پردازش فاصله نسبی بین ویژگی های دچار اختلال شده و ناحیه چشم در پروزوپاگنوزیای اکتسابی -دو روی یک سکه جامع نگر؟

Abstract Acquired prosopagnosia (AP) is characterized by impaired recognition of individual faces following brain damage. The nature of the functional impairment(s) underlying AP remains debated. Recent studies have demonstrated deficient processing of diagnostic information in the region of the eyes (Caldara et al., 2005); other studies suggest that patients fail to judge relative distances between facial features (Barton et al., 2002). We hypothesized that these apparently different observations are related to a common cause. More precisely, we suggest that AP arises due to an impairment of a process that reduces uncertainty about the nature/location of the diagnostic cues for face individualization: the ability to perceive multiple elements of a face as a single global representation (holistic processing). Being impaired at processing individual faces holistically, prosopagnosic patients would tend to perform relatively worse for processing facial areas containing multiple elements (i.e., the eyes), and for elements that are widely spaced apart. Here we tested PS, a single case of AP, at matching unfamiliar faces differing either with respect to local features or inter-feature distances, over the upper and lower areas of the face. A pilot study and Experiment 1 confirmed that PS was extremely poor at using information encompassing the eyes, but was also deficient at perceiving relative distances between features. When uncertainty about the location and nature of the diagnostic cue was removed in Experiment 2, PS' performance remained below normal range, but she improved substantially. Most interestingly, her pattern of performance across the different conditions appeared qualitatively identical to that of normal controls. In line with previous observations of PS and other cases of prosopagnosia, our findings indicate that the reduced reliance on the area of the eyes and on relative distances between features in AP may have a common underlying cause—the disruption of holistic processing of the individual face.

1. Introduction Normal face processing involves adequate perception of different cues that are thought to be diagnostic for face individualization. For instance local shape (Young et al., 1985) and surface (color/texture) (Lee and Perrett, 1997 and Russell et al., 2006) information can be derived for this purpose, with the region of the eyes/eyebrows conveying particularly critical sources of information (Gosselin and Schyns, 2001, Haig, 1985 and Sadr et al., 2003). The high efficiency with which we generally perceive and recognize faces masks a complexity which becomes apparent when this ability breaks down, as observed in acquired prosopagnosia (AP) (Bodamer, 1947). This rare neurological condition refers to the selective inability to recognize individual faces as a consequence of brain damage to bilateral or right unilateral occipito–temporal regions. Since the first observations (Quaglino et al., 2003 and Wigan, 1844; for recent reviews see Barton, 2003 and Mayer and Rossion, 2007) the clinical and anatomical conditions of AP have attained considerable notoriety as they provide a means to clarify the neuro-functional mechanisms of normal face processing. However, despite over 60 years of research on AP, the underlying functional basis of the observable deficits in deriving an adequate representation of an individual face remains a matter of debate. It has been proposed that AP involves a deficit in processing the face as a global representation, i.e., configural/holistic processing.1 For instance, based on their assessment of LH, Levine and Calvanio (1989) concluded that prosopagnosia represents a loss of visual “configural processing”, which they conceptualized as a deficit in visual perception, reflected by the inability to derive an “overview of sufficient features to allow structuring or crystallization of a coherent concept” (p. 151). This view has been supported by other studies of acquired prosopagnosic patients that used different paradigms to test the interdependence between facial features of the whole face (e.g., Boutsen and Humphreys, 2002, Saumier et al., 2001 and Sergent and Villemure, 1989). However, the different paradigms used and the variability among patients tested has hindered true significant progress with respect to validation of this hypothesis and thus of our understanding of the nature of this configural/holistic processing view of AP. Furthermore, the fact that different authors conceptualize configural/holistic processing differently (e.g., Farah et al., 1998; Maurer et al., 2002), poses additional problems. More recent studies indicate that prosopagnosia involves a deficit restricted to the processing of certain localized features of the face. Caldara et al. (2005) tested the acquired prosopagnosic patient PS (Rossion et al., 2003) by means of a learning paradigm followed by an identification task of faces revealed through random apertures (“Bubbles”, Gosselin and Schyns, 2001). Compared to normal observers, PS required much more information to achieve the same performance level and relied mostly on the mouth rather than on the eyes. In the same vein, Bukach et al. (2006) showed that the prosopagnosic patient LR was able to detect diagnostic changes in the mouth region, but was strikingly impaired at making such judgments based on the eyes of faces (see also Bukach et al., 2008 and Rossion et al., 2009). Also recently, other authors have reported several patients who were impaired at discriminating faces that differed with respect to distances between features (e.g., mouth-nose distance, inter-ocular distance, …) but could apparently process local features (e.g., eye color) efficiently (Barton et al., 2002, Barton and Cherkasova, 2005 and Joubert et al., 2003). Barton et al. (2002) therefore concluded that the perception of the relative distances between features of faces is impaired in patients with prosopagnosia, in particular when their lesions involve the right fusiform gyrus, and that this deficit contributes directly to their prosopagnosia. These last two hypotheses differ from the proposed holistic/configural hypothesis of AP described above. They suggest that prosopagnosia arises from the inability to process a certain type of information—local information conveyed by the eyes ( Bukach et al., 2006 and Caldara et al., 2005) or the relative distances between facial features in general ( Barton et al., 2002)—rather than from an impaired mode of processing (i.e., holistic, as opposed to analytical). One the one hand, it is tempting to attribute these different observations to the functional variability among acquired prosopagnosic patients (Schweich and Bruyer, 1993 and Sergent and Signoret, 1992), and to acknowledge that the main impairment observed in prosopagnosia—the inability to process faces at the individual level efficiently—has several different manifestations, which would presumably rely on the specific localization of a patient's lesion(s). On the other hand, another way to conceptualize these observations is to integrate all of them into a single theoretical framework. That is, while acknowledging the functional variability among prosopagnosic patients in terms of associated deficits, it may be that all of these patients share a common disrupted process, which characterizes their prosopagnosia. In line with previous studies and our interpretation of the observations made for the patient PS, we hypothesized that the primary cause of AP lies in the inability to process faces holistically/configurally. More precisely, all patients suffering from AP would be unable (or significantly less able) to “integrate the multiple features of an individual face simultaneously, into a unified perceptual representation” ( Tanaka and Farah, 1993 and Rossion, 2008a). Consequently, they would have to process a face feature-by-feature, analytically, or over a small spatial window at a time. Since the region of the eyes contains several elements (two eyes and two eyebrows, at least), a disruption of the ability to process these elements as a whole would be particularly detrimental for the diagnosticity of this facial region. In the same vein, processing a distance between features requires the processing of at least two elements over a wider spatial range than processing a localized single feature. Hence, the loss of the ability to process both the eye region of the face ( Bukach et al., 2006 and Caldara et al., 2005; Rossion et al., 2009) and the relative distances between features ( Barton et al., 2002 and Barton and Cherkasova, 2005) may not reflect distinct fundamental aspects of AP, but rather represent mere consequences of a single cause: a defective holistic processing mode. This view would have the advantage of accounting for the above outlined observations within a single theoretical framework. However, it remains quite speculative at this stage. One way to provide support for this hypothesis would be to demonstrate that (1) the same patients present difficulties in processing the area of the eyes and relative distances between features, and (2) that these two phenomena can be directly related to a disruption of the ability to process the face holistically. With respect to (1), we recently noted that in addition to their impairment at detecting eye changes in a delayed matching task, both PS' and LR's performance at detecting mouth changes was correct but slow relative to controls when the modification concerned the mouth-nose distance, but not the size of the mouth (Rossion et al., 2009). Hence, the patients may indeed exhibit a particular defect at both processing information at the level of the eye region, and of relative distances between features. In an exploratory investigation of PS' ability to discriminate relative distances between features we also noted a pattern of performance which supported a defect in processing relative distances between features. This experiment required same/different judgments of face stimuli that had been used in a study with normal observers (Goffaux and Rossion, 2007; see Fig. 1a). PS completed four blocks of 20 trials/condition, presenting a quite poor performance overall, but with large differences in performance between the perception of inter-feature distances and judgments which could be done locally: nose/mouth changes > eye changes > eye-nose distance > inter-ocular distance (unpublished data, Fig. 1b and c). Her performance for local feature changes (eyes, and nose/mouth) was reasonably above chance, which was in sharp contrast to her extremely poor performance at judging relative distances between these features. Thus, we had preliminary evidence that the patient PS was also impaired at processing relative distances between features. However, we noted that her performance for the different conditions changed across blocks, increasing for some conditions (e.g., vertical distances between features), but apparently at the expense of other conditions (Fig. 1). Moreover, it turned out that during extra sessions, PS verbally stated having suddenly realized the type of manipulation she had been repeatedly missing previously—the inter-ocular distance. Thereafter, PS' performance improved dramatically for this condition, even reaching perfect scores, albeit with extremely prolonged RTs. Exploratory investigation of face discrimination for the patient PS. a. Examples ... Fig. 1. Exploratory investigation of face discrimination for the patient PS. a. Examples of the stimuli used in an initial same-different experiment with two faces presented side-by-side. Four conditions (equalized for difficulty in normal observers) were used to test PS' ability to discern relative distances between features (inter-ocular distance, EH; eyes–nose distance, EV), as well as featural information (eyes only, Ef; nose/mouth NM), respectively. The stimuli used were identical as those employed by Goffaux and Rossion (2007); conditions were presented at random with unlimited time to respond. b–c. PS' accuracy and RTs (in sec) for “different” trials are displayed as a function of (consecutive) testing sessions (84 trials per block, half of which required a “different” response). PS' performance reflected a strong feature versus relative distance dissociation: for single and combined feature changes (Ef, NM) she was reasonably above chance, contrary to her extremely poor performance at judging relative distances, for which she generally displayed a strong bias for “same” responses (with lower performance for EH as compared to EV; only for block 3 did she detect any differences for EH). Note that performance varied across sessions: her initially high performance for NM decreased as it increased progressively for Ef and EV changes. Figure options Thus, even though PS did not automatically perceive the differences in relative distances between features, she was able to successfully discriminate them once she was aware of the modifications employed. This anecdotal observation seems in line with behavior of other prosopagnosic patients as reported in the literature. Patients 3 and 4 reported by Barton et al. (2002) “did markedly better” when required to discriminate relative distances in blocks of “mouth only trials”. Furthermore, when given unlimited time to discriminate faces, their performance also increased substantially. Similarly, Joubert et al. (2003) reported markedly improved performance when their patient FG, who presented with prosopagnosia following a degenerative disease extending to the fusiform gyrus, was made aware of the nature and location of the cues for discriminating faces (eyes color, mouth–nose distance, inter-ocular distance). Finally, Bukach et al. (2006) found that performance of their patient, LR, “improves substantially if eye trials are blocked”—specifically, the condition for which his performance was well below normal level when trials were presented randomly. To summarize, several cases of AP appear to perform poorly at discriminating relative distances as well modifications of the eyes, but can improve substantially if made aware of the nature and location of the diagnostic cues on the face (including the eyes and the relative distance between features). This suggests that the primary cause of their difficulties in processing individual faces may be the disruption of the ability to process all the diagnostic features of a face at once, in a single representation, i.e., their deficit may be an impairment of holistic face processing. 1.1. Goals and hypotheses of the present study The goal of the present study was to provide support for this view by testing the prosopagnosic patient PS' ability to discriminate individual faces based on features and relative inter-feature distances, in two different situations. Given the observations above, the rationale of our investigation was as follows. For a normal observer who has to discriminate between individual faces, an intact holistic processor is functional because any diagnostic cue (e.g., a change in the shape of the mouth, or inter-ocular distance) affects the perception of other (more or less distally located) features of the whole face ( Tanaka and Farah, 1993 and Tanaka and Sengco, 1997). Hence, if two faces differ only in terms of one element (the shape of one feature), the ability to process faces holistically allows rapid identification of the source and location of the information diagnostic for face individuation. In other words, for the normal observer “The general expression of a face is the sum of a multitude of small details, which are viewed in such rapid succession that we seem to perceive them all at a single glance. If any one of them disagrees with the recollected traits of a known face, the eye is quick at observing it, and it dwells upon the difference. One small discordance overweighs a multitude of similarities and suggests a general unlikeness” ( Galton, 1883). However, for the prosopagnosic patient who, presumably, has abnormal holistic processing, facial cues would be perceived one at a time. Thus, a sequential search among many different cues encompassing the entire face would be necessary in order to identify the diagnostic cue. If this notion were correct, indicating the nature of the cue diagnostic for face discrimination should not only improve the patient's performance, but furthermore his/her profile of response should become more similar to that of normal observers. Consequently, the relatively larger impairment for processing information at the level of the eyes and of distances between features in AP should be cancelled out, or at least strongly reduced. Here we tested this hypothesis with the patient PS, a case of AP following lesions to the right inferior occipital cortex and left middle fusiform gyrus, who has been reported in detail in previous studies (e.g., Caldara et al., 2005, Rossion et al., 2003, Schiltz et al., 2006 and Sorger et al., 2007). PS was tested with an individual face matching task in which we manipulated the kind of internal facial cue diagnostic for discrimination (vertical change at the level of the eyes or mouth, inter-ocular distance, featural changes of the mouth, nose or eyes). The two experiments reported in the following differed only with respect to potential a priori knowledge of changes to be discriminated (uncertainty: random trial presentation, vs certainty: trials blocked by condition with participants informed about the nature of the cue). Our main hypothesis was that the region-dependent processing deficit reported previously for PS (observed for the eyes) would disappear if she was informed about change location/type to be discriminated and most importantly that, overall, she would present a response profile similar to that of normal observers under this condition

5. Conclusion Different theoretical accounts have been proposed for two characteristics associated with AP, namely deficient processing of diagnostic information in the eye region, as well as impaired perception of the relative distances between facial features. Here we show that these two impairments, which were presented by a single case of AP, result from an inability to disambiguate the nature and location of the diagnostic cues when individualizing faces. Based on these observations, we suggest that impaired holistic face processing underlies the prosopagnosic deficit of this patient and presumably that of many other cases.