عزت نفس پایین ضمنی و ارتباطات اتوماتیک ناکارآمد در اختلال اضطراب اجتماعی

Abstract Background and Objectives Negative automatic associations towards the self and social cues are assumed to play an important role in social anxiety disorder. We tested whether social anxiety disorder patients (n = 45) showed stronger dysfunctional automatic associations than non-clinical controls (n = 45) and panic disorder patients (n = 24) and whether there existed gender differences in this respect. Methods We used a single-target Implicit Association Test and an Implicit Association Test to measure dysfunctional automatic associations with social cues and implicit self-esteem, respectively. Results Results showed that automatic associations with social cues were more dysfunctional in socially anxious patients than in both control groups, suggesting this might be a specific characteristic of social anxiety disorder. Socially anxious patients showed relatively low implicit self-esteem compared to non-clinical controls, whereas panic disorder patients scored in between both groups. Unexpectedly, we found that lower implicit self-esteem was related to higher severity of social anxiety symptoms in men, whereas no such relationship was found in women. Conclusions These findings support the view that automatic negative associations with social cues and lowered implicit self-esteem may both help to enhance our understanding of the cognitive processes that underlie social anxiety disorder.

. Introduction Individuals who suffer from social anxiety disorder (SAD) typically experience an intense and persistent fear of social situations in which they are exposed to unfamiliar people or to the possible scrutiny by others (APA, 2000). According to the cognitive model of social anxiety of Wells and Clark (1997), socially anxious individuals interpret social situations as threatening, because of negative beliefs about their selves and dysfunctional assumptions about their social performance; together with excessively high standards for social performance. Negative self-beliefs typically are unconditional negative statements about the self (e.g., “I'm stupid” or “I'm a failure”). Dysfunctional assumptions concerning social performance take the form of conditional beliefs about possible negative consequences of social behaviours (e.g., “if they see my anxiety, then they will think I'm a failure”). In support of the role of these dysfunctional cognitions in SAD, research showed that high socially anxious individuals indeed display more negative self-statements (e.g., Beidel, Turner, & Dancu, 1985; Cacioppo, Glass, & Merluzzi, 1979; Dodge, Hope, Heimberg, & Becker, 1988) and lower levels of self-esteem in social situations (e.g., Bouvard et al., 1999; Tanner, Stopa, & De Houwer, 2006) than low socially anxious individuals. According to the multi-process model of anxiety, not only dysfunctional assumptions and negative self-beliefs (belonging to so-called ‘rule-based processes’) constitute the cognitive vulnerability to anxiety disorders, but also dysfunctional associative processes play an important role in the development and maintenance of anxiety disorders (Ouimet, Gawronski, & Dozois, 2009). In response to anxiety-relevant stimuli, threat-related associations are thought to be directly activated via the spreading of activation from one concept to associated concepts in memory. Subsequently, the input from the associative system is assumed to be used for more deliberate, rule-based mental processing (Strack & Deutsch, 2004) which involves the more rational analysis of factual relationships between concepts. Associative and rule-based information processing systems are thought to jointly influence other cognitive processes (e.g., negative interpretive bias) and behaviours (e.g., attention bias or avoidance behaviours) that work together in a way to aggravate and/or maintain the anxiety disorder. Based on the model of Wells and Clark (1997) two types of automatic associations seem to be most important for SAD: first of all, negative automatic associations towards the self (i.e. implicit self-esteem) and secondly, associations between social cues and negative outcomes of social performance such as failure or rejection. In line with the presumed role of automatic associations in anxiety, automatic associations were found to predict experimentally-provoked anxiety behaviours in unselected student samples in the laboratory (e.g., Asendorpf, Banse, & Mücke, 2002; Egloff & Schmukle, 2002; Spalding & Hardin, 1999). In addition, three studies showed that automatic self-anxious associations were related to having an anxiety disorder diagnosis as well as to the maintenance and onset of anxiety disorders over time (Glashouwer & de Jong, 2010; Glashouwer, de Jong, & Penninx, 2011, 2012). Furthermore, one study showed that socially anxious students were characterized by stronger automatic self-anxious associations than non-anxious students and that these associations seemed to reduce following treatment (Gamer, Schmukle, Luka-Krausgrill, & Egloff, 2008). Up to now, only a few studies looked specifically at implicit self-esteem and associations with social cues in SAD and, in addition, these studies mainly relied on analogue samples. Two studies showed that high socially anxious female students indeed were characterized by relatively low implicit self-esteem (de Jong, 2002; Tanner et al., 2006). In addition, recently it was shown that for adolescent girls, but not boys, lower implicit self-esteem was related to more social anxiety symptoms (de Jong, Sportel, de Hullu, & Nauta, 2012). Furthermore, there is some evidence supporting the view that also threat-related automatic associations with social cues may be involved in SAD. High socially anxious female students displayed stronger negative automatic associations with social cues than low-anxious participants (de Jong, Pasman, Kindt, & van den Hout, 2001). Similar results were found in an adolescent sample, showing that social cues automatically elicited relatively strong threat-related associations in high compared to low socially anxious adolescents (de Hullu, de Jong, Sportel, & Nauta, 2011). Although the available evidence suggests that both dysfunctional automatic associations with respect to the self and social cues seem to be involved in SAD, some important questions still remain unanswered. Prior studies in this field typically compared analogue groups of high socially anxious individuals with low socially anxious individuals. Since there may be quantitative as well as qualitative differences between analogue and clinical samples (Emmelkamp, 1982), it seems important to replicate these findings in a clinical sample. In addition, prior studies did not include clinical-control groups, leaving open the question whether differences regarding automatic associations can indeed be attributed to SAD or have to be seen as more general characteristics shared among several anxiety disorders. Furthermore, most studies until now relied on female samples. However, there might be gender differences in the relationship between dysfunctional automatic associations and social anxiety symptoms. Earlier work showed that women are more likely than men to base their judgements on intuitions and gut impressions (e.g., Pacini & Epstein, 1999). Perhaps this could mean that, compared to men, women tend to rely more on their automatic associations as a guideline for their behaviour and self-judgements (cf. Pelham et al., 2005). Consequently, automatic associations could have stronger predictive validity in women than in men and one study indeed showed findings in this direction in a group of adolescents (de Jong et al., 2012). Finally, prior studies usually focused on one kind of automatic associations, making it impossible to examine independent contributions of different types of automatic associations for social anxiety symptoms. The main goal of the present study is to test whether SAD patients show stronger dysfunctional automatic associations (regarding self and social cues) than both clinical and non-clinical controls and whether there exist differences for females and males in this respect. Therefore, we included a clinical sample of treatment-seeking SAD patients, a non-anxious control group and a clinical sample of treatment-seeking panic disorder patients as a clinical control group. We hypothesize that automatic associations with social cues are more dysfunctional in the SAD group than in both control groups. In addition, we expect the SAD group to have a lower implicit self-esteem than controls. For the panic disorder group we have no clear expectations with respect to implicit self-esteem. Furthermore, we expect automatic associations to have stronger predictive validity for social anxiety symptoms in women than in men. Finally, an additional strength of the present study is that it is the first to examine two types of automatic associations in one socially anxious sample, allowing to test whether both types of associations are independently related to social anxiety symptom severity.

Results 3.1. Descriptive statistics LSAS data was missing for one participant of the socially anxious group. Means and standard deviations for the different groups on the SPAI, the LSAS, the BDI, the social stIAT, the self-esteem IAT and overall percentage of errors on stIAT/IAT are reported in Table 3. As expected, the socially anxious group scored significantly higher than the control groups on social anxiety symptom measures during baseline (SAD vs. NCC: SPAI: t(88) = 12.88, p < .001; LSAS fear: t(65.66) = 12.72, p < .001; LSAS avoidance: t(59.93) = 8.51, p < .001; SAD vs. PD: SPAI: t(67) = 4.75, p < .001; LSAS fear: t(66) = 4.96, p < .001; LSAS avoidance: t(66) = 3.35, p = .001). Table 3. Means and standard deviations of variables as a function of group. Variables Group SAD NCC PD SPAI, social phobia score 125.13 (33.21) 44.09 (26.03) 82.47 (39.63) SPAI, agoraphobia score 27.57 (16.87) 7.96 (9.00) 38.96 (14.99) LSAS, fear score 40.41 (13.33) 11.44 (7.17) 23.00 (14.76) LSAS, avoidance score 31.55 (14.74) 10.80 (6.74) 19.38 (13.47) BDI, total score 16.20 (9.88) 2.78 (4.36) 13.83 (6.86) Social stIAT, d-measure 0.46 (0.37) 0.64 (0.29) 0.66 (0.30) Social stIAT, % error trials overall 6.67 (4.88) 7.01 (4.73) 6.58 (3.62) Self-esteem IAT, d-measure 0.43 (0.45) 0.75 (0.33) 0.63 (0.44) Self-esteem IAT, % error trials overall 7.02 (5.37) 7.65 (5.79) 8.30 (7.40) Note. stIAT = single target Implicit Association Test; IAT = Implicit Association Test; SPAI = Social Phobia and Anxiety Inventory; LSAS = Liebowitz Social Anxiety Scale; BDI = Beck Depression Inventory; SAD = social anxiety disorder; NCC = non-clinical control; PD = panic disorder. Table options 3.2. Group differences on automatic associations 3.2.1. Social stIAT An overview of mean response times for the different groups for each pairing of the social stIAT is provided in Fig. 1. A 3 Group (SAD, PD, NCC) × 2 Gender ANOVA on social stIAT showed a significant main effect for Group (F(2,108) = 3.35, p = .039, partial η2 = .06). However, the main effect of Gender, and the interaction between Group and Gender were non-significant (p's > 0.4). As expected, planned comparisons with t-tests revealed that the socially anxious group displayed significantly less positive automatic associations with social situations (i.e., a relatively fast response when ‘social situation’ and ‘negative outcome’ shared the same response key) than both the non-anxious control group (t(82.6) = 2.46, p = .016, d = 0.54) and the panic disorder group (t(67) = 2.17, p = .034, d = 0.53). The control group and the panic disorder group did not significantly differ on the social stIAT (t(67) = 0.26, p = .799). Mean latencies of the trials per pairing of the social stIAT: Social anxiety ... Fig. 1. Mean latencies of the trials per pairing of the social stIAT: Social anxiety disorder, non-clinical controls and panic disorder. Note. RT = reaction time; stIAT = single target implicit association test; stIAT latencies include 600 ms penalty for wrong responses; error bars: ±2 * SE. Figure options In addition, to test whether depressive symptoms might explain the observed group differences, we compared social stIAT scores for socially anxious individuals with and without depressive symptoms. Both a classification based on MINI diagnosis (n depressed = 18; n non-depressed = 27) and based on median split on the BDI (median = 13; n depressed = 21; n non-depressed = 24) did not show meaningful differences between depressed and non-depressed SAD individuals with regard to their automatic associations with social situations (t's < 0.5; p's > 0.6). Furthermore, we calculated the correlation coefficients between BDI scores and social stIAT scores. However, neither in the entire sample (r = −.16, p = .099) nor in the subgroups this relationship reached significance (SAD: r = −.03, p = .845; PD: r = −.37, p = .072; NCC: r = .16, p = .292). 3.2.2. Self-esteem IAT An overview of mean response times for the different groups for each pairing of the self-esteem IAT is provided in Fig. 2. A 3 Group (SAD, PD, NCC) × 2 Gender ANOVA on self-esteem IAT showed a significant main effect for Group (F(2,108) = 5.35, p = .006, partial η2 = .09). However, the main effect of Gender, and the interaction between Group and Gender were non-significant (p's > 0.3). Planned comparisons with t-tests revealed that the socially anxious group displayed significantly less positive implicit self-esteem (i.e., a relatively fast response when ‘me’ and ‘negative’ shared the same response key) than the control group (t(80.6) = 3.85, p < .001, d = 0.86). The difference between the socially anxious group and the panic disorder group only showed a non-significant trend (t(67) = 1.74, p = .09, d = 0.43). The control group and the panic disorder group did not significantly differ on implicit self-esteem (t(67) = 1.30, p = .20, d = 0.32). Mean latencies of the trials per pairing of the self-esteem IAT: Social anxiety ... Fig. 2. Mean latencies of the trials per pairing of the self-esteem IAT: Social anxiety disorder, non-clinical controls and panic disorder. Note. RT = reaction time; IAT = implicit association test; IAT latencies include 600 ms penalty for wrong responses; error bars: ±2 * SE. Figure options In addition, we compared self-esteem IAT scores for socially anxious individuals with and without depressive symptoms. Both the classification based on MINI diagnosis and median split on the BDI did not show meaningful differences on implicit self-esteem between depressed and non-depressed SAD individuals (t's < 0.9; p's > 0.4). Furthermore, we calculated the correlation coefficients between BDI scores and self-esteem IAT scores. We found a negative correlation in the entire sample indicating that individuals that suffer from more depressive symptoms tend to show more negative implicit self-esteem (r = −.32, p < .001). However, within each of the subgroups, the correlation between BDI and self-esteem IAT scores was smaller and did not reach significance (SAD: r = −.16, p = .284; PD: r = −.18, p = .414; NCC: r = −.21, p = .173). 3.3. Relationship between automatic associations and social anxiety symptoms (as a function of gender) Because the subscales of the LSAS were somewhat skewed to the right, we first performed a square root transformation on both subscales. Correlation coefficients were calculated between stIAT, IAT and, the self-report measures of anxiety symptoms (see Table 4). Next, we performed forced entry hierarchical regression analyses to explore whether having dysfunctional automatic associations on both the social stIAT and the self-esteem IAT independently contributed to the level of social anxiety symptoms. In addition, we included interaction effects with gender expecting that the regression model would be more sensitive to possible gender differences. Social anxiety symptoms (SPAI) were included as dependent variable and standardized social stIAT, self-esteem IAT and gender as independent variables in the first three steps respectively. In the fourth step, the interactions between standardized social stIAT and self-esteem IAT and the interactions with gender were added to the model (see Table 5, upper part). Results show there was only a main effect of self-esteem IAT on social anxiety symptoms, indicating that lower implicit self-esteem was related to higher social anxiety symptom severity. Main effects of social stIAT and gender were non-significant. In addition, we found that the interaction between self-esteem IAT and gender showed predictive validity for severity of symptoms, although the ΔR2 of step 4 did not reach significance. The interactions between social stIAT and self-esteem IAT and between social stIAT and gender were non-significant. Table 4. Correlation matrix stIAT, IAT, SPAI and LSAS (N = 114). Measure 2. 3. 4. 5. 6. 1. Social stIAT, d-measure .02 −.04 .01 −.20* −.08 2. Self-esteem IAT, d-measure – −.38** −.21* −.35** −.32** 3. SPAI, social phobia score – .65** .94** .85** 4. SPAI, agoraphobia score – .58** .55** 5. LSAS, fear score – .88** 6. LSAS, avoidance score – Note. stIAT = single target Implicit Association Test; IAT = Implicit Association Test; SPAI = Social Phobia and Anxiety Inventory; LSAS = Liebowitz Social Anxiety Scale. *p < .05; **p < .01. Please note that the negative correlations between automatic associations and symptom measures are in line with our expectations, since positive effects for self-associations indicate a relatively stronger automatic association between me/social situation and positive. Table options Table 5. Summary of hierarchical regression analysis for variables predicting social anxiety measured with the social phobia subscale of the social phobia and anxiety inventory (SPAI; N = 114). Model Step Predictor B SE B β p All predictors includeda 1 social stIAT −2.11 4.55 −0.04 .644 2 social stIAT −1.84 4.23 −0.04 .665 self-esteem IAT −18.13 4.23 −0.38 <.001 3 social stIAT −2.18 4.24 −0.05 .608 self-esteem IAT −18.20 4.23 −0.38 <.001 gender 4.37 4.25 0.09 .306 4 social stIAT −1.67 4.26 −0.05 .679 self-esteem IAT −16.78 4.30 −0.38 <.001 gender 4.18 4.23 0.09 .326 stIAT × IAT interaction −1.41 4.34 −0.03 .746 IAT × gender interaction 9.39 4.45 0.19 .037 stIAT × gender interaction −1.13 4.39 −0.02 .797 Only significant predictors includedb 1 self-esteem IAT −18.23 4.22 −0.38 <.001 gender 4.19 4.22 0.09 .322 2 self-esteem IAT −16.59 4.22 −0.34 <.001 gender 4.11 4.15 0.09 .324 IAT × gender interaction 9.49 4.37 0.19 .032 Note. stIAT = single target Implicit Association Test; IAT = Implicit Association Test. a Step 1: R2 = .002, p = .644; Step 2: ΔR2 = .14, p < .001; Step 3: ΔR2 = .008, p = .306.; Step 4: ΔR2 = .035, p = .206. b Step 1: R2 = .15, p < .001; Step 2: ΔR2 = .035, p = .032. Table options Because we assumed that the number of parameters in the regression model could have reduced the power of the regression model, we decided to repeat the analysis with only the significant predictors: implicit self-esteem, gender, and the interaction term of implicit self-esteem and gender (see Table 5, lower part). Results showed that now the ΔR2 of adding the interaction effect did reach significance. To interpret these findings, the predicted values for the interaction term are presented in Fig. 3. Unexpectedly, the figure indicates that lower implicit self-esteem was related to higher social anxiety symptom severity in men, whereas no such relationship was found in women. Interaction effect of implicit self-esteem on symptoms of social anxiety as ... Fig. 3. Interaction effect of implicit self-esteem on symptoms of social anxiety as indexed by the social phobia and anxiety inventory (SPAI) for men (N = 67) and women (N = 47). Figure options Since correlational analyses showed that stIAT was only significantly related to LSAS fear, we repeated the regression analyses with LSAS fear as dependent variable for exploratory purposes. In this analysis, both social stIAT and self-esteem IAT showed independent predictive validity for the severity of social anxiety symptoms (stIAT: β = −0.19, p = .031; IAT: β = −0.34, p < .001). Other outcomes were similar to the SPAI, except for the interaction between self-esteem IAT and gender which appeared to be only marginally significant (β = 0.17; p = .056), yet in the same direction.