Holistic perception of the individual face is specific and necessary: Evidence from an extensive case study of acquired prosopagnosia

Abstract

We present an extensive investigation (24 experiments) of a new case of prosopagnosia following right unilateral damage, GG, with the aim of addressing two classical issues: (1) Can a visual recognition impairment truly be specific to faces? (2) What is the nature of acquired prosopagnosia? We show that GG recognizes nonface objects perfectly and quickly, even when it requires fine-grained analysis to individualize these objects. He is also capable of perceiving objects and faces as integrated wholes, as indicated by normal Navon effect, 3D-figures perception and perception of Mooney and Arcimboldo face stimuli. However, the patient could not perceive individual faces holistically, showing no inversion, composite, or whole-part advantage effects for faces. We conclude that an occipito-temporal right hemisphere lesion may lead to a specific impairment of holistic perception of individual items, a function that appears critical for normal face recognition but not for object recognition.

Introduction

The ability to recognize people from their face is a fundamental brain function which holds a high social value. It is also an extremely complex function, which is nevertheless performed quite well in human adults. The adult human brain has developed mechanisms allowing, for instance, recognizing a familiar person from its face in less than half a second (Bruce & Young, 1986), or encoding new faces in memory effortlessly during the entire life (e.g., Bahrick, Bahrick, & Wittinger, 1975). Yet, interestingly, the field of face recognition was originally based upon the study of people who, following brain damage, have lost this expertise in recognizing faces.

Difficulty in face recognition as a major symptom in patients with cerebral disease was first reported in the nineteenth century (Charcot, 1883, Quaglino and Borelli, 1867, Wigan, 1844, Wilbrandt, 1887). However, it was Bodamer (1947) who proposed to isolate the disorder on the basis of three cases, and introduced the term prosopagnosia from the Greek “prosopon” (face) and “a-gnosia” (without knowledge). Prosopagnosia is classically defined as the inability to recognize individual faces following brain damage, an impairment that cannot be attributed to intellectual deficiencies or low-level visual problems (Benton, 1980, Bodamer, 1947, Hécaen and Angelergues, 1962, Rondot and Tzavaras, 1969). Prosopagnosic patients also generally still retain their ability to recognize people by other cues: the voice or other visual traits such as gait, size, clothes, or even facial features (moustache, scar, freckles, …) or accessories (ear-rings, eyeglasses, piercings, …).

Over the years, tens of cases of prosopagnosia following brain damage have been reported, although extensive neuropsychological investigations of prosopagnosic patients remain quite rare (e.g., Anaki et al., 2007, Barton, 2008a, Delvenne et al., 2004, Lhermitte et al., 1972, Riddoch et al., 2008, Rossion et al., 2003, Sergent and Signoret, 1992a, Sergent and Villemure, 1989).

Both in traditional (cognitive) neuropsychology and in modern cognitive neuroscience, the lesion method is seen as an invaluable and unique way of understanding normal brain function (e.g., Caramazza, 1986, Damasio and Damasio, 1989, Farah, 1990, Farah, 2004, Humphreys and Riddoch, 1987, Shallice, 1988), in particular with respect to face recognition. Such patient studies contribute to shaping our knowledge and conceptions of the processes involved in normal face recognition and their underlying neural networks.

From a functional point of view, there are two main debates concerning prosopagnosia, which have direct implications for understanding face recognition: (1) Can the impairment truly be restricted to face recognition (i.e. face-specific)? (2) What is the nature of the disorder, that is, what is at the heart of our expertise in facial recognition, and which is lost in these patients? These two issues have proved quite difficult to resolve and are still debated (e.g., Barton, 2009, Damasio et al., 1982, De Renzi, 1986a, Hécaen, 1981, Riddoch et al., 2008, Rondot and Tzavaras, 1969, Sergent and Signoret, 1992a).

The issue of the specificity of the disorder has been complicated by the fact that most reported cases of prosopagnosia also present with difficulties in basic-level object recognition (e.g., Barton, 2008a, Boutsen and Humphreys, 2002, Damasio et al., 1982, Delvenne et al., 2004, Gauthier et al., 1999Levine and Calvanio, 1989, Steeves et al., 2006). In many other cases, object recognition abilities were not tested sufficiently (e.g., De Renzi, 1986a, Ettlin et al., 1992, Tohgi et al., 1994, Young et al., 1993). A brief but extensive review of the neuropsychological literature points to 13 prosopagnosic patients who could be considered as presenting with a face-specific recognition disorder (Table 1). De Renzi (1986a) presented patient 4 who performed in the normal range at object and figure recognition, figure–ground discrimination, visual closure and segmentation. Patient VA (De Renzi, Faglioni, Grossi, & Nichelli, 1991) could name objects and pictures (presented under usual and unusual view) in the normal range, and succeeded at tasks of visual closure, coin discrimination, and recognition of makes of cars and personal belongings. Another patient described by De Renzi, Perani, Carlesimo, Silveri, and Fazio (1994), OR, was documented to present with an absence of impairment with respect to object naming, Italian coins discrimination, and recognition of animals, fruits and vegetables (under usual and unusual views). Takahashi, Kawamura, Hirayama, Shiota, and Isono (1995), in a study of four patients, related the case of a patient with apparently no object recognition impairment: case 3 succeeded in different tasks including overlapping figures, Gestalt completion test, Kanizsa triangle and real object naming. Schweinberger, Klos, and Sommer (1995) and Henke, Schweinberger, Grigo, Klos, and Sommer (1998) showed that the performance of patient MT was preserved in numerous tasks: recognizing overlapping figures, Gestalt completion task, object naming, animal naming and different series of similar objects to name (fruits and vegetables, symbols of German industrial brands and cars brands). Patient WB (Buxbaum, Glosser, & Coslett, 1996) presented with preserved object naming (real objects and drawings) and memory for homogeneous category of objects (glasses, under different views). Patient Anna (De Renzi & di Pellegrino, 1998) succeeded in several tasks: objects naming (color photographs and line drawings), perceptual categorization, visual segmentation and closure, and memory for homogeneous category of objects (glasses, under different views). Another study (Wada & Yamamoto, 2001) also reported a prosopagnosic patient who could perform well the tasks of overlapping figures, picture copying, recognition of letters and symbols, visual space perception, object naming (real objects, pictures, line drawings; under usual and unusual view), animal face and famous place recognition. Prosopagnosic patient PS was able to recognize objects perfectly and rapidly (Rossion et al., 2003) and could perform within-category discrimination for nonface items in the normal range of performance and speed (Busigny et al., 2010, Schiltz et al., 2006). Barton and colleagues reported case 009 (Barton, 2008a, Barton, 2009, Barton and Cherkasova, 2005, Barton et al., 2004), a patient who had no low-level visual impairment and was able to recognize incomplete letters, overlapping figures, real objects, vegetables and fruits, presented with a classical Navon effect, and showed some ability to process configurations of dots. Bukach, Bud, Gauthier, and Tarr, 2006 and Bukach, Le Grand, Kaiser, Bub, and Tanaka, 2008 related another case of selective impairment for faces, LR, who succeeded easily in tasks of low-level visual processing, silhouettes and object naming (under usual and unusual views). Riddoch et al. (2008) presented the case of FB, who had preserved abilities in low-level visual processing, non-living and living (birds, flowers, vegetables and fruits) objects naming, and in a task of learning associations between names and novel multipart objects. Finally, Rivest, Moscovitch, and Black (2009) published the case of DC, who performed normally in segmented object recognition, object naming, recognition of famous buildings and dog breeds.

These pure cases of acquired prosopagnosia suggest that some processes may be necessary to recognize faces efficiently, and that these processes may be selectively disrupted by brain damage. While these processes might also be involved in object recognition, they would not be necessary for this function.

Regarding the nature of the impairment in prosopagnosia, an influential idea is that such patients have difficulties in perceiving a face as a whole, or a Gestalt. This long-standing view (Galli, 1964) is inspired originally from the Gestaltist approach of visual perception (e.g., Koffka, 1963, Kohler, 1929, Kohler, 1971, Wertheimer, 1967). According to the Gestaltist view and its more modern revival (e.g., Kubovy and Poremantz, 1981, Navon, 1977; for a review see Kimchi, 1992), a whole item is qualitatively different from the sum of the components, the whole exceeding the sum of its parts. Hence, what takes place in each single part already depends upon what the whole is: objects are not only made of featural elements, but also defined by the interactions between these constituents, a property that is called configuration or (w)holistic property (e.g., Navon, 2003). For instance, a face is a typical visual stimulus made of parts (eyes, nose, mouth, …) that are organized in a whole configuration (a symmetrical structure with two eyes on top, above a central nose and mouth).

The idea that acquired prosopagnosic patients lose their ability to perceive faces holistically is supported at four levels.

First, many patients have been described as presenting with a configural/holistic¹ processing impairment, that is, an inability to integrate simultaneously different features into a coherent global representation (RB, Davidoff, Matthews, & Newcombe, 1986; HJA, Boutsen and Humphreys, 2002, Riddoch and Humphreys, 1987; LH, Levine & Calvanio, 1989; BM, Sergent & Villemure, 1989; WL, Spillmann, Laskowski, Lange, Kasper, & Schmidt, 2000; AR, Saumier, Arguin, & Lassonde, 2001; RC, Wilkinson et al., 2009; PS, Ramon, Busigny, & Rossion, 2010). For example, Levine and Calvanio (1989) described the patient LH as being unable to “get an immediate overview of a face […] as a whole at a single glance” (p. 159). They conceptualized this loss of visual “configural [i.e. holistic] processing” 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”. In the same vein, Spillmann et al. (2000) described the prosopagnosic patient WL as following: “he was unable to form a holistic percept of a given face that would have revealed its bearer's identity. Rather, he used conspicuous features for recognition […]. Recognition was based on characteristic details, not faces per se […]. What seems to be lacking in WL is the ability to create an integrated, unitary percept or a gestalt of a human face enabling him to assign identity to an individual” (pp. 93, 98).

Second, very few cases of acquired prosopagnosia with preserved holistic face processing have been reported (PV, Sergent & Poncet; PC, Sergent & Signoret, 1992a; LR, Bukach et al., 2006). Moreover, holistic processing was not tested extensively and with particularly sensitive tests in these patients, so that it remains unclear to what extent their holistic processing of faces was truly preserved (see Ramon et al., 2010). This issue will be addressed more extensively in Section 4.

Third, there is to date no solid and more accurate alternative hypothesis to account for the functional impairment characterizing acquired prosopagnosia. For instance, the few alternative proposals in terms of low-level processing are no longer valid. Indeed, an account of prosopagnosia – or visual agnosia – in terms of sensory or low-level visual impairments (Bay, 1953, Ettlinger, 1956), has been dismissed for some time (De Haan et al., 1995, Rondot and Tzavaras, 1969), and many prosopagnosic patients do not suffer from low-level visual problems (e.g., Bukach et al., 2006, Buxbaum et al., 1996, Delvenne et al., 2004, Dixon et al., 1998, Eimer and McCarthy, 1999, Schweinberger et al., 1995, Sergent and Poncet, 1990, Wada and Yamamoto, 2001). Even when low-level vision is impaired, such as color vision (i.e. achromatopsia, as in many cases of prosopagnosia, see Bouvier & Engel, 2006), or visual defects in the left upper quadrant (Bouvier and Engel, 2006, Hécaen and Angelergues, 1962, Meadows, 1974) these associated defects cannot account for the face recognition impairment (Rondot & Tzavaras, 1969).

Fourth and finally, alternative views of prosopagnosia which consider this syndrome as a high-level visual defect can be easily integrated into a holistic processing impairment account. For instance, it has been suggested that the processing of the region of the eyes in faces is particularly problematic for prosopagnosic patients (Gloning, Gloning, Hoff, & Tschabitscher, 1966), a proposal which has received recent empirical support by studies showing a reduced diagnosticity of the region of the eyes of faces for the patients PS (Caldara et al., 2005, Rossion et al., 2009) and LR (Bukach et al., 2006, Bukach et al., 2008). However, the reason why these patients do not rely on the eyes region, and fixate this region less often than normal observers during face recognition (Orban de Xivry, Ramon, Lefèvre, & Rossion, 2008), may be directly related to their inability to process individual faces holistically. Indeed, for a patient who cannot encode the individual features of the face as a single representation, it may be better to focus on an isolated feature (e.g., the mouth) which may contain in itself more information than each of the elements of the eye region considered in isolation (see Caldara et al., 2005, Orban de Xivry et al., 2008, Rossion et al., 2009, Van Belle et al., 2010a). In the same vein, an impairment in perceiving relative distances between features in prosopagnosia (Barton, 2009, Barton and Cherkasova, 2005, Barton et al., 2002) may be a consequence of the difficulty to perceive the face as a whole (Ramon and Rossion, 2010, Rossion, 2008a, Rossion, 2009, Sekunova and Barton, 2008). These issues will be developed further in the present paper.

Given these considerations, the holistic perception account of (acquired) prosopagnosia can be considered to be the dominant view. However, there is at least one important issue that needs to be resolved: since object recognition is based – at least to some extent – on the ability to perceive an object or a general visual pattern holistically (e.g., Kimchi, 1992, Kimchi, 2000, Navon, 1977), how could prosopagnosia be specific to faces if what characterizes prosopagnosic patients is an impairment in holistic perception? This paradox is reinforced by the fact that most case studies who support the holistic account of prosopagnosia have reported patients who suffer from important object recognition impairments, and who were actually tested with non-face objects (e.g., HJA, Riddoch & Humphreys, 1987; LH, Levine & Calvanio, 1989; WL, Spillmann et al., 2000; AR, Saumier et al., 2001; CR, Behrmann and Williams, 2007, Gauthier et al., 1999; SM, Behrmann and Kimchi, 2003, Behrmann and Williams, 2007, Gauthier et al., 1999; RN, Behrmann & Kimchi, 2003; NS, Delvenne et al., 2004).

To resolve this issue, one may consider that nonface objects are perceived in a part-based manner (Biederman, 1987, Marr, 1982, Treisman, 1986), while faces only would be perceived holistically (Biederman and Kalocsai, 1997, Moscovitch et al., 1997; see also McKone, Martini, & Nakayama, 2003). However, there are numerous instances in which a nonface object pattern is perceived holistically (e.g., Kimchi, 1992, Navon, 1977).

Another way to resolve this issue is by proposing that faces are processed more holistically than other objects (Farah, 1990, Farah, 2004). That is, while object perception would depend on both part-based and holistic processes, faces would depend exclusively on holistic processes. Depending on severity of the impairment, patients would be either prosopagnosic only, or prosopagnosic and object agnosic (Farah, 1990, Farah, 2004). However, it is unclear based on this account how “the severity” of the impairment can be assessed, and thus how the recognition impairment may be completely restricted to faces in certain cases. In fact, this view was inspired by investigations carried out on the patient LH, a case of prosopagnosia who also presented with a severe impairment in object recognition (Farah et al., 1995b, Levine and Calvanio, 1989, Levine et al., 1980).

Finally, since most case studies supporting the holistic account of prosopagnosia have reported patients who are impaired at non-face object recognition (Behrmann and Kimchi, 2003, Behrmann and Williams, 2007, Delvenne et al., 2004, Gauthier et al., 1999, Levine and Calvanio, 1989, Saumier et al., 2001, Spillmann et al., 2000), it may be that the few cases of pure prosopagnosia reported in the literature (Table 1) do not suffer from an impairment at holistic processing. However, while holistic object perception has been tested and found normal in some of these patients (see Table 1), their ability to process faces holistically remains largely unclear. Moreover, there is evidence collected from separate studies performed in at least one of these cases of prosopagnosia (PS) that visual recognition difficulties can be truly restricted to faces (Rossion et al., 2003, Schiltz et al., 2006), and yet concern holistic perception of faces (Busigny and Rossion, 2010, Ramon et al., 2010, Van Belle et al., 2010a).

Here we suggest that none of the above proposals is satisfying, and that the key issue is not that face recognition would be “more” holistic than object recognition, or that objects would not be processed holistically. Rather, we suggest that both basic-level face and object categorization (“it is a face”; “it is a banana”) rely on holistic processes. These processes are impaired in patients suffering from integrative visual agnosia, and these patients also suffer from prosopagnosia. When an individual item of a visual category has to be identified and/or differentiated from other individuals from the same category, normal observers would rather rely largely on part-based processes. In contrast, individualizing faces would still require the ability to perceive (the individual item) holistically. That is, contrary to nonface objects, fine-grained discrimination of individual faces would also depend critically on the ability to perceive faces holistically (Biederman & Kalocsai, 1997).

If this hypothesis is true, a patient with pure prosopagnosia following brain-damage, should have (1) preserved basic-level object recognition, even when holistic processing is required; (2) preserved face detection, even when holistic processing is required; (3) preserved individual level object recognition, that is, when fine-grained discrimination is required. However, the patient should be impaired at individual face recognition/discrimination, and be insensitive – or significantly less sensitive than normal observers – to the classical effects measuring holistic processing of the individual face (Tanaka and Farah, 1993, Young et al., 1987).

As mentioned above, over the last few years, we have reported evidence supporting most of these claims from investigations of the prosopagnosic patient PS in separate studies (Busigny and Rossion, 2010, Busigny and Rossion, in press, Ramon et al., 2010, Rossion et al., 2003, Schiltz et al., 2006). Here, our goal was to fully address this crucial issue in a complete and exhaustive case study of a new case of prosopagnosia.

To this aim, we report the patient GG, who suffered from a stroke in 2003 in the territory of the right posterior cerebral artery, causing focal posterior right hemisphere brain damage. No controlateral (left hemisphere) insult was evidenced on structural (MRI) and perfusion (SPECT) brain studies. This patient is particularly interesting to test our hypothesis, because, while being a very conscientious and dedicated man who was alert and cooperative throughout the study, he did not and does not complain at all of object recognition difficulties in real life. Moreover, like many acquired prosopagnosic patients, when he had to describe his face recognition difficulties he mentioned that he was no longer able to build a “global picture” of faces. Rather, in order to recognize faces, he reported trying to identify a particular feature (pimple, wart, very big green eyes, red hair, freckles, …). GG thus provided us with a unique opportunity to better understand the nature of the face processing impairment in prosopagnosia.