خانواده ای در معرض خطر: پروزوپاگنوزیا مادرزادی، تشخیص چهره فقیر و نقص های دیداری ادراکی در یک خانواده

Abstract Congenital prosopagnosia (CP) describes a severe face processing impairment despite intact early vision and in the absence of overt brain damage. CP is assumed to be present from birth and often transmitted within families. Previous studies reported conflicting findings regarding associated deficits in nonface visuoperceptual tasks. However, diagnostic criteria for CP significantly differed between studies, impeding conclusions on the heterogeneity of the impairment. Following current suggestions for clinical diagnoses of CP, we administered standardized tests for face processing, a self-report questionnaire and general visual processing tests to an extended family (N=28), in which many members reported difficulties with face recognition. This allowed us to assess the degree of heterogeneity of the deficit within a large sample of suspected CPs of similar genetic and environmental background. (a) We found evidence for a severe face processing deficit but intact nonface visuoperceptual skills in three family members – a father and his two sons – who fulfilled conservative criteria for a CP diagnosis on standardized tests and a self-report questionnaire, thus corroborating findings of familial transmissions of CP. (b) Face processing performance of the remaining family members was also significantly below the mean of the general population, suggesting that face processing impairments are transmitted as a continuous trait rather than in a dichotomous all-or-nothing fashion. (c) Self-rating scores of face recognition showed acceptable correlations with standardized tests, suggesting this method as a viable screening procedure for CP diagnoses. (d) Finally, some family members revealed severe impairments in general visual processing and nonface visual memory tasks either in conjunction with face perception deficits or as an isolated impairment. This finding may indicate an elevated risk for more general visuoperceptual deficits in families with prosopagnosic members.

. Introduction Individual recognition of familiar faces is one of the most important and demanding abilities for humans in social life (e.g., Farah et al., 1998 and Young et al., 2008). The very high performance in this skill is assumed to be subserved by cortical networks specialized on the processing of faces (e.g., Haxby et al., 2000 and Kanwisher et al., 1997). Lesions within these cortical networks can lead to a state in which patients are dramatically impaired in recognizing faces, despite normal lower-level vision, object identification skills, and semantic knowledge. This severe neurological impairment has been called prosopagnosia or face blindness and has attracted a lot of interest in the last decades, both in the scientific community and in the general population. Prosopagnosia provides evidence that face processing is a cognitive function that may be dissociated from general visual processing or object processing (e.g., Farah, 1996 and Moscovitch et al., 1997). 1.1. Congenital prosopagnosia Individuals with an isolated face recognition deficit, which manifests itself in early childhood but is not attributable to overt neurological, neuropsychological, or psychiatric abnormalities, have been categorized as congenital, developmental, or hereditary prosopagnosics (e.g., Behrmann and Avidan, 2005, Duchaine and Nakayama, 2006b, Jones and Tranel, 2001, Kennerknecht et al., 2006 and Kress and Daum, 2003). In line with our earlier publications, we will use the term congenital prosopagnosia (CP) to emphasize its presumed presence from birth and/or its hereditary origin. A significant number of case descriptions, but also group- and family studies on this condition have been published in recent years (e.g., Dobel et al., 2007, Duchaine et al., 2007a, Grueter et al., 2007, Kennerknecht et al., 2008a, Kennerknecht et al., 2008b, Kennerknecht et al., 2007 and Schmalzl et al., 2008). Independent research groups have estimated the prevalence of CP at 2–3% in the general population (Bowles et al., 2009, Kennerknecht et al., 2006 and Kennerknecht et al., 2008a). However, it is not clear whether these findings indicate a dichotomous, bimodal distribution of face processing skills in the population, or whether CPs reflect the lower end of a normal distribution, which would imply a continuous representation of face processing skills in the general population. Recent evidence of so-called super recognizers who perform approximately 2 SDs above the mean of the general population, but also population-based assessments with sensitive behavioral tests point towards the latter interpretation of the results ( Kennerknecht et al., 2011, Russell et al., 2009 and Wilmer et al., 2012). Given that many CPs report on first-degree relatives who are also impaired in face recognition, most researchers argue for a hereditary contribution to CP (Behrmann and Avidan, 2005, De Haan, 1999, Dobel et al., 2007, Galaburda and Duchaine, 2003 and Kennerknecht et al., 2008b). Support for a genetic contribution to face recognition skills in the general population arises from studies in behavioral genetics (Wilmer et al., 2010 and Zhu et al., 2010). These studies have compared performance of mono- and dizygotic twins on different tests of face cognition and have estimated the specific impact of genetic variation in face recognition to be as high as 39%. Such findings imply that CP and face recognition deficits more generally may be predominantly found in certain families. Regarding the underlying cognitive mechanism of the impairment, there is some evidence that persons with CP display abnormalities in what is called configural or holistic processing of faces (e.g, Avidan et al., 2011, Palermo et al., 2011, Robbins and McKone, 2007, Tanaka and Farah, 1993 and Van Belle et al., 2010). Usually people perceive an upright face as an indecomposable whole, despite its constitution of individual features with complex spatial relations to each other ( Maurer, Grand, & Mondloch, 2002). Behavioral evidence for this special cognitive treatment of upright faces comes most prominently, from the face inversion effect ( Yin, 1969). The face inversion effect describes reduced recognition rates for faces that are presented upside down (i.e., inverted) compared to upright faces. This disproportion is considerably larger for faces compared with other objects. Supposedly, the inversion of a face as well as a misalignment of the bottom half (composite face effect, e.g., Young, Hellawell, & Hay, 1987) interferes with the interactive processing of its parts, leading to a feature-based, analytic encoding strategy, which is less efficient than a holistic approach regarding accurate and fast recognition. In prosopagnosic subjects however, these usually robust behavioral effects are often not found; in fact, many described cases even display better recognition rates for inverted faces ( Avidan et al., 2011, Busigny and Rossion, 2011, Dobel et al., 2008, Duchaine et al., 2007b, Farah et al., 1995, Lee et al., 2010 and Schmalzl et al., 2009). 1.2. Heterogeneity of CP as a clinical condition Whereas an impairment of face recognition is by definition at the core of CP, evidence on associated neuropsychological deficits in persons classified as CP is scattered and often conflicting, suggesting that CP may be a heterogeneous clinical condition (Dobel et al., 2007, Le Grand et al., 2006 and Schmalzl et al., 2008). Among the reported perceptual deficits in nonface visual domains are intraclass object agnosia (i.e., difficulties in discriminating between members of other semantic categories such as houses or cars; Behrmann et al., 2005 and Duchaine et al., 2007b), impaired perception of biological motion including lip reading (Dobel et al., 2007 and Lange et al., 2009), and visual imagery deficits (Tree & Wilkie, 2010). In other cases, however, the face recognition deficit was reported to be isolated, or at least not associated with deficits in object processing or domain-general visual abilities (e.g., Duchaine and Nakayama, 2005 and Stollhoff et al., 2011). One cause of this heterogeneity regarding associated deficits may be genuine diversity in the investigated cases themselves (i.e., subjects may present with various subtypes of CP depending on genetic and/or environmental factors). A second cause for the reported heterogeneity may be a poor comparability between the employed methods and diagnostic criteria to classify individuals as CP. Such methods range from self-reports (e.g., Dinkelacker et al., 2010, Grueter et al., 2007 and Kennerknecht et al., 2006) to interpreting significant group differences in tailor-made experimental tasks (e.g., Behrmann et al., 2005, Dobel et al., 2007, Duchaine and Nakayama, 2005 and Le Grand et al., 2006). Moreover, among the latter types of studies the tested domains and neuropsychological functions vary considerably in their level of specificity and difficulty, ranging from low-level face perception of gender or emotion to highly abstract and difficult tasks on nonface perceptual organization. From a practitioner׳s point of view, basic research on CP has suffered from a lack of consensus on clear criteria both for diagnosis and for exclusion (Gainotti, 2010 and Herzmann and Danthiir, 2008). However, the situation has been improved by Bowles et al. (2009), who suggested consensual clinical diagnostic criteria based on neuropsychological face processing tests which provide normative data, cut-off scores and a high level of psychometric quality. With this study, we attempt to contribute to the current debate on heterogeneity of CP as a clinical entity. We analyzed the patterns of performance on face processing tasks and general visual processing tasks across a large family sample (N=28) in which many members reported face recognition difficulties. The studied sample is highly similar with regard to genetic and environmental factors, especially within core families (e.g., a father and his offspring). This high group homogeneity allows us to largely control for the impact of genetic and environmental variability on performance. Previous work on CP within family samples concluded that CP is a primarily heterogeneous condition regarding associated neuropsychological deficits and/or underlying cognitive functions ( Lee et al., 2010 and Schmalzl et al., 2008). We were now interested in whether the use of the recently suggested standardized diagnostic criteria and clinical cut-off scores ( Bowles et al., 2009) might yield a more homogeneous picture of the condition within families by reducing the possibility of falsely diagnosing CP e.g., in ambiguous cases. For single-case diagnoses, such a normative account has several advantages over a group comparison of tailor-made experimental tasks. First of all, normative samples are usually larger, allowing for a more precise quantitative classification of the results compared to smaller control group samples. Second, standardized tasks have usually been tested for their psychometric quality. Most importantly, published and standardized neuropsychological tests provide a basis for reproducible results and comparisons between studies and thus constitute the standard procedure in clinical settings. With this normative account, we furthermore aimed to contribute to the current debates on whether CP is a categorical or a continuous phenomenon as well as comment on the relevance and practicability of self-reports in CP research. 1.3. Diagnostic criteria for CP As recommended by Bowles et al. (2009), we employed two tests of face processing abilities for the diagnosis of CP: The Cambridge Face Memory Test (CFMT; Duchaine & Nakayama, 2006b) and the Cambridge Face Perception Test (CFPT; Duchaine et al., 2007b). These computer-based tests have overcome various shortcomings of older tests, such as feature matching of facial and nonfacial information (Duchaine and Weidenfeld, 2003 and Duchaine and Nakayama, 2004) by using unfamiliar natural faces without hairlines as stimuli. Recent studies from different laboratories on the psychometric qualities of the CFMT establish this test as a reliable instrument for assessing face memory and even its subtle impairments on an individual basis (Bowles et al., 2009, Herzmann and Danthiir, 2008 and Russell et al., 2009). The CFPT moreover provides a measure of a perceptual face inversion effect (i.e., the hallmark of holistic face processing) by contrasting performance in trials with upright and inverted faces ( Duchaine et al., 2007b). Despite obvious differences between the two tasks, as well as recent evidence for dissociations between face perception and face memory abilities ( Wilhelm et al., 2010), the CFMT and CFPT are strongly correlated, and the authors of the tests suggest using both for a diagnosis of CP. In line with Bowles et al. (2009), we consider a subject as clinically impaired if performance in both the CFMT and the CFPT is lower than 2 SDs below the predicted mean for the person׳s age. We also employed a standardized self-report questionnaire consisting of 15 items on everyday face processing skills, which has been used in previous studies ( Kennerknecht et al., 2007 and Kennerknecht et al., 2008a). The aim of this self-report questionnaire was to test the predicitve validity of such a quick-and-dirty screening method and compare it to the diagnoses based on established neuropsychological tests. The major advantage of a screening for prosopagnosic symptoms lies in its potential economic efficiency: Whereas in-depth neuropsychological testing may ascertain a diagnosis, short screening questionnaires may be used to preselect subjects that report deficits. This can be done in a fast and efficient way, e.g. via telephone or online. Such procedures have already proven to be useful for a range of other neurocognitive syndromes including early dementia and developmental dyslexia (e.g. Ramlall et al., 2013 and Snowling et al., 2012). Given the lack of clear guidelines for exclusion criteria, we decided on a pragmatic procedure to screen for possible impairments that may result in face processing dysfunctions which cannot be attributed to CP. Since face recognition and/or face perception deficits are assumed to be common symptoms in several developmental (e.g., autism spectrum disorders; Weigelt et al., 2012 and Wilson et al., 2010) and neurodegenerative disorders (e.g., frontotemporal dementia; Omar, Rohrer, Hailstone, & Warren, 2011), we specifically asked subjects for any birth complications, developmental abnormalities as well as their neurological and psychiatric history. To screen for major impairments in basic visual integrity and visuoperceptual abilities which may impede a reasonable evaluation of face processing abilities, we employed two widely used neuropsychological assessment tools. First, we administered the Visual Object and Space Perception Battery (VOSP; Warrington & James, 1992) to preclude the possibility of a general visual agnosia as a cause for prosopagnosic symptoms. Following diagnostic recommendations by Von Cramon, Mai, and Ziegler (1995), the VOSP combines the most important tests for the diagnosis of visual agnosias. Second, we employed the Rey–Osterrieth Complex Figure Test (RCFT; Meyers & Meyers, 1995) to screen for impairments in perceptual organization and in figural memory for nonface material. Analogously to the face processing tests, we only regard major impairments (i.e., an age-corrected score more than 2SDs below the norm) in either the RCFT or the VOSP as an exclusion criterion for a CP classification. 1.4. Goals of the current study In this study we aim to investigate familial patterns of face and nonface processing deficits in an extended family sample of suspected CPs. We focus on standardized neuropsychological assessment and the use of operational diagnostic criteria, in order to reduce the risk of falsely diagnosing CP. We are specifically interested in the following questions: (a) Will we find similar neuropsychological profiles for tests of face and nonface visual processing in close relatives suggestive of CP? This question relates to the aspect of homogeneity vs. heterogeneity of CP within a family sample. (b) Will those family members who are not suspected of having CP perform normally in face processing tasks compared to the general population? This addresses the question of whether deficient face processing is transmitted in a dichotomous all-or-nothing or in a continuous fashion. (c) How are scores on standardized face processing tests reflected in self-report data? This addresses the question whether self-report data can be used as a viable screening instrument to classify persons with face processing deficits.

. Results Table 1 summarizes the results of all 28 subjects on the 12 neuropsychological measures and the self-report. Table 1. Individual z scores for all test variables. Full-size image (97 K) Note. z Scores for all 28 subjects on neuropsychological tests and self-rating for face processing. Subjects of the younger generation were indented to the right, beneath their parent. Cell colors represent severity of deficits. Black cells represent values more than 2 SDs lower than the mean of the normative sample. Dark grey cells indicate values between −1.5 and −2.0. White cells represent scores within or above average (i.e., z>−1.5). Rows below the table contain mean values for each test and whether these significantly deviate from the normative sample (expressed as t-, and corresponding p-values, *p<.05, **p<.01). CFMT: Cambridge Face Memory Test; CFPT: Cambridge Face Perception Test; RCFT: Rey-Complex Figure Test; VOSP: Visual Object And Space Battery. Table options 3.1. Performance in face processing tasks and CP classifications z Scores for the CFMT as well as the upright version of the CFPT and the perceptual face inversion effect measure are shown in the first three columns of Table 1. Three first-degree relatives – a father and his two sons (M46, M16b, M13) – each scored lower than z=−2 on both the CFMT and the upright CFPT, indicating a clear CP classification. Additionally, these three subjects displayed z values around −2 on the CFPT inversion effect measure, indicating that they did not show the typical perceptual face inversion effect. In fact, for M13 the face inversion effect was itself inverted—he made more errors on the upright than on the inverted trials of the CFPT (raw scores: CFPT upright=64; CFPT inverted=58). None of these three subjects showed significant deficits on any measures of the VOSP or the RCFT, with all z scores higher (i.e., better performance) than −2. M16b displayed a single marginal z score of −1.63 in the object identification subtest of the VOSP. Table 2 furthermore demonstrates that all three CPs display significant differences between face and nonface tasks, indicating that the observed discrepancies in these subjects are highly unlikely to be merely due to measurement errors of the employed tasks. Table 2. Tests of significance between individual test scores in face and nonface tasks. Full-size image (60 K) Note. Values represent z scores for testing the significance of differences between two individual test scores ( Payne & Jones, 1957). Values larger than z=±1.96 indicate, that the observed difference is highly unlikely to be due to measurement errors only (p<.05). Algebraic sign of the z score indicates the direction of the difference. For columns 1–4 thus, black cells represent a discrepancy of face over nonface tasks. Grey cells indicate a discrepancy in the opposite direction (nonface>face tasks). 1We included this analysis to substantiate our observation of partly dissociable face processing abilities in several individuals. For the comparison CFMT vs. CFPT upright, black cells indicate a significant discrepancy of CFMT over CFPT performance. Grey cells contrarily indicate a significantly worse performance in the CFPT compared to the CFMT. Table options Besides this homogeneous group of three first-degree relatives with an impairment in face processing, another four subjects (M23, M16a, M27, M21; note that M27 and M21 are brothers) displayed a very similar pattern of results on face tasks. They all showed impaired or at least marginally impaired performance in the CFMT and in at least one measure of the CFPT, though without reaching the conservative diagnosis cut-off of z=−2.00 on both tests. Of these four subjects, and in contrast to the CP triplet, M21 and M23 also revealed deficient scores on two trials of the RCFT, pointing towards deficits in figural memory functions and general perceptual organization skills. We will refer to this group of subjects as bad recognizers. Finally, a pattern of normal face recognition (CFMT) but deficient facial similarity judgment abilities (CFPT upright) was found in a number of subjects (F48, F23, M50, M49, M18; Table 1). For two of these subjects (F48, M18), this discrepancy between CFMT and CFPT upright was found to be statistically significant (Table 2, column 5). While these subjects are certainly somewhat limited in their ability to perceptually judge similarities of faces, their face recognition ability – the core feature of CP – is spared. This important issue of partly dissociated skills underlying the CFMT and CFPT will again be addressed in a correlational analysis in Section 3.5. Given that all three classified CP and all of the bad recognizers were males, we have further investigated the gender distribution of face recognition deficits in our data despite the fact that we had no a priori hypotheses concerning this aspect.2 The percentage of subjects that were classified as CP did not significantly differ by gender, χ2(1, N=28)=1.86, p=.17. However there was a significant gender effect when CP and bad recognizers were both considered as impaired, χ2(1, N=28)=5.18, p=.03 suggesting that male subjects were more likely to be classified as CP or bad recognizers in our sample. 3.2. Face processing of the family sample as a group We next compared the family׳s group mean against the mean of the general population, taken from available normative data (Bowles et al., 2009). As can be seen in the bottom row of Table 1, the result indicates a significantly lower group mean for both face processing tests, compared with the general population mean—CFMT (M=−.90, SD=.87), t(27)=−5.51, p<.001; CFPT upright (M=−1.20, SD=1.30), t(27)=−4.94, p<.001). When the three CPs (M46, M16b, M13) as well as the group of bad recognizers who failed to reach all diagnostic criteria (M23, M16a, M27, M21) were taken out of this analysis, the group means of the remaining subjects on the CFMT and the CFPT upright still remained significantly below average—CFMT, (M=−.55, SD=.67), t(20)=−3.70, p<.001; CFPT upright, (M=−.67, SD=1.00), t(20)=−3.06, p=.006. In order to rule out the possibility that these remaining subjects may suffer from more general cognitive impairments, which globally affect their performance on both face as well as nonface tasks, we furthermore directly compared mean performance on face tasks vs. nonface tasks in this group. Face tasks (CFMT, CFPT upright) and nonface tasks respectively were aggregated by summing up z scores and dividing by the number of tasks. A paired t-test revealed that for the group of subjects who were neither classified as CP nor as bad recognizers (i.e., remaining subjects), scores on the aggregated face tasks were significantly lower than on the aggregated nonface tasks, t(20)=3.13, p=.005. Thus, the family as a whole displayed a tendency to perform below average on both face processing tasks, even without those individuals classified as CP or as bad recognizers. 3.3. Comparison of self-report data and results in face processing tests In order to compare the self-report data with the neuropsychological test data for face processing, we adopted a twofold strategy—a single-subject and a group analysis. Regarding individual self-reports of the classified CPs, subjects M46 and M16b consider their ability to recognize familiar persons as severely impaired, with z scores of −2.12 and −3.4, respectively. This roughly matches their individual performances in the CFMT and CFPT. M13 judges himself less impaired, but his self-rating is also marginally impaired, z=−1.58. Of the four subjects who failed to reach all diagnostic criteria and were thus classified as bad recognizers, only M27 rates his own face processing abilities as marginally impaired, z=−1.58. In the remaining three subjects the somewhat conspicuous findings in the CFMT and CFPT are not clearly reflected in their self-reports. For the rest of the family, the self-reports vary considerably, with values ranging from z=.79 (M18) to z=−4.31 (M52). Just as with the neuropsychological test data, the mean self-rating of the family as a group is significantly lower than in the available normative sample, (M=−1.07, SD=1.30), t(27)=−4.53, p<.001. Also analogously to the results in standardized test data, this even holds true when the three CPs and the four subjects, which we categorized as bad recognizers are excluded from this analysis, (M=−.94, SD=1.4), t(21)=−3.07, p=.006). Thus, the general tendency of this family to perform below average in face processing tests is mirrored in self-report data. Correlations between tests for face processing and the self-reports are displayed in Table 3. The CFMT is moderately correlated with self-reported impairment, r(26)=−.5, p=.007, suggesting that performance in the CFMT is at least partly reflected in the self-report data. Considering the generally reduced means on the face processing measures and the consequent restriction of variance in this family, the strength of this correlation is impressive. Although we were able to replicate the finding by Bowles et al. (2009) of a strong association between the CFMT and the upright CFPT, r(26)=.57, p=.002, neither the perceptual face inversion effect measure from the CFPT, nor the upright CFPT is reflected in the self-report data, indicated by low non-significant correlations. The fact that the items in our self-report battery ( Kennerknecht et al., 2007) mainly focus on face recognition abilities and not on perceptual facial similarity judgement, may explain this result. We furthermore calculated a receiver operating characteristic (ROC) curve to estimate sensitivity and specificity as well as the optimal cut-off score for the self-report questionnaire given the CP diagnoses based on the neuropsychological tests. The self-report questionnaire used in the current study shows an area under the curve (AUC) of .83 (95% confidence interval: .66–.99) for the distinction between CP and non-CP subjects. The optimal cut-off score for the self-report questionnaire as a first screening procedure was 34 in the current sample with a sensitivity of 100% and a specificity of 72%. Thus, with this score it is possible to identify all CPs with the self-report questionnaire, however at the expense of a substantial rate of false positives. These false positives are then classified by using neuropsychological tests. Table 3. Intercorrelations of face processing measures. Measure CFMT total CFPT upright errors CFPT inversion effect CFPT upright errors −.57⁎⁎ CFPT inversion effect .30 −.63⁎⁎ Self reported deficits −.50⁎⁎ .20 −.05 Note. Results of correlational analyses with data from all 28 subjects. Note that for the CFPT upright and the self report, higher scores indicate greater impairment. ⁎p<.05. ⁎⁎p<.01. Table options In summary, impaired face processing skills and self-reported impairment show a considerable overlap in this family. All three classified CPs reported at least subtle impairments in recognizing faces. Although in some subjects the results diverged strongly between assessed face recognition and self-report (i.e., M49, M52), the high correlation between CFMT and self-reported face recognition deficits as well as the accurate classification parameters indicated by the ROC analysis suggests that self-report data can be a useful screening tool for CP classifications. 3.4. General visuoperceptual deficits Even though we had no a priori hypotheses about general visuoperceptual deficits in the studied family, an inspection of Table 1 (columns 5–12) necessitates a more detailed analysis. Of the 28 tested subjects, four individuals displayed pronounced deficits on at least one subtest of the object part of the VOSP, indicated by a z score below −2. Their scores suggest difficulties with basic-level object identification. The most pronounced deficits were seen on the subtest Object Decision, with a z score as low as −3.5 for one subject (F48). On this subtest, subjects were asked to select the silhouette of a real object out of an array that included three distractor items (nonsense objects). At the group level, the means in the subtests Object Decision and (to a lesser extent) Silhouettes both revealed performances significantly below average (see Table 1). Thus, as with face processing, the family as a whole displayed a general impairment in basic-level object identification skills. Regarding spatial perception (subtests Number Location and Cube Analysis; Table 1, column 11 and 12), three subjects displayed deficits on the subtest Number Location. However, the group means for the two subtests of the space perception part of the VOSP were either average (Number Location) or even above average (Cube Analysis). Although the group means for the four RCFT measures were not significantly impaired, the profiles of M28, F23, M21, M50, F19b, and again F48 revealed major deficits in perceptual organization skills and figural memory, indicated by z scores below −2 on at least one measure of the RCFT ( Table 1). Whereas in some of these profiles, poor results appeared in isolation (both, F19b and M50 scored well within normal ranges on three out of the four RCFT measures), three other subjects displayed consistently abnormal results on the RCFT (F48, F23, M28), although they were unimpaired in face recognition as measured by the CFMT. A clinical interpretation of these values would suggest a moderately-to-severely impaired memory for complex visual stimuli ( Meyers & Meyers, 1995). Fig. 2 exemplary shows the results of subject F48 on three measures of the RCFT and gives an impression of how items seem to be recalled in an unsuccessful, feature-based manner that misses the overall organization of the figure. z Scores for testing significance of test differences moreover indicate that F48 shows a significant discrepancy between face recognition as measured by the CFMT and nonface tasks. Also for F23 and M28, the difference between CFMT and RCFT immediate recall score is highly unlikely to be explainable by measurement error ( Table 2). Subject F48׳s performance on the RCFT trials. Fig. 2. Subject F48׳s performance on the RCFT trials. Figure options 3.5. Correlations between general visual processing and face perception skills Given both, the accumulation of these diverse visuoperceptual impairments in the investigated sample, as well as our finding of subjects with impaired scores in the CFPT but not in the CFMT, we further explored possible relationships between the general visual processing tasks and the face processing tasks by means of a correlational analysis. Whereas the CFMT did not share a significant amount of variance with any nonface task in our battery, the CFPT upright was significantly correlated with the immediate recall trial of the RCFT, r(26)=−.41, p=.031, suggesting a relationship between nonface perceptual organization skills and the process of judging perceptual similarities in faces: Subjects with high recall scores on the RCFT generally made fewer errors on the CFPT, and vice versa. Correlations between all three RCFT drawing scores and the inverted face trials of the CFPT (CFPT inverted) were even stronger copy trial, r(26)=−.63, p<.001; immediate recall, r(26)=−.55, p=.003; delayed recall, r(26)=−.50, p=.006. This further supports the interpretation of a connection between general perceptual organization skills and performance in the CFPT, independent of whether the faces that are compared in the CFPT are upright or inverted. No other correlations between face- and nonface tests reached significance