پیدا کردن یک صورت در میان جمعیت: تست اثر برتری خشم در سیندروم اسپرگر

Social threat captures attention and is processed rapidly and efficiently, with many lines of research showing involvement of the amygdala. Visual search paradigms looking at social threat have shown angry faces ‘pop-out’ in a crowd, compared to happy faces. Autism and Asperger Syndrome (AS) are neurodevelopmental conditions characterised by social deficits, abnormal face processing, and amygdala dysfunction. We tested adults with high-functioning autism (HFA) and AS using a facial visual search paradigm with schematic neutral and emotional faces. We found, contrary to predictions, that people with HFA/AS performed similarly to controls in many conditions. However, the effect was reduced in the HFA/AS group when using widely varying crowd sizes and when faces were inverted, suggesting a difference in face-processing style may be evident even with simple schematic faces. We conclude there are intact threat detection mechanisms in AS, under simple and predictable conditions, but that like other face-perception tasks, the visual search of threat faces task reveals atypical face-processing in HFA/AS.

Faces are one of the most important visual stimuli and are potent facilitators for social interaction and communication. Facial expressions of emotion convey critical signals for inferences about the intentions and motivations of others (Blair, 2003 and Darwin, 1872/1965). Although faces provide a wealth of information, people are generally able to extract important information rapidly and efficiently and to produce appropriate responses. Humans are so adept at various aspects of face processing it is suggested we may have evolved special face processing modules (Ekman, 2003 and Young, 1998). However, not all humans are proficient at face processing. Autism and Asperger Syndrome (AS) are neurodevelopmental conditions characterised by severe social and communication difficulties, as well as restricted behaviours and interests (APA, 1994 and ICD-10, 1994). From the earliest descriptions of these disorders, striking abnormalities were noted in social-emotional behaviour including difficulties with face processing and social interactions (Asperger, 1944 and Kanner, 1943). Face-processing deficits are likely to relate to the social difficulties, in that people with autism spectrum conditions (ASC) have not developed the same expertise with faces as typical controls (Grelotti et al., 2005). Various differences have been reported about how people with ASC process faces. For example, while typically developing people normally use a more holistic style when processing faces and emotional expressions (Tanaka and Farah, 1993 and Yin, 1969), people with autism rely on a more feature-based style of processing faces (Hobson et al., 1988, Langdell, 1978 and Weeks and Hobson, 1987). People with ASC also focus their attention more on the mouth region when processing faces, while typical controls rely on the eye region that provides more information about the emotional states of others (Baron-Cohen et al., 1997, Joseph and Tanaka, 2003 and Weeks and Hobson, 1987). Results concerning emotional expression processing in ASC have been less consistent and suggest an uneven profile. People with ASC show difficulties on tasks with complex mental states and emotions (Baron-Cohen et al., 1993, Baron-Cohen et al., 2001 and Baron-Cohen et al., 1997), whereas their accuracy in recognising basic emotions may be intact, particularly when high-functioning participants are studied (Adolphs et al., 2000, Baron-Cohen et al., 1997, Golan et al., in press, Grossman et al., 2000 and Volkmar et al., 1989). However, other studies have reported deficits by people with autism in processing basic emotions (Bolte and Poustka, 2003, Celani et al., 1999 and Howard et al., 2000), somewhat confusing the current understanding of emotion recognition in autism (Frith, 2003). Research involving the more automatic and implicit emotional processing mechanisms in people with ASC has been lacking. When task variables and procedures are kept simple and predictable, people with autism show evidence of normal configural face-processing strategies (Joseph and Tanaka, 2003 and Teunisse and de Gelder, 2003). These findings suggest that the ability to process faces using configural or holistic processing styles may not be completely absent in ASC, rather that people with these conditions are just more likely to process information using a ‘cognitive style’ characterised by enhanced local feature detection over holistic processing (Frith, 2003 and Happe, 1999). Facial expressions of emotion are important for non-verbal communication, and facial threat is a particularly potent social signal. The emotional expression of anger is a typical example and is a very potent facial warning signal. Threatening facial expressions have considerable power to recruit attention and are processed rapidly and efficiently (Vuilleumier & Schwartz, 2001). Neuroimaging experiments have shown that angry and fearful expression faces activate the amygdala (Breiter et al., 1996, Morris et al., 1996 and Phillips et al., 1998), a brain area important for detecting threat and producing appropriate responses (Aggleton, 2000 and LeDoux, 1996). The amygdala activation occurs even when threatening faces are masked and below the level of awareness (Morris et al., 1998 and Whalen et al., 1998), demonstrating a quick and direct sub-cortical route to the amygdala for automatic processing of threat (Morris, Ohman, & Dolan, 1999). These findings are consistent with investigations of humans with amygdala damage, who show deficits in perceiving fearful expressions (Adolphs et al., 1995, Adolphs et al., 1999 and Calder et al., 1996). Amygdala patients also judge people rated negatively by typical controls as being more trustworthy and approachable (Adolphs, Tranel, & Damasio, 1998), and do not show the enhanced perception of emotionally significant stimuli normally seen in control participants (Anderson & Phelps, 2001). The face-processing deficits seen in amygdala patients are similar in some ways to emotion processing deficits seen in people with ASC, including reports that the deficits may involve complex emotions more than basic emotions (Adolphs et al., 2002, Adolphs et al., 2000, Baron-Cohen et al., 1997 and Golan et al., in press). Neuroimaging studies of people with ASC have shown decreased amygdala activation while processing faces, including threatening expressions (Ashwin et al., in press, Baron-Cohen et al., 1999, Critchley et al., 2000 and Pierce et al., 2001). A simple task developed to investigate the attention capturing abilities of threat is the ‘face-in-the-crowd’ visual search paradigm. A pioneering study by Hansen and Hansen (1988) found that an angry face was detected more quickly and accurately than a happy face in a crowd of distracter faces. They further found this ‘anger superiority’ effect was unaffected by the number of distracter faces in the display, supporting the notion that facial threat detection may elicit pre-attentive processing involving ‘pop-out.’ The pop-out effect shows reaction times (RT’s) that do not vary greatly with increasing size of the distracters, as shown by a search slope (the increase in RT divided by the increase in number of distracters) in the range of 5–6 ms/item (Hershler and Hochstein, 2005 and Treisman and Souther, 1985). However, further studies with similar stimuli did not replicate these findings and criticisms were raised about the results involving visual confounds and lack of a condition with neutral face crowds (Nothdurft, 1993 and White, 1995). To address these problems and further test the face-in-the-crowd effect, researchers have developed schematic faces for visual search experiments. By using schematic faces it is possible to eliminate many low-level perceptual variations found in emotional expression photographs and to allow for greater control over experimental variables. The features of angry and happy schematic faces can be matched very closely and easily manipulated to test a variety of factors in a consistent way. Naturally, this greater control comes at the cost of a lack in ecological validity. Several studies have shown schematic threatening faces are found more quickly and accurately than schematic friendly faces, strengthening the idea that social threat captures attention (Eastwood et al., 2001, Fox et al., 2000 and Ohman et al., 2001). Another intriguing effect is longer response latencies for non-target displays containing all-angry faces compared to all-happy displays, which is thought to reflect that each angry face captures attention to a greater degree than each happy face (Fox et al., 2000, Vuilleumier and Schwartz, 2001 and White, 1995). However, other studies have failed to replicate the findings of ‘pop-out’ for schematic threatening faces (Ohman et al., 2001) and for longer dwell times for all-angry displays (Ohman et al., 2001), suggesting further replication studies of this type are needed. For facial threat to elicit the rapid and efficient extraction of information and adaptive responses, there have to be clear and prototypical features (Darwin, 1872/1965, Lundqvist et al., 1999 and Ohman and Soares, 1993). Lundqvist et al., 1999 and Lundqvist et al., 2004 have investigated the role of individual schematic facial features and configurations of schematic features in their ability to convey threat. Results showed that V-shaped eyebrows are the most salient individual feature for conveying threat, followed by down-turned mouths (Lundqvist et al., 1999). In addition, they found individual features do not convey the same degree of threat as configurations of facial features, particularly compared to the most negatively rated configuration of V-shaped eyebrows and down-turned mouths together. From their findings Lundqvist et al. (2004) illustrated different levels of face-processing including single features, feature configurations, and holistic shapes, and the importance of configurations of features for conveying highly effective threat. These ideas of different levels of information extraction from faces is consistent with a prominent model of face-processing (Bruce & Young, 1986) that posits face recognition involves different modules extracting various aspects of facial information in parallel. Results from visual search studies with schematic faces show the anger superiority effect does not emerge from single feature detection, as the effect is not seen when threatening features are presented in isolation. Fox et al. (2000) found no difference in RT’s to detect angry (down-turned) versus friendly (up-turned) mouths when they were presented in isolation. Similarly, others have found that people are no quicker to detect threatening (V-shaped) eyebrows than friendly eyebrows when presented in isolation (Tipples, Atkinson, & Young, 2002). Therefore, V-shaped eyebrows need a configuration containing other internal facial features in order to produce a facilitated detection effect. Consistent with this, Eastwood, Smilek, and Merikle (2003) found the detection of threatening schematic faces was actually associated with a disruption in feature-processing and longer latencies, and a facilitation in the configural processing of faces. Therefore, it appears the anger superiority effect involves some degree of configural and/or holistic level of face-processing ( Fox et al., 2000, Lundqvist et al., 2004, Ohman et al., 2001 and Tipples and Young et al., 2002). Research with visual search tasks have also reported normal or even superior ability by people with ASC compared to controls in non-social paradigms ( O’Riordan and Plaisted, 2001, O’Riordan et al., 2001, Plaisted et al., 1998a and Plaisted et al., 1998b). Visual search studies have typically found children and adults with ASC are better able to detect target stimuli among distracters. This is important as it reflects a paradigm where people with ASC are known to perform at levels comparable to controls, probably because it is a simple and predictable task with limited requirements of language ability or executive function. These factors may confound other studies of social-emotional functioning in ASC ( Teunisse & de Gelder, 2003). However, to date no published research has reported findings of visual search paradigms with social-emotional stimuli in autism. This gap in the evidence is filled by the experiments reported here. This is important because it allows us to test 4 questions: (a) Are the rapid, evolved aspects of threat detection intact in HFA/AS? (b) Is basic emotion discrimination intact in HFA/AS? (c) Do the savant skills in visual search extend to detecting odd-one-out faces? And finally (d); Is there any evidence of a difference in cognitive style in HFA/AS (e.g., under conditions of face-inversion)? 1.1. Aims and hypotheses The present study investigated the effects of the following variables on RT and accuracy to detect threatening and happy faces in crowds of distracter faces: the size of the matrix, time of presentation, and inverted presentation. The experiments had the following aims: (1) To replicate previous studies reporting facilitated detection of threatening faces over friendly faces; (2) to test if previous findings of longer search times for displays containing all angry faces would replicate; (3) to investigate the pop-out effect for angry faces; and (4) to determine if people with HFA/AS show the same attention biases for the detection of threatening faces compared to control participants. We predicted the typical control group would be quicker and more accurate in the detection of the threatening faces compared with the friendly faces, and would take longer and be less accurate to detect non-target displays comprised of all angry faces, consistent with previous studies. If the evolutionary model of threat detection is true, we expected to find evidence for pre-attentive processing of the threatening faces consistent with a pop-out effect. We were open to how the group with HFA/AS would perform, given previous evidence from some studies that basic emotion processing is intact (Adolphs et al., 2000, Baron-Cohen et al., 1993, Baron-Cohen et al., 1997 and Golan et al., in press). If people with ASC are processing faces using only single features, we would not expect them to show evidence of the anger superiority effect. However, if they are extracting facial information using a configural and/or holistic level of processing, then we would expect to see evidence of an angry face superiority. Further, if people with ASC are extracting facial information using a lower level of processing than controls (e.g., feature configural versus holistic), we would predict they will show less evidence of the anger superiority effect than the control group. We expected differences in processing style would be most evident in the face inversion condition, consistent with previous findings of face orientation (Langdell, 1978). 2. Experiment 1 Previous research has shown that threatening faces are detected faster and more accurately than friendly faces (Hansen and Hansen, 1988 and Ohman et al., 2001), and that all-angry displays are detected slower and less accurately than all-happy displays (Fox et al., 2000 and White, 1995). In Experiment 1, the effect of exposure time on threat detection was investigated by setting the display time of the matrices to be either short (1 s) or long duration (2 s). People with autism have difficulty making social judgments, which may be more evident under shorter time constraints. We tested (1) whether the control group would show faster and more accurate detection of threatening compared to friendly faces, (2) if there would be longer dwell times for all-angry displays, and (3) if the HFA/AS group would show the same behavioural effects as the control group.