia 0884^22^550Ð63[ ð02Ł Farah MJ\ McClelland JL[ A computational model of semantic memory impairment[ Modality!speci_city and emergent category!

ð15Ł

ð16Ł

ð17Ł ð18Ł

ð29Ł

ð20Ł

ð21Ł

ð22Ł

ð23Ł

ð24Ł

ð25Ł

38

speci_city[ Journal of Experimental Psychology] General 0880^019]228Ð46[ Farah MJ\ Wallace MA[ Pure alexia as a visual impairment] A reconsideration[ Cognitive Neuropsychology 0880^7]202Ð23 Gomori AJ\ Hawryluk GA[ Visual agnosia without alexia[ Neu! rologica 0873^23]836Ð49[ Goodglass H\ Kaplan E[ Boston Naming Test[ 1nd ed[ Phi! ladelphia] Lea + Febiger\ 0861[ Hanley JR\ Kay J[ Does letter by letter reading involve the spelling system< Neuropsychologia 0881^29]126Ð45[ Howard D\ Patterson K[ The Pyramids and Palm Trees Test[ Bury St[ Edmunds\ U[K[\ Thames Valley Test Co[\ 0881[ Humphreys GW\ Riddoch MJ[ To see but not to see] A case study of a visual agnosia[ Hove\ U[K[] Lawrence Erlbaum Associates\ 0876[ Humphreys GW\ Riddoch MJ\ Donnelly N\ Freeman T\ Boucart M\ Muller HM[ Intermediate visual processing and visual agnosia[ In] Farah\ MJ\ Ratcli}\ G\ editors[ The neuropsychology of high! level vision[ Hillsdale\ NJ] Lawrence Erlbaum\ 0883[ Humphreys GW\ Riddoch MJ\ Quinlan PT[ Cascade processes in picture identi_cation[ Cognitive Neuropsychology 0877^4]56Ð092[ Humphreys GW\ Rumiati RI[ Stimulus speci_city in visual rec! ognition] Agnosia without prosopagnosia or alexia[ Cognitive Neuropsychology\ in press[ Kay J\ Hanley R[ Simultaneous form perception and serial letter recognition in a case of letter by letter reading[ Cognitive Neu! ropsychology 0880^8]138Ð62[ Kay J\ Lesser R\ Coltheart M[ PALPA[ Lawrence Erlbaum Associ! ates\ Hove\ 0881[ Kinsbourne M\ Warrington EK[ A disorder of simultaneous form perception[ Brain 0851^74]350Ð75[ Lloyd!Jones TJ\ Humphreys GW[ Perceptual di}erentiation as a source of category e}ects in object processing] Evidence from naming and object decision[ Memory and Cognition 0886^14]07Ð 24[ Lissauer H[ Ein fall von seelenblindheit nebst einem Beitrage zur Theori derselben[ Archiv fur Psychiatrie und Nervenkrankheiten 0789^10]111Ð69[ Logothetis NK\ Sheinberg KL[ Visual object recognition[ Annual Review of neuroscience 0885^08]466Ð510[ Monsell S[ On the relation between lexical input and output path! ways for speech[ In] Allport A\ Mackay D\ Prinz W\ Scheerer E\ editors[ Language perception and production] Relationships between listening\ speaking\ reading\ and writing[ London] Aca! demic Press\ 0876[ Patterson K\ Kay J[ Letter by letter reading] Psychological descrip! tions of a neurological syndrome[ Quarterly Journal of Exper! imental Psychology 0871^23A]300Ð30[ Rapcsak SZ\ Rubens AB\ Laguna JF[ From letters to words] Procedures for word recognition in letter by letter reading[ Brain and Language 0889^27]493Ð03[ Riddoch MJ\ Humphreys GW[ Visual object processing in Optic Aphasia] A case of semantic access aphasia[ Cognitive Neu! ropsychology 0876^3]020Ð74[ Roach A\ Schwartz MF\ Martin N\ Grewal RS\ Brecher A[ The Philadelphia Naming Test] Scoring and rationale[ Clinical Aphasi! ology 0885^13]010Ð22[ Rumiati R\ Humphreys GW\ Riddoch MJ\ Bateman A[ Visual object agnosia without alexia or prosopagnosia[ Visual Cognition 0883^0]070Ð114[ Sartori G\ Job R[ The oyster with four legs] A neuropsychological study on the interaction of visual and semantic information[ Cog! nitive Neuropsychology\ 0877^4]094Ð21 Warrington EK\ James M[ Visual Object and Space Perception Battery[ Thames Valley Test Co[\ Bury St[ Edmunds\ Su}olk\ UK\ 0880[

49

L[J[ Buxbaum et al[:Neuropsycholo`ia 26 "0888# 30Ð49

ð26Ł Warrington EK\ McCarthy R[ Categories of knowledge] Further fractionation and an attempted integration[ Brain 0876^009]0162Ð 85[ ð27Ł Warrington EK\ Shallice T[ Category!speci_c semantic impair! ments[ Brain\ 0873^096]718Ð42[

ð28Ł Warrington EK\ Taylor AM[ The contribution of the right parietal lobe to object recognition[ Cortex\ 0862^8]041Ð53[ ð39Ł Wilson BA\ Cockburn J\ Halligan P[ Rivermead Behavioral Inat! tention Test[ Thames Valley Test Co[\ Titch_eld\ UK\ 0876[