الگوی نابجای اسکن در پروزوپاگنوزیا نشان دهنده اختلال در چهره پردازی

Abstract Visual scanpath recording was used to investigate the information processing strategies used by a prosopagnosic patient, SC, when viewing faces. Compared to controls, SC showed an aberrant pattern of scanning, directing attention away from the internal configuration of facial features (eyes, nose) towards peripheral regions (hair, forehead) of the face. The results suggest that SC’s face recognition deficit can be linked to an inability to assemble an accurate and unified face percept due to an abnormal allocation of attention away from the internal face region. Extraction of stimulus attributes necessary for face identity recognition is compromised by an aberrant face scanning pattern.

Introduction Prosopagnosia is a rare neurological deficit characterised by an inability to recognise facial identity. A hallmark of the disorder is impaired processing of configural information involving the unique spatial relationships between the facial features. The failure to recognise faces has sometimes been attributed to a complete loss of the configural strategy (Saumier, Arguin, & Lassonde, 2001). Consequently, prosopagnosics are said to perceive faces in terms of analytic processing of facial features in contrast to a more integrated approach using configural information as engaged in normals. It is not known what causes the configural deficit in prosopagnosia. Depending on the nature of damage impaired face identity recognition may manifest at different stages of face processing (Bruce and Young, 1986 and Fox et al., 2008). Two main subtypes of prosopagnosia have been defined depending on the locus of damage. These include apperceptive and associative forms (Barton et al., 2004, De Renzi et al., 1991 and Sergent and Signoret, 1992b). In apperceptive prosopagnosia the main deficit is in generating an adequate percept of a seen face with which to match against a stored memory. The deficit here is therefore in encoding the facial stimulus. In contrast, in associative prosopagnosia early perceptual processes are considered intact, but the individual incapable of accessing identity-semantic information in response to face stimuli (Bruyer, 1991, De Renzi et al., 1991 and Farah, 1991). Impairment in these cases is thought to stem from deficient processing at stages which follow encoding including, matching the sensory representation to a stored representation or an inaccessibility of memory representations, or both. An individual’s visual scanpath record corresponds to the pattern of eye movements or saccades that are made during visual processing and informs as to the nature of information acquired from a scene or stimulus (Gordon et al., 1992, Manor et al., 1999 and Norton and Stark, 1971). Scanpath parameters such as the pattern of eye fixations can be used to inform where attention is being allocated (Rayner, 1995). Perception results from the integration of information acquired across eye fixations and as such is governed by the efficiency and distribution of gaze control (Bloom and Mudd, 1991 and Henderson and Hollingwort, 1999). Eye movement studies to different stimuli such as scenes and objects have found that the sequence of eye movements or visual exploration pattern is not random (Noton and Stark, 1971 and Rayner, 1995). Indeed, face identity recognition in non-patient groups has been linked to a stereotyped sequence of viewing directed towards the internal configuration of facial features including the eyes, mouth and nose respectively, and away from peripheral regions including the hair/hairline and ears (Henderson et al., 2001, Mertens et al., 1993, Rizzo et al., 1987, Schwartz et al., 1999 and Walker-Smith et al., 1977). This visuo-cognitive strategy has been associated with the formation of a unified configural percept involving information from the internal facial features and their configuration. This is thought to be necessary for accurate face identity recognition (Norton & Stark, 1971). Abnormalities of visual scanning of expressive faces in schizophrenia (Kee et al., 1998, Loughland et al., 2002, Manor et al., 1999, Rosse et al., 1998, Schwartz et al., 1999 and Streit et al., 1997), autism (Pelphrey et al., 2002), social phobia (Horley, Williams, Gonsalvez, & Gordon, 2003), and individuals with amygdala damage (Adolphs et al., 2005), have been linked to deficits in emotional recognition in these patient groups. Indeed, compared to non-patient groups the scanpaths tend to be non-focal, more erratic, and undirected, often reflecting the processing of limited internal featural information. As a result, it has been argued in these groups that those stimulus attributes linked to information from the internal face region are not adequately extracted and there is insufficient information to inform an accurate response. While in different patient groups scanpath parameters have been examined with respect to facial expression recognition, similar technology may be used to assess the pattern of scanning during face identity recognition in prosopagnosic patients. Remarkably, given the interest in visual scanpath data in psychiatric populations, to date, few studies have examined the eye movement pattern of prosopagnosic patients. The results have been mixed. Rizzo et al. (1987) showed that compared to controls there was no deviation in the eye movement pattern during face learning and recognition in two patients tested: saccades, fixation and feature scanning of scenes and faces were normal. The study did however identify individual differences in the mode of scanning familiar compared to unfamiliar faces. This result is comparable to findings of covert recognition in other prosopagnosics using autonomic and behavioural indexes (Barton and Cherkasova, 2003, Barton et al., 2001, Baurer, 1984, de Haan, 1999, Diamond et al., 1994, Sergent and Signoret, 1992a, Wallace and Farah, 1992 and Young, 1994). Indeed, qualitative differences in information processing of familiar and unfamiliar faces have been proposed in normal subjects. Behavioural studies have found that familiar faces are more accurately recognised from the internal region including the eyes–nose–mouth, than from external features such as the hair and chin, which are more important for initial processing of unfamiliar faces (Bruce and Young, 1998, Clutterbuck and Johnston, 2002 and Ellis et al., 1979). The inner face advantage is thought to be associated with the many social attributes that can be inferred from this region which carries greater configural information and is less likely to change with age (Campbell et al., 1999 and Want et al., 2003). Le, Raufaste, and Demonet (2003a) investigated face perception in a visual agnosic and prosopagnosic patient, SB, and found a difference in the exploration sequences and fixation number compared to controls (see also, Le, Raufaste, Roussel, Puel, and Demonet (2003b)). SB showed an aberrant scanning pattern, preferentially attending to the hair and hairline. Furthermore, in a subsequent face detection task controls fixated fewer areas of interest, a result linked to the ability to engage configural-holistic processes thought to be unavailable to SB. SB showed no scanpath evidence of covert recognition of overtly unrecognised faces. A similar discrepancy in gaze behaviour, dispersed across the whole face was also found by Schwarzer et al. (2007) in a group of hereditary prosopagnosics. The presence of covert recognition would support evidence of a face-specific memory deficit in some instances of prosopagnosia and the absence of covert recognition, a deficit that is perceptual in nature (Barton et al., 2001, Newcombe et al., 1989, Sergent and Signoret, 1992a and Sperber and Spinnler, 2003). Indeed, in both cases tested by Rizzo et al. (1987) the disturbance appeared to be a dysfunction of memory recall/retrieval rather than perception, as evidenced by normal performance on perceptual tests of face processing (see also McNeil and Warrington (1991)). In SB perceptual processing was severely impaired while his scanpath pattern failed to demonstrate covert effects. The presence or absence of evidence of covert recognition in the scanpath pattern may therefore correlate with the locus of the functional impairment. Here we investigated the pattern of eye movements in a 38-year-old patient, SC, with severe acquired prosopagnosia. The focus of this experiment was on how SC processes visual information from faces during an identity task. The aim was to determine whether abnormalities in exploratory eye movements are associated with impaired face identity recognition. On the basis of the research reviewed above and previous findings of impaired configural processes in SC in identity and expression recognition (Stephan, Breen, & Caine, 2006), it was hypothesised that SC would show scanpath disturbances relative to controls, resulting in fewer fixations to the internal configuration of facial features and a pattern characteristic of a featural-based processing strategy.

. Conclusion The measurement of eye movements made during identity recognition in prosopagnosic patients allows a quantitative assessment of which information these patients use to identify faces currently not widely available. While visual scanning has been assessed in other prosopagnosic patients the findings have been mixed. The results of this study advance our understanding of the mechanism underlying the prosopagnosic deficit in this case: the prosopagnosic impairment in SC arises as a consequence of insufficient acquisition of face information from the internal face region. Whether intervention aimed at directing eye movements can improve identity recognition will have important implications for overcoming the face recognition deficit in those cases where impairment can be linked to damaged, but modifiable perceptual processes.