COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2), 171–180

PERSON-SPECIFIC KNOWLEDGE AND KNOWLEDGE OF BIOLOGICAL CATEGORIES Janice Kay University of Exeter, UK

J. Richard Hanley University of Essex, Colchester, UK

We report new findings from a neurological case described by Hanley, Young, and Pearson (1989). In the original study, the patient, BD, had impaired biographical knowledge of people which appeared to extend to a general impairment of knowledge of “living things”. In more recent work, we present evidence which confirms Hanley et al.’s finding that BD has impaired person-specific knowledge, but we suggest that this is not associated with a more general impairment to do with knowledge of biological categories. We propose that an artefactual explanation of the original account is more likely based on differences in “age of acquisition” between items from “living” and “nonliving” categories. We conclude that biographical knowledge of people is represented separately from knowledge of biological categories in semantic memory.

INTRODUCTION In recent years, a considerable number of published neuropsychological studies have reported cases of apparent category-specific dissociations within semantic memory. The most commonly reported pattern is one in which biological or living categories (e.g. animals, birds, fruits, vegetables) are less well preserved than nonliving, inanimate categories (e.g. furniture, kitchen utensils, clothes); see McCarthy and Warrington (1994) for a review. Although it has been demonstrated that exemplars from biological categories can be less frequently encountered, less familiar, and more visually complex than exemplars from inanimate categories (e.g. Funnell & Sheridan, 1992; Stewart, Parkin, & Hunkin, 1992), it is clear that an artefactual account of a “living–nonliving” dissociation is not sufficient to explain all cases, since the dissociation can

be found even when cognitive variables such as familiarity and visual complexity are controlled (e.g. Farah, Meyer, & McMullen, 1996). Neither can an artefactual account explain (much rarer) reports of the reverse dissociation (e.g. Hillis & Caramazza, 1991; Sacchett & Humphreys, 1992; Warrington & McCarthy, 1987). What counts as a “biological” category has to a large extent been driven not by a priori theoretical assumptions, but experimentally by the patterns of disorder that have been reported. It is clear that in some cases, the profile of impairment includes difficulties with categories that certainly would not be regarded as biological (e.g. musical instruments). The findings that have been observed through this evidence-collecting approach have led to interpretations of the disorder that emphasise ways in which different categories of knowledge may be acquired (e.g. the relative importance of perceptual vs.

Requests for reprints should be addressed to Dr. J.M. Kay, School of Psychology, University of Exeter, Perry Road, Exeter EX4 4QG, UK (Tel: +44 1392 264663; Fax: +44 1392 264623; email: [email protected]).

Ó 1999 Psychology Press Ltd

171

KAY AND HANLEY

functional attributes in biological versus nonbiological categories), rather than one which postulates a taxonomic structure for semantic memory (e.g. Farah & McClelland, 1991; Warrington & McCarthy, 1987). Given this emphasis, it is interesting to explore the limits of what may be deemed “biological”, and what kinds of impairments in biological knowledge tend to co-occur. In particular, we have been keen to discover whether the representation of biological knowledge also includes biographical knowledge about people, and whether disorders in which person-specific knowledge is impaired extend to comprehension difficulties of living things in general. There have been rare reports of acquired neurological damage specifically affecting biographic knowledge of familiar people. For example, Kapur, Heath, Meudell, and Kennedy (1986) describe the case of ED, who presented with a permanent retrograde amnesia (without corresponding anterograde amnesia), which particularly affected knowledge of familiar people, assessed through photographs and names. Ellis, Young, and Critchley (1989) report the performance of a patient, KS, with a highly specific deficit in identifying people familiar to her, whether from voices, photographs, or names. A third case has also been investigated by Hanley et al. (1989) in which a post-encephalitic patient, BD, was again found to have profound difficulties in identifying people from their face, name, and voice. The impairment affected people he had known before his illness and people that he had encountered in the period following his illness. More recently, Evans, Heggs, Antoun, and Hodges (1995) have reported a case of a patient, VH, who presented with an impaired ability to recognise faces, which progressed over several months to a “cross-modality loss of personbased semantic knowledge”. Finally, Kartsounis and Shallice (1996) have described the case of a patient, ABR, whom they claim has a modal1

ity-specific semantic memory loss for “unique objects”, including familiar people. It is important to emphasise that the impairment from which these patients suffer is not confined to recognising people from their faces. For example, the experimental investigations of BD (Hanley et al., 1989) and KS (Ellis et al., 1989) revealed that they were equally poor at recognising famous names and voices as they were at recognising famous faces. This is quite different from what is observed in cases of prosopagnosia, in which the core deficit is a profound difficulty in recognising familiar faces together with a preserved ability to identify familiar names (e.g. Young & De Haan, 1988). In the majority of these cases, right-hemisphere damage, and specifically damage to temporal lobe 1 structures, is clearly implicated. KS (Ellis et al., 1989) had difficulty in identifying familiar people following a right temporal lobectomy for intractable epilepsy. The person identification difficulties faced by BD (Hanley et al., 1989) resulted from herpes simplex encephalitis, which involved damage to anterior structures of the right temporal lobe. VH’s (Evans et al., 1995) progressive disorder of person recognition and identification was found to be associated with selective right temporal lobe atrophy. However, despite some evidence for similar neuroanatomical impairments in the majority of these patients, there appear to be theoretically important behavioural differences between them. One of relevance to the present investigation is whether the disorder of person-specific knowledge is also accompanied by more general difficulties in identifying living things. BD, for example, performed poorly in identifying exemplars from “living” categories, relative to control subjects, on a number of picture naming and naming-to-definition tasks, and Hanley et al. conclude that he had difficulties in gaining access to detailed semantic information about members of biological categories, including

In the case of ED (Kapur et al., 1986), CT and MRI scanning failed to reveal evidence of structural lesion. EEG investigation showed temporal lobe abnormalities, “more prominent in the left temporal region,” indicative of temporal lobe epilepsy. On the basis of this evidence alone, one cannot rule out right hemisphere temporal lobe damage. MRI scanning of patient ABR (Kartsounis & Shallice, 1996) showed the patient to have bilateral multiple small lesions in keeping with small vessel ischaemia following anoxia, which resulted from major myocardial infarction.

172

COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

PERSON-SPECIFIC KNOWLEDGE

people. KS, on the other hand, did not appear to have problems in naming items from living categories; her performance was consistently within the normal range when asked to name such items. Data from the other patients are less clear. Relevant tests were not given to ED. VH’s performance in comprehension and naming was not tested across a range of categories, but the authors do show that her knowledge of flowers and buildings was well preserved relative to that of person-specific knowledge (although her naming of flowers was somewhat weak). ABR was mildly impaired in naming items from two biological categories (flowers and dogs), compared with one nonliving category (cars), although this may have been wholly attributable to associated problems in word retrieval. One might conclude from these findings that in spite of Hanley et al.’s proposal that “people might be considered to form a particular subgroup of ‘living things’”, evidence from other cases, and particularly that of KS, does not support this claim. However, it is unclear how far the particular patterns of disorder displayed by these patients result from difficulties at the same cognitive level. For example, Hanley et al. suggest that BD’s comprehension difficulties reflect a problem in gaining access to semantic knowledge, since they were able to show that he had covert knowledge of people he could not identify explicitly. Although this important test was not able to be given to KS, Ellis et al. claim that her comprehension difficulties are compatible with a problem within semantic memory itself. On the basis of consistency of their patient’s performance across testing sessions, Evans et al. propose that VH, like KS, had lost person-specific knowledge, although there was also evidence that such knowledge was not always completely lost, and could be retrieved through cueing. Given such possible differences between the patients, it is of interest to establish whether the co-occurrence of a disorder of person-specific knowledge with one of impaired knowledge of living things is reliable in the case of BD. This question is particularly pertinent given the recent findings of Funnell and Sheridan (1992) and Stewart et al. (1992), that exemplars from biological categories can be more familiar, have more frequent

names, and can be more complex structurally than exemplars from nonliving, inanimate categories. These authors demonstrated that apparent differences in comprehending and naming items from living and nonliving categories disappeared when stimulus sets were matched on these variables. We have recently had the opportunity to work again with BD. Our initial aim was to discover whether his problems in recognising people remained as severe as they had been previously. To this end, we compared his ability to recognise faces and names with his ability to recognise items from common noun categories using tasks with similar presentation and response formats. Our main aim, however, was to discover whether his semantic memory impairment would selectively impair comprehension of the category of living things when variables such as familiarity, word frequency, and visual complexity were controlled (Funnell & Sheridan, 1992; Stewart et al., 1992). BD was therefore asked to name pictures of members of living and non-living categories using the same materials as those employed by Funnell and Sheridan (1992) and Farah et al. (1996).

CASE REPORT BD is a 68-year-old man living in the North-East of England. In 1974 he was diagnosed as suffering from herpes simplex encephalitis, and a subsequent CAT scan showed an area of damage in the right temporal region, lying at the anterior end of the inferior longitudinal fascisculus (no additional left temporal lobe damage was evident on the scan). BD’s primary complaint, investigated extensively by Hanley et al. (1989), was an inability to identify people other than his wife, his immediate family, and close friends. He also presented with severe topographical problems, and complained that he was no longer able to remember information that he had read in books and newspapers. BD has fluent spontaneous speech and conversational abilities; clinical tests carried out when he was assessed in 1976, and by Hanley et al., did not reveal evidence of language abnormality. COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

173

KAY AND HANLEY

Hanley et al. describe anecdotally the difficulties that BD faces in failing to recognise, either by face or voice, familiar people. They report that, “the severity of BD’s difficulties in remembering people can be seen in the fact that he still only recognises one of us (Richard Hanley), though we have been to see him many times”. Eight years later, and without any contact in the intervening years, BD was unable to recognise Richard Hanley, although he recalled that someone had visited him in connection with his memory difficulties. BD and his wife illustrated his difficulties by telling us that he failed to recognise his young grandchildren by face or voice without the context of his home or the grandchildren’s home to help him. On tests of person identification published in their 1989 paper, Hanley et al. were able to show that BD was impaired not only in recognising, but also in providing identifying information about, familiar people (see Table 1). On the Lancaster/Radcliffe faces “line-up” test, for example, BD’s performance in recalling the correct occupation of a famous face or name was substantially impaired compared with that of control subjects. Although he was more successful with highly familiar people (e.g. Margaret Thatcher, Queen Mother), than with less familiar people (e.g. Lucille Ball, Marlene Dietrich), his performance was still worse than that of control subjects with both types of item. His knowledge of familiar people was impaired whether he was tested with faces, voices, or names. To test his knowledge of categories other than people, BD was given items from a variety of “living” and “nonliving” categories to name. He was shown to be impaired at naming items from some of the “living” categories (e.g. fruit, flowers, vegetables, and breeds of dog), but not “nonliving” categories (e.g. musical instruments, tools) compared with control subjects. Although the authors consider that differences in word frequency might explain the pattern of results, this account is rejected on the grounds that frequencies of the “nonliving” items were comparable with those of “living” items. Findings from more recent studies (e.g. Funnell & Sheridan, 1992), however, indicate that it is important to control carefully for a variety of cognitive variables

174

COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

in testing for differences between different categories in naming.

EXPERIMENTAL TESTS Comprehension of Exemplars from Common Noun Categories (see Table 2 for Summary) The Pyramids and Palm Trees (PPT) test (Howard & Patterson, 1992). The PPT test examines the ability to make semantic associations on the basis of general semantic knowledge (e.g. associating a picture of a pyramid with a pine tree rather than a palm tree). There are 52 items in the test and non-brain-damaged elderly controls make 3 errors or fewer. On the three-picture version of this test, BD scored 49/52, which is similar to his performance when tested by Hanley et. al (1989) (see Table 1). He is not impaired on this task relative to the performance of elderly control subjects.

Word–Picture Matching (Kay, Lesser, & Coltheart, 1992) On the spoken word–picture matching test from the PALPA battery, BD scored 38/40, producing two errors that were both close semantic distractors (cobweb ® spider, stirrup ® saddle). This is within the normal range (five errors or fewer) and is the same score as that recorded by Hanley et al. (1989). Picture Decision This test, adapted from the design and procedure suggested by Howard and Orchard-Lisle (1984), presents the subject with a different name for a picture on three separate occasions: the correct name, a name of a semantic coordinate, and an unrelated name. Shown a picture of a lock, for example, the subject is asked: “Is it a lock?”, or, “Is it a key?”, or, “Is it a mug?”. The picture set consisted of 40 black-and-white photographs. The set was presented to BD in three testing sessions, and the order of correct name, coordinate, and unrelated name

PERSON-SPECIFIC KNOWLEDGE

Table 1. BD’s Performance on Selected Tests from Hanley et al. (1989) BD Retrieval of occupations of famous faces and names High familiar people Faces .60 Names .70 Low familiar people Faces .20 Names .50 WAIS Verbal IQ Perfomance IQ

Controls

SD

.94 .98

.06 .04

.65 .96

.28 .06

Normal Range

107 99

Pyramids & Palm Trees

49/52

49–52

PALPA Word-Picture Matching test

37/40

35–40

Naming pictures of members of semantic categories Musical Instruments .75 Garden Tools 1.00 Animals .83 Fruits .50 Vegetables .56 Flowers .42

was randomised across sessions. The test was given in an auditory version, in which the stimulus name was spoken to BD. Overall, he accepted 36/40 (90%) of correct names, and rejected 40/40 (100%) of unrelated names. Of the semantically related names, although he rejected correctly 34/40 (85%), 6/40 (15%) were accepted as correct. As non-brain-damaged controls made no errors, BD shows evidence of a slight impairment on this task. Synonym Judgement (Kay et al., 1992) On an auditory synonym judgement task in which the subject is asked to decide whether pairs of spoken words are either closely related in meaning or unrelated, BD scored 58/60. The test examines the effect of imageability on synonym judgements, but the results show that this variable had no significant effect on BD’s performance: 30/30 correct in judging high-imageability pairs and 28/30 correct in judging low-imageability pairs.

.84 .84 .89 .88 .78 .78

.11 .11 .07 .10 .17 .13

Person Identification and Semantics (see Table 2 for Summary) Next we examined BD’s comprehension of biographical knowledge using tests with a similar structure to those just described. In Hanley et al.’s original investigation, BD’s ability to identify familar people was tested by requiring him to produce biographical knowledge and names (e.g. occupations of famous people in the Lancaster/Radcliffe Famous Faces Line-Up test, and identifying famous people by voice; Meudell, Northen, Snowden, & Neary, 1980), and to carry out a forced-choice task of matching name and occupation. The current tasks test comprehension without requiring a spoken response and also make use of distractors that are closely related semantically to the target person. The following tasks were specifically designed to match the tasks in the previous section in terms of presentation format and the nature of the response that was required. COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

175

KAY AND HANLEY

Table 2. BD’s Performance on Tasks Requiring Comprehension of Members of Common Noun Categories and Person Identification and Comprehension Controls ———————————– BD Range Mean Comprehension of common noun categories Pyramids & Palm Trees .94 Word-Picture Matching .95 Picture Decision .95 Synonym judgement .97 Person identification and semantics Semantic association .53 (famous people) Face–Name matching .13 Face decision chance

.94-1.00 .88-1.00 1.00 .97-1.00

.84–1.00 1.00 .97–1.00

Semantic Association Using Famous Names A version of the Pyramids and Palm Trees Test was constructed using the names of celebrities of whom a person of BD’s age would be aware. Pairs of names were selected which were those of famous couples, either married (e.g. Lauren Bacall and Humphrey Bogart, Dulcie Gray and Michael Dennison), or who performed together (e.g. Dawn French and Jennifer Saunders, Jane Torvill and Christopher Dean). While some of these pairs had strong verbal associative links (e.g. Lennon and McCartney, Abbott and Costello), others had not (e.g. Muir and Norden, Madeley and Finnegan). One out of each pair was given as a stimulus name, the other acted as the target. For each pair, a distractor name was then chosen using careful criteria (that target and distractor were of the same gender and age, and closely matched in occupation, length of time of celebrity, etc.); for example: Dawn French (Stimulus), Jennifer Saunders (Target), Victoria Wood (Foil); Lauren Bacall (Stimulus), Humphrey Bogart (Target), Robert Mitchum (Foil). Nineteen triplets of names were constructed and presented using the PPT format for three written names. BD produced a correct response for 10 of the items, but made the wrong decision for 9 of them. He therefore performs at chance on this forced-choice task, a finding that accords with his own claim that the

176

COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

names were unfamiliar to him and that he was merely guessing. Fifteen control subjects (with a mean age of 79 years) had a mean score of 16.02 items correct on this test (a range of 0 to 3 errors). Name–Face Matching This test was designed to parallel the word–picture matching test from the PALPA described earlier. BD was asked to match a spoken name with its corresponding face. On each trial, BD was given a choice of five photographs from which to select the target: the target itself, a photograph of someone who is semantically confusable with the target, a more distantly related foil, a photograph of someone who looks visually similar to the target (judged by comparison of the photographs used in the test), and a photograph of someone who is unrelated to the target (but of the same gender). Examples of the items are: Paddy Ashdown (target), David Steel (close distractor), Dennis Healey (distant distractor), Charles Dance (visual distractor), Russell Harty (unrelated distractor); Penelope Keith (target), Felicity Kendal (close distractor), Nerys Hughes (distant distractor), Sue Lawley (visual distractor) and Edwina Currie (unrelated distractor). Again, BD claimed that the names and faces had little familiarity for him and this seems to be borne out by his score of 2/15 correct (chance level on this test would be a score of 3/15). On the other hand, while he made 2 “unrelated” errors and 1 “visual” error, 7/15 of his errors were closely semantically related and 3/15 were distantly semantically related. Overall, then, 80% (12/15) of his responses were either correct or from the same broad semantic category as the target; 60% (9/15) were either correct or closely related semantically to the target. Control subjects were at ceiling on this test. Face Decision We devised a test which was similar to the “Is it a?” task, but with photographs of famous faces. BD was either given the correct name, the name of someone who might be confused with the target, and someone who is completely unrelated to the target: Ronnie Barker (target), Ronnie Corbett (distractor), Tony Curtis (unrelated). There are 36

PERSON-SPECIFIC KNOWLEDGE

photographs in the test, and it was designed so that each photograph is presented three times in separate testing sessions (with order of presentation randomised across items). BD, however, found the task very distressing and it was discontinued after the first session. He felt unable to make a decision to 25/36 face-name pairings. Of the ones to which he did respond, 6/11 were correct (either accepting the correct name, or rejecting the wrong name), and 5/11 were incorrect. On this test, BD shows no evidence of face or name recognition and performs at chance level.

Discussion In comparison with his comprehension of exemplars from common noun categories, and using tests with similar formats, BD is severely and substantially impaired in comprehending famous and highly familiar people. However, his performance on the name–face matching test, in which his errors were almost all related semantically to the target (rather than distributed randomly across error categories), is in line with Hanley et al.’s (1989) claim that he has covert biographical knowledge of people whom he is unable to identify explicitly. Given that we have established that BD’s comprehension profile is essentially identical to that reported by Hanley et al., we turned next to re-examine his performance in identifying exemplars from living versus nonliving categories. Consistent with the previous findings of Hanley et al. (1989), BD shows well-preserved comprehension of exemplars from common noun categories used in these tests (although he did show a slight impairment on the picture decision task). None of the tests systematically manipulated items from living and nonliving categories, and so we cannot tell from them whether this variable will play an important part in his naming ability.

Naming to Confrontation of Pictures from Living and Nonliving Categories BD was given the two stimulus sets of living and nonliving items constructed by Funnell and Sheridan (1992), one of which controls for word

frequency and familiarity (their Appendix 2), and the other for word frequency, familiarity, and visual complexity (their Appendix 3). Both pictures and ratings were taken from the Snodgrass and Vanderwart (1980) norms. Each picture consisted of a black-and-white line drawing and was presented individually to BD, centred on A4 plain paper. He was given unlimited time to name the picture, although his responses were rapid and fluent. Given that Farah et al. (1996) have shown that, using Funnell and Sheridan’s materials, a living–nonliving dissociation can appear cumulatively over repeated presentations, BD was given each set twice. Funnell and Sheridan’s Appendix 2 set consists of 68 items (34 living and 34 nonliving). Twelve of the items in each group are of low familiarity and low word frequency (e.g. duck, drum), 12 are of high familiarity and low word frequency (e.g. squirrel, scissors) and 10 are of high familiarity and high word frequency (e.g. eye, sun). BD’s performance in each of the testing sessions is summarised in Table 3. The first time he was given the pictures to name, he made 12 errors on items that are “living”; 11/12 of these items had names that are low in frequency. At the same time, he made five errors on items that are “nonliving,” and all of these had names that are low in frequency. Although there is a small difference between living and nonliving exemplars (22/34 and 29/34 correct, respectively), this difference is not significant statistically (Fisher exact test, z = 0.28). On the second occasion, he incorrectly named 6/34 exemplars from “living” categories, and 6/34 exemplars from “nonliving” categories (combining across categories, 11/12 errors again had low frequency names). There is therefore no clear evidence from these results that BD’s identification and word retrieval is affected by whether something is living or nonliving. Funnell and Sheridan’s Appendix 3 set consists of 24 items (12 living and 12 nonliving) matched for familiarity, word frequency and visual complexity. BD’s performance (across testing sessions) is also summarised in Table 3. The first time he was given the pictures to name, he made three errors on items that are “living” and four errors on items that are “nonliving.” In the second session he again proCOGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

177

KAY AND HANLEY

Table 3. Number of Error Responses Made by Patient BD on Naming Items from Living/Nonliving Categories, Matched for Familiarity and Word Frequency (Funnell & Sheridan, 1992; Appendix 2). and also for Visual Complexity (Funnell & Sheridan, 1992; Appendix 3)

Appendix 2 Pictures Low familiarity, Low frequency (N = 12) High familiarity, Low frequency (N = 12) High familiarity High frequency (N = 10) Totals

Living ————— T1 T2 a

Non-Living ————– T1 T2

4

2

5

5

7

3

0

1

1

1

0

0

12

6

5

6

3

3

4

4

Appendix 3 Pictures a

Test Session

duced three errors on “living” items and four errors on “non-living” items. He therefore performs no differently in naming living and nonliving items from this set.

was that of age of acquisition (b = – .26; P =.03). Note that age of acquisition appears to be a better predictor of BD’s naming success than either familiarity or word frequency, as there were no additional effects of these variables in the analysis (familiarity: b = – .04, P = .79; frequency: b = – .11, P = .39), once age of acquisition had been taken into account. None of the other variables, including animacy, approached significance. From this analysis, it appears that BD’s naming is influenced by age of acquisition, rather than other factors such as animacy. In line with this finding, a re-examination of the Funnell and Sheridan Appendix 2 set revealed that although living and nonliving items are matched for familiarity and word frequency, the set of living things has names that are on average acquired significantly later than those of the nonliving set (Wilcoxon Signed Rank, z = – 3.45; P < .001). Age of acquisition effects in spoken word retrieval therefore need to be ruled out if an advantage for naming items from nonliving categories is shown using these materials (e.g. Farah et al. 1996).

GENERAL DISCUSSION Discussion Overall, there is no clear evidence for a living–nonliving dissociation using these materials. On the other hand, we were concerned by the finding that BD was slightly more successful in naming nonliving items in the first testing session using the Appendix 2 pictures, even though the difference did not reach statistical significance. One way of addressing this concern was to carry out a further analysis on the naming accuracy data. Using multiple regression, we examined the effects of a number of predictor variables (animacy, word frequency, familiarity, age of acquisition, name agreement, phoneme length, and visual complexity), on the items named correctly across the two testing sessions (0, 1, or 2 correct). Ratings for familiarity, word frequency, age of acquisition, name agreement, and visual complexity were taken from the norms accompanying the Snodgrass and Vanderwart (1980) pictures. With all the independent variables entered simultaneously, the only significant effect

178

COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

We have found no evidence to indicate that BD’s spoken naming is influenced by whether an item is a member of a biological category. Neither is there any sign that such an effect might appear over repeated presentations. We suggest that the original finding of Hanley et al., that BD has difficulty in comprehending items from “living” categories, does not turn out to be the case on closer examination. Indeed, it seems that this finding may be attributable to other differences between the materials, such as age of acquisition. By the same token, however, our results suggest that BD has a secondary difficulty in spoken word retrieval (which is consistent with the possibility that he not only has right temporal lobe damage, but also impairment to left temporal structures). This difficulty may be attributable to the mild comprehension problem that we observed on the picture decision test, for example. Alternatively, it may stem from an impairment to the process of lexical retrieval itself, a possibility that is compatible with

PERSON-SPECIFIC KNOWLEDGE

findings from other case studies of impaired word retrieval which suggest a role for age of acquisition in accessing lexical/phonological representations in 2 word production (e.g. Hirsh & Ellis, 1994) . It is important to note that although our findings suggest that BD may have a mild semantic memory problem, it seems to be nowhere near as severe as his problems in identifying people, and is no worse for living things than for nonliving things. Even when tasks are comparable, with similar formats and response demands, BD is mildly impaired at comprehending/naming objects, but is at chance at identifying people. BD’s impairment of biographical knowledge does not therefore constitute part of a wider impairment to do with the representation of “living things.” We have seen that, apart from BD, none of the patients with selective impairment of biographical knowledge appears also to have a more widespread impairment in knowledge of biological categories. We have shown that in BD’s case, the original finding may have an artefactual explanation. From the rare cases that exist in the neuropsychological literature, we conclude that disorders of person-specific knowledge do not occur in the context of a more general impairment to do with “living things.” Caramazza and Shelton (1998, p. 1) have recently argued that data from patients with selective semantic memory deficits can be best explained by assuming that the deficts genuinely reflect semantic category boundaries rather than “damage to noncategorically organised visual or functional subsystems.” The conclusion that BD is suffering from a problem in remembering semantic information about people that does not extend to other living things appears to support this view. The recognition of people, just like the recognition of all living things, would seem likely to require an extensive knowledge of visual properties. If, therefore, the core deficit in a patient such as BD is an impairment to the representation of sensory knowledge (Farah & McClelland, 1991; Warrington & McCarthy, 1987), then one would not have expected that his impairment would largely be restricted to knowl-

edge of people. Caramazza and Shelton suggest that the ability to categorise an object rapidly as an animal, plant life, or neither, together with the ability to distinguish different types of animal and different types of plant, may confer important evolutionary advantages. Consequently, they argue that the categories of animals and plant life may be subserved by distinct dedicated neural subsystems separate from those involved in the representation of artefacts such as tools or furniture. Cases such as that of BD strongly suggest that knowledge of people, like that of animals and plants, may be represented by its own distinct semantic subsystem. Moreover, the evolutionary argument for a system that can rapidly distinguish different people from one another and from other animals would appear to be equally strong. Manuscript received 31 March 1998 Revised manuscript received 5 August 1998 Manuscript accepted 6 August 1998

REFERENCES Caramazza, A., & Shelton, J.R. (1998). Domainspecific knowledge systems in the brain: The animate-inanimate distinction. Journal of Cognitive Neuroscience, 10, 1–34. Ellis, A.W., Young, A.W., & Critchley, E.M.R. (1989). Loss of memory for people following temporal lobe damage, Brain, 112, 1469–1483. Evans, J.J., Heggs, A.J., Antoun, N., & Hodges, J.R. (1995). Progressive prosopagnosia associated with selective right temporal lobe atrophy. Brain, 118, 1–13. Farah, M., & McClelland, J. (1991). A computational model of semantic memory impairment: Modality specificity and emergent category specificity. Journal of Experimental Psychology: General, 120 , 339–357. Farah, M., Meyer, M.M., & McMullen, P.A. (1996). The living/non-living distinction is not an artefact: Giving an a priori implausible hypothesis a strong test. Cognitive Neuropsychology, 13, 137–154.

2

It is pertinent to note at this point that while BD’s semantic comprehension (of items other than people) is considerably more preserved than that of Hirsh and Ellis’ (1994) patient, both make primarily semantic errors in spoken naming. COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

179

KAY AND HANLEY

Funnell, E., & Sheridan, J. (1992). Categories of knowledge? Unfamiliar aspects of living and nonliving things. Cognitive Neuropsychology, 9, 135–153. Hanley, J.R., Young, A.W., & Pearson, N. (1989). Defective recognition of familiar people. Cognitive Neuropsychology, 6, 179–210. Hillis, A., & Caramazza, A. (1991). Category-specific naming and comprehension impairment: A double dissociation. Brain, 114, 2081–2094. Hirsh, K., & Ellis, A.W. (1994). Age of acquisition and lexical processing in aphasia: A case study. Cognitive Neuropsychology, 11, 435–458. Howard, D., & Orchard-Lisle, V. (1984). On the origin of semantic errors in naming: Evidence from a case of a global aphasic. Cognitive Neuropsychology, 1, 163–190. Howard, D., & Patterson, K.E. (1992). The Pyramids and Palm Trees test. Bury St. Edmunds: Thames Valley Test Co. Kapur, N., Heath, P., Meudell, P., & Kennedy, P. (1986). Amnesia can facilitate memory performance: Evidence from a patient with dissociated retrograde amnesia. Neuropsychologia, 24, 215–221. Kartsounis, L.D., & Shallice, T. (1996). Modalityspecific semantic knowledge loss for unique items. Cortex, 32, 109–119. Kay, J., Lesser, R., & Coltheart, M. (1992). Psycholinguistic Assessments of Language Processing in

180

COGNITIVE NEUROPSYCHOLOGY, 1999, 16 (2)

Aphasia. Hove, UK: Lawrence Erlbaum Associates Ltd. McCarthy, R., & Warrington, E. (1994). Disorders of semantic memory. Philosophical Transactions of the Royal Society of London. B, 346, 89–96. Meudell, P., Northen, B. Snowden, J., & Neary, D. (1980). Long-term memory for famous voices in amnesic and normal subjects. Neuropsychologia, 18, 133–139. Sacchett, C., & Humphreys, G.W. (1992). Calling a squirrel a squirrel but a canoe a wigwam: A category-specific deficit for artifactual objects and body parts. Cognitive Neuropsychology, 9, 73–86. Snodgrass, J.G., & Vanderwart, M. (1980). A standardised set of 260 pictures: Norms for name agreement, image agreement, familiarity and visual complexity. Journal of Experimental Psychology: Human Perception and Performance, 6, 174–215. Stewart, F., Parkin, A., & Hunkin, N. (1992). Naming impairments following recovery from herpes simplex encephalitis: Category-specific? Quarterly Journal of Experimental Psychology, 44, 261–284. Warrington, E., & McCarthy, R. (1987). Categories of knowledge: Further fractionations and an attempted integration. Brain, 111, 1273–1296. Young, A.W., & de Haan, E.H.F. (1988). Boundaries of covert recognition in prosopagnosia. Cognitive Neuropsychology, 5, 317–336.