اسکن بیش از حد و اجتناب در اختلال اضطراب اجتماعی: مسیر اسکن بصری در طول صحبت های عمومی

Abstract Social anxiety disorder (SAD) is a debilitating mental illness which is thought to be maintained in part by the aberrant attentional processing of socially relevant information. Critically however, research has not assessed whether such aberrant attentional processing occurs during social-evaluative contexts characteristically feared in SAD. The current study presents a novel approach for the assessment of the visuocognitive biases operating in SAD during a social-evaluative stressor. For this task, clinically socially anxious participants and controls were required to give a brief impromptu speech in front of a pre-recorded audience who intermittently displayed socially positive or threatening gestures. Participant gaze at the audience display was recorded throughout the speech. Socially anxious participants exhibited a significantly longer visual scanpath, relative to controls. In addition, socially anxious participants spent relatively longer time fixating at the non-social regions in between and around the confederates. The findings of the present study suggest that SAD is associated with hyperscanning and the attentional avoidance of social stimuli.

. Introduction Social anxiety disorder (SAD) is a relatively common and debilitating psychiatric disorder characterized by an excessive fear of negative social evaluation, with an estimated lifetime prevalence of 12.1% (Kessler et al., 2005). Cognitive accounts suggest that SAD is maintained by biases in information processing (Williams et al., 1997 and Mogg and Bradley, 1998). It has been suggested that the fear of negative evaluation in SAD may be reflected in an excessive vigilance for threat cues in the environment which may confirm such a fear (Beck et al., 1985 and Beck and Clark, 1997). While a large body of research has confirmed that social anxiety is associated with an attentional bias to threat (cf. Bar-Haim et al., 2007), several studies have also observed that socially anxious individuals may avoid such information (Mansell et al., 1999, Chen et al., 2002 and Stirling et al., 2006). A two-stage vigilance–avoidance model of attentional bias has been proposed to integrate these findings (Mogg et al., 1997 and Amir et al., 1998). According to this model, socially anxious individuals are posited to be initially vigilant for threat, but subsequently avoid such information in an attempt to alleviate anxious symptoms. Previous attentional bias research has typically employed reaction time based assessments, such as the dot probe task (MacLeod et al., 1986). In a typical task, threat–neutral stimulus pairs are briefly presented, and then replaced by a probe appearing in the location vacated by one of the two stimuli. Socially anxious individuals have been found to be faster to respond to probes appearing in threat, relative to neutral, locations, which is taken to reflect an attentional bias to threat (Asmundson and Stein, 1994 and Musa et al., 2003). While such reaction time based tasks have been useful in establishing the presence of attentional bias in SAD, they offer limited flexibility in the assessment of the more dynamic patterns of attention. Gaze-based measures of attentional bias have been increasingly utilized, as eye tracking may provide a relatively direct measure of visual attention in real time (Duchowski, 2002). While probe tasks typically index biases in covert attention, eye tracking provides a measure of foveal vision, providing an assessment of biases in overt attention (Armstrong and Olatunji, 2012). Recent attentional bias research, incorporating gaze-based measures, has demonstrated that anxious individuals have a greater propensity, or are faster, to initially orient gaze towards threat stimuli (Mogg et al., 2000, Calvo and Avero, 2005, Garner et al., 2006 and Armstrong and Olatunji, 2012), consistent with the notion that social anxiety is associated with an attentional bias to threat. Previous studies have additionally assessed the total fixation time towards socially relevant stimuli in order to examine the preferences in selective attention over time. Clinically socially anxious individuals, in comparison to controls, have shown reduced total fixation time to emotional social stimuli (Chen et al., 2012 and Weeks et al., 2013). Anxious non-clinical individuals have exhibited similar total fixation time reductions towards threat (Rohner, 2002 and Calvo and Avero, 2005). The findings suggest that SAD may be associated with the attentional avoidance of emotional social stimuli, when considering the overall attentional preference for a given period of stimulus presentation. While the aforementioned research demonstrates the biased attentional processing with regard to emotional stimuli, it is further possible that SAD may be associated with a general aberration of the visual scanpath. The visual scanpath is the trace of eye movements made while viewing a complex stimulus, and consists of a sequence of fixations and saccades. While several processes influence the visual scanpath, it may essentially be taken to reflect the manner in which information is attended to, reappraised and integrated (Toh et al., 2011). Hyperscanning for instance, which is the extensive and potentially excessive monitoring of visual information in the environment, has been suggested to reflect a vigilant strategy (Green et al., 2003 and Horley et al., 2003). In terms of fixations and saccades, hyperscanning may be reflected by saccades of greater amplitude, and attenuated fixations with regard to duration and quantity, which subsequently may result in a longer scanpath length. Consistent with this notion, clinically socially anxious individuals, in comparison to controls, have exhibited longer scanpaths and reduced fixation duration while viewing emotional faces (Horley et al., 2003 and Horley et al., 2004). Similarly, the scanning of threatening faces has been associated with a greater mean spatial distance between consecutive fixations (Green et al., 2003). While existing research has incorporated gaze-based measures in order to assess the biased attentional processes in SAD, minimal research has been conducted to assess whether attentional avoidance and hyperscanning operate in SAD during conditions of social-evaluative stress. Given that such conditions are characteristically feared in SAD, it is imperative to ascertain whether the attentional biases observed during controlled laboratory experiments indeed occur during practically relevant social-evaluative tasks. The speech task has commonly been used in social anxiety research, as it necessitates social performance and exposes the participant to potential negative social evaluation. Socially anxious individuals have been found to give negative self-appraisals of their speech performance, experience heightened distress, and also exhibit greater startle reactivity in response to the speech task (Rapee and Lim, 1992, Cornwell et al., 2006 and Cornwell et al., 2011). We have recently shown that the implementation of eye tracking during a speech task may yield crucial insights into the attentional processes which occur during conditions of social-evaluative stress (Alvares et al., 2012 and Lowe et al., 2012). In this task, broadly referred to as the Sydney Eye Movement and Speech Assessment (SEMSA), participants give a brief speech in front of a pre-recorded audience of confederates, who intermittently display positive and threatening social gestures, while eye gaze is recorded. The audience display presented during the SEMSA may provide a useful socially relevant stimulus to examine visual attention. However, this assessment has yet to be applied to SAD-related attentional anomaly. Hence, the present study sought to determine whether aberrant attentional processing is associated with SAD while processing dynamic social information during a practically relevant task. In light of previous research, it was predicted that clinically socially anxious individuals, relative to controls, would exhibit hyperscanning of the audience display, indicated by an increased scan length during the speech. To further examine the component factors which contribute to the visual scanpath, we additionally considered the mean distance between consecutive fixations (MDBF), total fixation count and the mean fixation duration. It was anticipated that if SAD was associated with a longer scan length, such hyperscanning may further be associated with a greater MDBF, reduced total fixation count, or shortened mean fixation duration. We additionally assessed the relative distribution of attention towards the audience display. If SAD is associated with the avoidance of sustaining attention towards social stimuli throughout the speech, reduced total fixation time toward social display regions and greater total fixation time at the remaining non-social regions of the audience display are expected.

Results 3.1. Group characteristics As expected, SAD participants, in comparison to controls, reported higher scores on all questionnaire measures, smallest F(1,33)=9.42, P=0.004. Groups did not differ on mean age, F(1,33)=1.04, P=0.316, or gender, χ2(1, N=35)=0.85, P=0.358. A group (SAD vs. control) by time (pre vs. post) ANOVA was conducted on SUDS scores. Participants reported higher subjective distress immediately before, compared to after the speech, F(1,33)=6.04, P=0.019. SAD participants reported higher subjective distress compared to controls, F(1,33)=10.84, P=0.002. No group by time interaction was observed, F(1,33)=1.41, P=0.244. Statistics are provided in Table 1. Table 1. Means, standard deviations and F-scores for questionnaires completed by SAD and control participants. Questionnaire Control SAD F(1,33) M S.D. M S.D. DASS-21 depression 8.00 10.11 18.78 10.67 9.42⁎⁎ DASS-21 anxiety 6.00 8.87 20.09 6.74 27.79⁎⁎⁎ DASS-21 stress 11.11 9.03 22.91 9.62 14.00⁎⁎⁎ LSAS social fear 9.78 5.76 21.10 7.21 26.50⁎⁎⁎ LSAS performance fear 7.81 6.42 21.38 6.01 41.55⁎⁎⁎ LSAS social avoidance 7.06 4.32 18.23 8.06 26.55⁎⁎⁎ LSAS performance avoidance 6.24 5.80 19.71 6.50 41.88⁎⁎⁎ LSAS total 30.89 20.84 80.42 25.06 40.62⁎⁎⁎ SUDS pre 34.28 22.03 58.76 22.92 10.39⁎⁎ SUDS post 29.00 20.05 43.63 20.66 4.52⁎ Age 23.78 3.14 25.24 5.15 1.04 ⁎ p<0.05. ⁎⁎ p<0.01. ⁎⁎⁎ p<0.001. Table options 3.2. Visual scanpath Due to significant violations of normality based on the Kolmogorov–Smirnov and Shapiro–Wilk tests, a logarithmic transformation was applied to scan length scores. A one-way ANOVA was calculated to assess group (SAD vs. control) differences on scan length during the speech. SAD participants were found to have significantly greater mean scan length in comparison to controls, F(1,33)=4.27, P=0.047. Given that this group difference in scan length may derive from a longer MDBF, reduced total fixation count, or shortened mean fixation duration, one-way ANOVAs were conducted to assess group (SAD vs. control) difference on these variables. Mean fixation duration values were logarithmically transformed due to violations of normality based on the Kolmogorov–Smirnov and Shapiro–Wilk tests. No significant group differences were found, the largest F(1,33)=2.41, P=0.130, see Table 2 for descriptive statistics. Table 2. Visual scanpath measures and inferential statistics for SAD and control participants. Measure Control SAD F(1,33) M S.D. M S.D. Scan length (m) 57.41 32.23 74.68 24.88 4.27⁎ Mean distance between fixations (m) 0.18 0.06 0.21 0.05 2.41 Total fixation count 316.50 106.70 354.88 104.83 1.15 Mean fixation duration (ms) 333.99 118.60 278.83 42.59 2.10 ⁎ p<0.05. Table options 3.3. Distribution of attention Total fixation times to positive, threat, neutral and non-face regions of the audience display were calculated to examine the distribution of attention throughout the speech. Descriptive statistics are provided in Table 3. Due to significant violations of normality based on the Kolmogorov–Smirnov and Shapiro–Wilk tests, a logarithmic transformation was applied to all total fixation time scores. A mixed-design ANOVA considering group (SAD vs. control) and region (positive vs. threat vs. neutral vs. non-face) on total fixation time was conducted. Mauchly׳s test indicated a violation of the assumption of sphericity, χ2(5)=30.93, P<0.001, hence Greenhouse–Geisser adjusted values have been reported (ε=0.67). No main effect of group was observed, F(1, 33)=0.89, P=0.354, indicating that groups did not differ in the total fixation time recorded throughout the speech. A main effect of region was found, F(1.99,65.79)=7.96, P=0.001. Pairwise comparisons with Bonferroni corrections confirmed that all participants exhibited greater total fixation time at threat, Mdiff=−0.58, SEdiff=0.19, P=0.032, and neutral regions, Mdiff=−0.65, SEdiff=0.17, P=0.002, compared to non-face regions. Critically, a significant group by region interaction was evident, F(1.99,65.79)=3.81, P=0.027, illustrated in Fig. 2. Pairwise comparisons with Bonferroni corrections confirmed that socially anxious individuals showed greater total fixation time at non-face regions compared to controls, Mdiff=−0.75, SEdiff=0.29, P=0.013. However, groups did not differ in total fixation time to positive, neutral or threat regions, the largest Mdiff=0.15, SEdiff=0.25, P=0.568. Table 3. Total fixation time scores for SAD and control participants. Total fixation time (s) Control SAD M S.D. M S.D. Positive 24.42 16.38 22.13 11.76 Threat 33.56 20.78 25.17 11.50 Neutral 32.57 19.73 28.85 11.70 Nonface 14.66 15.18 23.00 13.91 Table options Mean total fixation time to positive, threat, neutral and non-face display ... Fig. 2. Mean total fixation time to positive, threat, neutral and non-face display regions during the speech for SAD and control participants. Figure options 3.4. Analysis of initial neutral period The initial 50 s period of the audience presentation which did not contain any emotional gestures was additionally considered. Social anxiety was not found to modulate any of the gaze variables of interest, the largest F(1.95,64.20)=2.09, P=0.134. 3.5. Potential influence of educational levels To address the potential extraneous influence that educational levels may have on the attentional measures, participant educational levels were initially dummy coded (1=currently engaged with or completed tertiary education; 0=secondary education attained). Partial correlations were then calculated between educational level, and all total fixation time and scan length measures, while controlling for social anxiety group influences. Educational level was not found to significantly influence any of the attentional measures, largest r=−1.78, P=0.329.