پروزوپاگنوزیا تکاملی: مطالعه سه بیمار

Abstract We studied perception in three patients with prosopagnosia of childhood onset. All had trouble with other ‘within-category’ judgments. All were deficient on face matching tests and severely impaired on tests of perception of the spatial relations of facial features and abstract designs, indicating a deficit in the encoding of coordinate relationships, similar to adult-onset prosopagnosia with lesions of the fusiform face area. Two had difficulty perceiving feature colour, which correlated with reduced luminance sensitivity. In contrast to adult-onset patients, saturation discrimination was spared in two and spatial resolution impaired in two. Curvature discrimination was relatively spared. Contrast sensitivity showed variable reductions at different spatial frequencies. We conclude that developmental prosopagnosia is similar to the adult-onset form in encoding deficits for the spatial arrangement of facial elements. Deficits in luminance perception and spatial resolution are more associated with defective encoding for basic object-level recognition, as shown on tests of object and spatial perception.

1. Introduction Prosopagnosia is the inability to recognize familiar faces in the absence of more generalized cognitive dysfunction. While some degree of visual (de Haan, Young, & Newcombe, 1987) or memory disturbance (Bauer & Verfaellie, 1988; Bruyer et al., 1983; de Haan et al., 1987; Malone, Morris, Kay, & Levin, 1982; Rizzo, Hurtig, & Damasio, 1987) for stimuli other than faces can be present, the recognition deficit is more severe and sometimes highly specific for faces. Most cases of this unusual problem are acquired through lesions of the right (de Renzi, 1986; Landis, Cummings, Christen, Bogen, & Imbof, 1986; Takahashi, Kawamura, Hirayama, Shiota, & Isono, 1995) or both(Damasio, Damasio, & van Hoessen, 1982; Meadows, 1974) occipitotemporal cortices, involving regions ranging from the anterior temporal lobe to the fusiform and lingual gyri. Just as normal face recognition is thought to involve a series of processing stages, acquired prosopagnosia is considered now to have several functional subtypes (Damasio, 1985; Takahashi et al., 1995). The broadest and most frequently used division is into two categories (Damasio, Tranel, & Damasio, 1990; de Renzi, Faglioni, Grossi, & Nichelli, 1991): apperceptive prosopagnosia, in which impaired formation of the facial percept prevents recognition, and associative prosopagnosia, in which an adequately formed facial percept cannot be compared with stores of facial memories. Though rare, there are numerous well-studied cases of acquired adult-onset prosopagnosia in the literature. Developmental or childhood-onset prosopagnosia, however, has not been reported as often. The first case of this developmental form, AB, was described in 1976 (McConachie, 1976) and in later reports (Campbell, 1992; de Haan & Campbell, 1991). Since then, at least seven additional patients have been reported in single case studies. These cases are summarized in Table 1. Table 1. Patients with developmental prosopagnosia in the literature Reference LG KD AB HD YT EP BC S AV Ariel and Sadeh, 1996 Young et al., 1989 Campbell, 1992 Kracke, 1994 Bentin, Deouell, and Soroker, 1999 Nunn et al., 2001 Duchaine, 2000 Temple, 1992 de Gelder, 2000 Origin localization EEG imaging Bi-occipital Meningitis Familial R posterior Familial Familial Familial Hydrocephalus n “Small” n n IQ Verbal 142 144 110 127 132 136 Performance 90 100 105 113 135 147 Other abilities Reading Mildly abn n Superior Superior Colour “n” abn n n n Topographagnosia Yes Yes No No No Yes Face perception Age abn abn abn n n n Gender abn abn abn n n n Expression abn n abn abn n n BFRT “Poor” 37, slow 39 n, slow 41 46, slow 43, slow n 34 Warrington face 28 24 32 41 46 43 34 Covert recognition Absent Absent Absent Object perception Photographs abn abn n, slow n Drawings abn abn n, slow n n n n Unusual views abn abn abn n n n n Overlapping figures abn n, slow n n n Incomplete figures n n n n “Within category” abn abn n abn Table options Although the numbers are few, comparisons with adult-onset prosopagnosia reveal some interesting points. The majority of the developmental cases are thought to have deficits at the level of structural encoding of the facial percept. Supportive evidence for this stems mainly from impaired performance on the matching of unfamiliar faces (the Benton Face Recognition Test, or BFRT), poor judgments of facial affect, gender or age (Ariel & Sadeh, 1996; de Haan & Campbell, 1991; Kracke, 1994; Young & Ellis, 1989), and deficient recognition of objects under more demanding conditions, as in line drawings, unusual views, and overlapping or incomplete figures (Ariel & Sadeh, 1996; de Haan & Campbell, 1991; Young & Ellis, 1989). Also, within-category judgments for non-face objects were impaired in at least three patients (de Haan & Campbell, 1991; Temple, 1992; Young & Ellis, 1989), suggesting that the defect was not as face-specific as reported for some adult-onset prosopagnosic patients (Farah, Levinson, & Klein, 1995; McNeil & Warrington, 1993). Despite these parallels with adult-onset prosopagnosia, imaging in the few developmental cases in which this was obtained has not shown lesions of occipitotemporal cortex similar to those in the adult-onset form. Also, other deficits commonly associated with occipitotemporal lesions are unusual in childhood prosopagnosia. Only one patient had dyschromatopsia and superior altitudinal field defects (Young & Ellis, 1989), and only two had topographagnosia (de Haan & Campbell, 1991; Young & Ellis, 1989) while two others did not (Kracke, 1994; Temple, 1992). Residual covert or unconscious knowledge of familiar faces has not been found in any developmental prosopagnosic patients tested (Bentin et al., 1999; de Haan & Campbell, 1991; Young & Ellis, 1989), though it has been shown in several adult-onset cases (Bruyer, 1991; Young, 1994). The paucity of other functional markers or imaging evidence of occipitotemporal damage raise the question of whether the type of perceptual encoding deficits in developmental prosopagnosia are similar to those in adult-onset prosopagnosia. It is not always the case that developmental disorders share the same pathophysiology as their adult counterparts. For example, childhood-onset dyslexia and the acquired pure alexia of left occipital lesions are considered different disorders. In adult-onset prosopagnosia, the type of perceptual encoding dysfunction continues to be debated. We have recently found that patients with lesions of the fusiform face area are severely impaired in their ability to discern the spatial relations of facial features (Barton, Press, Keenan, & O’Connor, 2002). These spatial relations are important in distinguishing the subtle variations in the basic configuration of features that all faces share. An orientation-specific expertise with perceiving these spatial relations likely accounts for the difficulty normal subjects have in recognizing upside-down faces—the ‘inversion effect’ (Barton, Keenan, & Bass, 2001b; Cooper & Wojan, 2000; Freire, Lee, & Symons, 2000; Leder & Bruce, 2000). Apart from the indirect evidence noted above, the nature of perceptual encoding has rarely been studied in detail in developmental prosopagnosia. One patient, GA, was found to have a deficit in curvature perception (Kosslyn, Hamilton, & Bernstein, 1995). It was speculated that this reflected selective anoxic damage to end-stopped cells in primary visual cortex. Our study of seven adult-onset prosopagnosic patients (Barton, Cherkasova, Press, Intriligator, & O’Connor, 2003) found normal curvature perception in all but one, suggesting that curvature misperception seldom contributes to defective perceptual encoding in adult cases with defined extra-striate damage. In this report we describe studies on GA and an additional two patients with developmental prosopagnosia. The evidence suggests that all three have problems with perceptual encoding of faces. We conducted a detailed review of many perceptual functions, using the same tests employed in our sample of seven adult-onset prosopagnosic patients with localizable lesions on imaging. These tests probed perception of not only spatial relations but also a range of other basic functions of potential importance to encoding complex three-dimensional structures, such as luminance, contrast, saturation, spatial resolution, and curvature. Our goal was to determine whether the pattern of deficits in these developmental patients was similar to that in adult-onset prosopagnosia, or if it indicated a very different type of apperception.

3. Results (Table 2) 3.1. Characterization of prosopagnosia GA and KBN were in the low normal range (11th percentile) on the BFRT, whereas KT was severely impaired (<98% limit). All scored below the 95% limit on the Warrington Face recognition test. On tests of covert recognition reported elsewhere (Barton et al., 2001a), none of these subjects showed any residual unconscious access to facial identity. 3.1.1. Discrimination of facial features and spatial configuration All three patients were significantly impaired in judgments of the spatial position of features, in both time-limited and unlimited duration trials (Fig. 3). Neither KBN nor KT were able to reach threshold (67% correct) with even the largest shifts of feature position. GA was able to achieve a borderline normal score for mouth position with unlimited duration, but at the cost of very prolonged reaction times (Fig. 4). Accuracy for discrimination of facial feature colour and spatial configuration. ... Fig. 3. Accuracy for discrimination of facial feature colour and spatial configuration. Data for normal controls are shown in the top graphs, patients in the bottom graphs. Left graphs are for eye brightness changes, middle graphs are for eye position shifts, and right graphs for mouth position shifts. Viewing durations are: 2s, 2 seconds; 4s, 4 seconds; unlim, unlimited time. Figure options Reaction times for discrimination of facial feature colour and spatial ... Fig. 4. Reaction times for discrimination of facial feature colour and spatial configuration, from blocks with unlimited viewing duration. Data for eye colour in left graph, eye position in center graph, and mouth position in right graph. Grey lines indicate means with one standard deviation as error bars, for the normal sample. All three patients take over 10× as long as normal subjects to respond. Figure options For eye colour (brightness), KBN and KT were equally impaired. GA, on the other hand, while impaired on the time-limited trials of 2 and 4 s, scored near the normal mean with unlimited time, with reaction times that were only slightly prolonged compared to normal subjects, and much more rapid than his reaction times for spatial judgments. When subjects could restrict their attention to changes in mouth position, GA improved his performance dramatically to the low normal range, but KBN and KT did not (Fig. 5). Accuracy for discrimination of mouth position shifts. Left graph shows data from ... Fig. 5. Accuracy for discrimination of mouth position shifts. Left graph shows data from trials illustrated in Fig. 3, where any of the three facial changes are equally likely. Right graph shows data from the block with only changes to mouth position. Solid and dashed grey lines show normal data. With the focus narrowed to mouth position, GA can perform normally, but KT and KBN cannot. Figure options 3.1.2. Within-category recognition of fruits and vegetables Of the eight normal subjects, two made 3 errors and six made none out of 36 items. All three patients were abnormal. GA made 7 errors, KBN 19, and KT 11. 3.2. Non-face perceptual functions 3.2.1. Contrast sensitivity The contrast sensitivity function was different for each patient. GA had only a slight reduction at the highest spatial frequency tested, whereas KT showed a notch for the middle spatial frequency of 4.1 cycles/°. KBN was the most severely affected at the frequencies of 4.1 and 8.2 cycles/° (Fig. 6). Contrast sensitivity. This is shown for four different spatial frequencies. Mean ... Fig. 6. Contrast sensitivity. This is shown for four different spatial frequencies. Mean (dotted grey line) and lower 95% prediction limit (solid grey line) of normative data are shown. No consistent pattern is found. KBN is most severely impaired, in mid to high spatial frequencies, GA is mildly impaired at 8.2 cpd, while KT shows the converse, impairment at 4.1 but not 8.2 cpd. Figure options 3.2.2. Luminance discrimination GA had normal luminance discrimination, but KBN and KT were markedly impaired. This finding is consistent with the pattern of performance in these three patients on the test of perception of eye brightness changes (Fig. 7). Thresholds (75% correct) for brightness and saturation discrimination. Left ... Fig. 7. Thresholds (75% correct) for brightness and saturation discrimination. Left graph shows data for luminance (brightness) and the right graph data for saturation. Thresholds for left hemifield, central vision, and right hemifield are given. The top of dark grey bar indicates the mean of control subjects, and the top of the light grey bar is the upper 95% prediction limit. Asterisks indicate default values for impairments so severe that thresholds exceeded the testing range—defaults are 70% for luminance and 26% for saturation. Inset shows a schematic representation of a test stimulus. Figure options 3.2.3. Saturation discrimination This was normal for GA and for KBN, except for a mild deficit in her left hemifield. KT was impaired in all locations tested (Fig. 7). 3.2.4. Spatial resolution (Landolt C) GA, who had a Snellen visual acuity of 20/20 in the clinic, was only mildly impaired in his peripheral field, and low-normal centrally. KBN and KT were significantly deficient (Fig. 8). Thresholds (75% correct) for spatial resolution (Landolt C). Thresholds for left ... Fig. 8. Thresholds (75% correct) for spatial resolution (Landolt C). Thresholds for left hemifield, central vision, and right hemifield are given. The top of dark grey bar indicates the mean of control subjects, and the top of the light grey bar is the upper 95% prediction limit. KBN and KT are impaired, GA mildly so in his peripheral field. Inset shows a schematic representation of a test stimulus. Figure options 3.2.5. Dot displacement discrimination All three subjects were impaired in this task, though KBN was able to achieve one normal value above threshold (Fig. 9). Dot displacement discrimination. Data for horizontal and vertical shifts are ... Fig. 9. Dot displacement discrimination. Data for horizontal and vertical shifts are combined. Mean (dotted grey line) and lower 95% prediction limit (solid grey line) of normative data are shown. Only KBN has one normal measure at values above threshold for normal subjects. Figure options 3.2.6. Curvature discrimination Accuracy on this test was generally good (Fig. 10). Only three points (two for KBN and one for GA) fell outside the 95% prediction intervals, the rest of their points falling in the low-normal range. KT did very well on accuracy, but took a longer time to do so, suggesting that a mild deficit could not be excluded, whereas the reaction times for KBN and GA were well in the normal range. Curvature discrimination. Left graph shows accuracy, right graph shows reaction ... Fig. 10. Curvature discrimination. Left graph shows accuracy, right graph shows reaction time. Mean (dotted grey line) and 95% prediction limit (solid grey line) of normative data are shown. A dot shift of one corresponds to either a 0.2° horizontal decrease in the distance between the upper two of the four dots or a 0.1° decrease in the vertical distance between the lower two dots. GA is normal, KBN borderline low normal. KT has accuracy above the average for normal controls, but at the cost of increased reaction time. Inset shows a schematic representation of a test stimulus.