eACS: کنترل توجه در حضور احساسات

Abstract We present a questionnaire – The Emotional Attentional Control Scale (eACS) – An adaptation of the original Attentional Control Scale (ACS) that assesses the voluntary control of attention. A low score on the ACS has been associated with high levels of anxiety and depression. As the ACS items are affectively neutral, some people scoring high on the ACS may still show low levels of attentional control (AC) in more emotionally-demanding situations. We propose that the eACS, which focuses on the emotional modulation of AC, may explain additional variance in AC deficits associated with psychopathology. The eACS showed one general factor for emotional AC. Both the ACS and eACS showed a negative correlation with trait anxiety (STAI-T) and depressive symptoms scores (BDI-II). In regression analyses, when accounting for the shared variance between the STAI-T and BDI-II, both the eACS and ACS explained independent variance in STAI-T scores (β = −.23, and β = −.15, p < .001, respectively). The eACS has clear benefit in measuring AC deficits that are associated with psychopathology. Individual differences in AC in emotionally-demanding situations could be an important, and as yet underappreciated, aspect of psychopathology. Recommendations for future research are given.

1. Introduction Resource constraints in the attentional system limit the ability to process all environmental information at once. Instead, people must selectively attend to information that is relevant to their current goals (Eysenck, Derakshan, Santos, & Calvo, 2007). In the presence of threat, people often selectively attend to the sources of the threat at the expense of attending to other things (Berggren & Derakshan, 2013). This selectivity is typically more pronounced in high- versus low-anxious people (Massar, Mol, Kenemans, & Baas, 2011). High-anxious people show deficits in attentional control (AC) with enhanced orienting towards the sources of threat, which in turn disrupts directing attention elsewhere (Derryberry & Reed, 2002). Sources of threat can be physical or social (Garner, Mogg, & Bradley, 2006), so AC might be an important transdiagnostic mechanism underlying psychopathology including anxiety and depression (Ólafsson, Smári, Guðmundsdóttir, Olafsdóttir, Harðardóttir, & Einarsson (2011)). A widely-used self-report assessment of AC is the Attentional Control Scale (ACS; Derryberry & Reed, 2002). It uses affectively neutral items relating to mundane tasks and emotionally insignificant distractions. These items assess individual differences in the ability to focus attention and shift it between tasks. Research exists in support of the relationship between AC deficits and depression and anxiety (see Ólafsson et al., 2011), but Derryberry and Reed (2002) showed evidence of a subgroup of high trait anxious people who also exhibit good AC as assessed by the ACS. Not all people who exhibit anxiety or depression also show deficits in AC in affectively neutral tasks (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & Van IJzendoorn, 2007). In certain situations, some people with high trait anxiety can exhibit compensatory control allowing them to inhibit their dominant attentional tendencies and to perform comparably to their low-anxiety counterparts (Berggren & Derakshan, 2013). For example, relative to non-anxious controls, anxious people have shown heightened activation of the brain’s fronto-parietal attentional system in tasks such as the Stroop. This is thought to reflect increased attempts to inhibit task-irrelevant information (Basten, Stelzel, & Fiebach, 2011). However, there are limits to the compensatory mechanisms used by high anxious people (Eysenck et al., 2007). When resources are further constrained, as in stressful or emotionally provocative situations, some trait-anxious people may have difficulty controlling their attention and preventing distraction. Performance in a demanding, emotional task such as the emotional Stroop task may then reflect that of someone low in AC even if they have a high score on the affectively-neutral ACS (Reinholdt-Dunne, Mogg, & Bradley, 2009). High trait-anxious/high ACS scorers who show low AC in emotional situations may be as vulnerable to developing attentional biases and anxious and depressive disorders as their high trait-anxious/low ACS scoring counterparts. However, perhaps for high trait-anxious/high ACS scorers, deficits in AC may be less apparent in mundane tasks than they are for low ACS scorers because of their use of compensatory neural mechanisms (Reinholdt-Dunne et al., 2009). The ACS might be inadequate in explaining much of the variance in performance of trait-anxious people in emotionally-demanding situations. Indeed, the relationship between trait anxiety and attentional effects in an affective task remained significant even when statistically controlling for ACS scores (Massar et al., 2011). If differences in AC explained a large portion of the variance in performance this would not be the case. There is value, therefore, in developing an assessment of attention in emotional situations. Herein we present the development and analysis of an adaptation of the original ACS. Whereas the original ACS was concerned with assessing voluntary attentional processes in neutral situations (Derryberry, 2002), the Emotional Attentional Control Scale (eACS) is concerned with attention in the presence of emotion. We used multiple regression analyses to determine the associations of emotional AC with anxiety and depression. Given the well-known association between depression and anxiety, it is important to consider how their shared and unshared variances overlap with AC. To explore the unique variance in trait anxiety and depression that is explained by AC, the regression analyses accounted for the shared variance between anxiety and depression (Ólafsson et al., 2011). We expected the eACS to have one general factor related to the overall effect of emotion on AC and for this general factor to correlate with the original ACS. However, although the ACS should explain significant variance in trait anxiety and depressive symptoms, we expected that the eACS would explain additional, independent, variance, as well as correlate strongly with anxiety and depression even for those who scored highly on the original ACS.

. Results 3.1. Exploratory factor analysis Parallel analysis on the eACS showed three factors with eigenvalues of 6.14, 2.50 and 1.31, exceeding the eigenvalues of the raw data, which were 1.43, 1.35 and 1.28, respectively. From this, we extracted one, two and three factors using EFA. Table 2 presents the results of the EFA. A comparison between the factor loadings in Table 2 and the content of the eACS items (Table 1) suggested that a bifactor model might provide the best fit for the data. In other words, it appeared that there was a high degree of shared variance among all of the eACS items, in addition to two potential group factors. To test this hypothesis and to re-express the model from the EFA, we performed a confirmatory factor analysis (CFA) using the same sample and a bifactor structure. Several items were excluded on the basis of high correlation with the content of other items and/or loading poorly on the general factor (see Table 2 dispositions for details on exclusion). Thus, 14 items remained for inclusion in the bifactor model. This model had a good fit for the data with CFI = .98 and TLI = .97 and showed a good RMSEA (.07, p < .01) and these items had strong internal consistency when considered as one scale (.87). See Fig. 1 for the CFA factor loadings. Table 2. Exploratory factor analysis (EFA). F1 F1 F2 F1 F2 F3 Reason for exclusion eACS1 0.74 0.63 0.26 0.65 0.01 0.21 eACS 2 0.40 0.39 0.07 0.56 0.14 0.06 Poor loading on g factor eACS 3 0.72 0.68 0.14 0.71 0.02 −0.07 eACS 4 0.67 0.84 −0.36 0.16 0.74 −0.01 eACS 5 0.84 0.76 0.22 1.01 −0.23 −0.01 eACS 6 0.88 0.79 0.23 0.96 −0.14 0.04 eACS 7 0.54 0.16 0.59 0.13 −0.01 0.61 eACS 8 0.72 0.65 0.19 0.66 0.02 0.15 eACS 9 0.67 0.82 −0.30 −0.00 0.86 0.14 Correlation with eACS4 eACS 10 0.60 0.68 −0.10 0.27 0.48 0.10 eACS 11 0.53 0.00 0.75 0.04 −0.11 0.75 eACS 12 0.57 −0.06 0.84 −0.16 0.01 0.90 eACS 13 0.59 0.03 0.78 0.00 −0.04 0.80 eACS 14 0.50 0.04 0.67 −0.04 0.03 0.71 eACS 15 0.53 0.56 0.01 0.49 0.13 0.00 eACS 16 0.57 0.62 −0.03 0.48 0.22 −0.00 Poor loading on g factor eACS 17 0.70 0.72 0.02 0.65 0.16 −0.00 eACS 18 0.56 0.71 −0.24 0.24 0.57 −0.30 Correlation with eACS4 Note: Results from EFA of Emotional Attentional Control Scale (eACS) items, testing 1, 2 and 3 factor solutions, and information on exclusions. GEOMIN rotated factor loadings greater than or equal to 0.3 are displayed in bold. Table options Structural model from a confirmatory factor analysis of Emotional Attentional ... Fig. 1. Structural model from a confirmatory factor analysis of Emotional Attentional Control Scale items using a bifactor model with 2 factors for reverse-scored (F1) vs non-reverse-scored (F2) items, and one general factor for emotional attentional control. Figure options 3.2. Attentional control and emotion Total eACS scores were computed using the 14 items included in the previous analyses (see Table 2) (Mean: 35.44; SD: 6.42). Comparison with the original ACS showed large positive correlations with total ACS scores (Mean: 47.03; SD: 7.75) (r = .62, p < .001) and the focusing (Mean: 17.14; SD: 3.77) and shifting sub-factors (Mean: 28.15; SD: 4.05) (r = .57 and r = .45 respectively, p < .001). As expected, people who reported being more able to focus and shift their attention voluntarily were more likely to report that they could control their attention during and around emotional situations and stimuli. 3.3. Emotional attentional control, anxiety and depression There were large negative correlations between scores on the eACS and STAI-T (r = −.53, p < .001) and the ACS and STAI-T (r = −.50, p < .001). BDI-II scores also correlated with eACS and ACS scores (r = −.38 and r = −.36 respectively, p < .001). Greater trait anxiety and depressive symptoms were associated with lower control of attention in emotional situations as well as lower control in neutral situations. STAI-T (Mean: 43.48; SD: 9.93) and BDI-II (Mean: 11.05; SD: 6.68) scores also showed a large positive correlation (r = .70, p < .001). Table 3 presents the regression analyses. In predicting trait anxiety, neither age nor gender explained a significant amount of variance at any step in the analysis. BDI-II score remained a significant predictor of STAI-T score at all steps. ACS score also remained a significant predictor throughout, but the amount of variance in STAI-T score that it explained was reduced with the inclusion of the eACS. Both the ACS and eACS were significant predictors of STAI-T score, although their interaction was not significant. There was some shared variance between the two measures of AC, but both the ACS and eACS explained a significant amount of independent variance. The eACS explained more of the variance in STAI-T scores than the original ACS. Table 3. Hierarchical regression analysis. B SE (B) β Dependent variable: STAI-T Step 1 (ΔR2 = .00) Age −.26 .39 −.03 Gender −1.36 1.10 −.05 Step 2 (ΔR2 = .49) BDI .84 .06 .57∗∗∗ Step 3 (ΔR2 = .07) ACS −.19 .06 −.15∗∗ Step 4 (ΔR2 = .03) eACS −.35 .07 −.23∗∗∗ Step 5 (ΔR2 = .00) ACS∗eACS interaction −.01 .01 −.05 Dependent variable: BDI-II Step 1 (ΔR2 = .02) Age .32 .29 .05 Gender 2.06 .81 .11∗∗ Step 2 (ΔR2 = .48) STAI-T .47 .03 .69∗∗∗ Step 3 (ΔR2 = .00) ACS −.02 .05 −.03 Step 4 (ΔR2 = .00) EACS −.01 .06 −.01 Step 4 (ΔR2 = .00) ACS∗eACS interaction .00 .01 .02 Note: Results from two regression analyses using emotional attentional control and neutral attentional control measures to predict anxiety and depression scores. All B, SE (B) and β values are from the final step of each regression analysis. ∗p < .05; ∗∗p < .01; ∗∗∗p < .001. Table options When predicting depressive symptoms, gender was a significant predictor, but age was not. A one-way ANOVA revealed that females showed significantly higher scores on the BDI-II (Mean: 11.40, SD: 6.70) than did males (Mean: 9.12, SD: 6.29), F(1, 310) = 4.80, p < .05. STAI-T scores explained a large amount of the variance in BDI-II scores. Neither ACS nor eACS scores were significant predictors of BDI-II scores at any step in the regression (β = .01, and β = −.01, p > .05, respectively), nor was their interaction term. When all other predictors were removed from the regression, the ACS and eACS now explained a significant amount of BDI-II variance (β = −.50, and β = −.53, p < .001, respectively). As with STAI-T scores, the regression coefficient for the eACS was slightly larger than the coefficient for the original ACS. When selecting participants high in AC – that is, they scored above the median ACS score (48 or above) – a significant negative correlation was still present between eACS and STAI-T and BDI-II scores (r = −.42, p < .001 and r = −.27 p < .005, respectively). The mean STAI-T score for high ACS scorers was 38.34 (SD: 8.51) whereas when selecting all participants it was 43.48 (SD: 9.93). An increase was also apparent for mean BDI-II scores from high ACS scorers (Mean: 8.55; SD: 5.83) to all participants (Mean: 11.05; SD: 6.68).