Executive attention control impairments and social anxiety symptoms in children

Introduction

The primary aim of the present study was to investigate the role of executive attention control in children with heightened levels of social anxiety symptoms. Specifically, the study assessed whether social anxiety levels in children were associated with (1) actual performance in an objectively measured executive attention control task, (2) perception of performance on the task, and (3) a subjective measure of executive attention control. A secondary aim of the present study was to examine the unique contribution of the different executive attention control types to predict social anxiety. We chose to study social anxiety for several reasons. First, social anxiety is highly prevalent in typically developing children, with a peak onset in early adolescence (Kessler et al., 2005). Second, childhood social anxiety disorder is associated with lower well-being and social functioning (Cillessen & Mayeux, 2004), including impairments in education and early dropout (Stein & Kean, 2000). Knowing more about the underlying attentional mechanisms that drive social anxiety is important to direct interventions to help overcome social anxiety. Specifically, objective versus subjective evaluations of social or performance situations might help us to understand whether children with social anxiety show actual or perceived deficits in attention.

Executive attention control is a heterogeneous concept which can be challenging to define. It includes the abilities to inhibit processing of task-irrelevant information, to focus on relevant information, and to shift attention when necessary. According to Attentional Control Theory, executive attention control is impaired in children with heightened levels of anxiety (Eysenck & Derakshan, 2011; Eysenck et al., 2007). Anxiety-prone children with poorer executive attention control are thought to have greater difficulties in inhibiting negative thoughts and regulating emotional responses, relative to those with better executive attention control (Rueda et al., 2004). This impairment in executive attention control is purported to explain greater levels of attentional bias for emotional stimuli in adults and children with high levels of anxiety (Eysenck et al., 2007). Attentional bias is the allocation of processing resources to certain stimuli or aspects of the environment, generally at the expense of processing other concurrent stimuli or aspects. Empirical findings support this notion in adults (e.g., Reinholdt-Dunne, 2009) as well as in children with varying levels of anxiety (Muris et al., 2008). Executive attention control dysfunction in children with high levels of social anxiety is thought to be due to prolonged processing of information related to worry and rumination about social situations and the self (Brozovich & Heimberg, 2008; Heimberg et al., 2014). However, research has yet to determine the specific role that executive attention control mechanisms play in childhood social anxiety.

To date, only a few studies have investigated the relation between executive attention control and social anxiety. These studies found support for an association between impaired executive attention control and higher levels of social anxiety, in both adults (e.g., Moriya & Tanno, 2008; Morrison & Heimberg, 2013; Taylor et al., 2016) and youth (Pergamin-Hight et al., 2016). However, all four studies are greatly limited by only using a self-report measure of executive attention control. The participants were asked to self-report about their own executive attention control, and no objective measurement or an external view on their executive attention control was administered. This is unfortunate, especially in studying social anxiety, as socially anxious children fear social and performance situations and evaluate themselves as being less capable than their peers. Indeed, several studies suggest that children with social anxiety might partly underestimate their social and performance skills (see also Miers et al., 2011), and some studies even found no significant relation between social anxiety and performance skills deficits (Baartmans et al., 2019). To the best of our knowledge, only one study in adults, and none in children, has examined executive attention control and social anxiety using an objective measure of executive attention control. Morrison et al. (2016) used an attentional blink paradigm (Martens & Wyble, 2010) in 166 patients with social anxiety disorder and a group of 37 healthy controls. Interestingly, their findings showed that only patients with comorbid depression exhibited impaired executive attention control abilities. Their study, however, was limited by unequal sample sizes, and further investigations are needed. Some studies have examined the role of executive attention control in socially anxious adults under high and low perceptual load, using an experimental measure of executive attention control (Moriya & Tanno, 2010, 2011). However, provided that perceptual load is purported to modulate early pre-processing at involuntary levels of executive attention control (Berggren & Derakshan, 2013), these findings may not be related to the focus of the present paper. The current study examines executive attention control at the cognitive voluntary level of processing and not at the pre-conscious level of awareness. The role that executive attention control impairment may play in social anxiety is thus an area relatively unexplored using objectively measured executive attention control in both adults and children, and it highlights the need for further investigation.

In the current study, we investigated the relation between social anxiety and executive attention control, while focusing on subjective versus objective executive attention. We also tested which aspects of executive attention control would best predict social anxiety: objective executive attention control, subjective perceived attention control, and/or the evaluation of one’s own performance on the objective control task. Executive attention control was measured objectively with the Attention Network Task (ANT; Fan et al., 2002) and subjectively with the Attentional Control Scale (ACS; Derryberry & Reed, 2002) in children aged 8–13. The ANT is a non-stressful computer task of approximately 10 minutes that uses reaction times to assess alerting, orienting, and control of executive attention.

Based on studies investigating executive attention control and social anxiety using self-report methodology (e.g., Pergamin-Hight et al., 2016), we hypothesized to find negative correlations between the levels of social anxiety and performance on the ANT, self-reported executive attention control, and evaluation of the performance on the ANT. In addition, we expected that performance on the Attention Network Task would be positively related to self-reported executive attention control scores. Regarding our secondary aim, given the nature of social anxiety, we would expect that social anxiety was best predicted by the three variables in combination (i.e., objective ANT performance, subjective self-reported attention control, and evaluation of the ANT performance). To summarize, high levels of social anxiety are hypothesized to be related to (1) lower levels of objective attention control, (2) underestimation of performance, and (3) lower levels of self-reported perceived attention control, which should all explain unique variance, and thus distinct aspects, of social anxiety (see also Baartmans et al., 2019, 2020).

Methods

Results

Relations with age and gender

There was a significant correlation between age and ANT performance scores (r = .28, p = .001), older children performed better than younger children. There was no significant effect of gender, F (1,132) = 1.01, p = .316, η² = .01. The mean total score on the ANT-question was 5.3 (SD = 1.0). There was no significant correlation between age and ANT-question scores (r = −.15, p = .094), but there was a relation with gender, F (1,132) = 2.56, p = .011, η² = .05, boys rated their performance on the ANT significantly higher than girls. The mean total score on the ACS was 52.9 (SD = 8.8). There was a significant correlation between age and ACS scores (r = .20, p = .019), older children reported higher executive attention control than younger children, but there was no relation with gender, F (1,132) = 1.21, p = .228, η² = .01. The mean total score on the SASC-R was 41.8 (SD = 15.8). There was neither a significant correlation between age and SASC-R scores (r = .04, p = .622), nor with gender, F (1,132) = .14, p = .887, η² = .01 (see Table 1).

OPEN IN VIEWER

Table 1.

Descriptive statistics for age, error rates, missing values, SASC-R, ACS, ANT, and ANT-question (n = 134).

	Mean	(SD)
Age	10.0	(1.1)
Error (% trials)	4.7	(6.3)
Missing (% trials)	5.9	(6.2)
SASC-R	41.8	(15.8)
ACS	52.9	(8.8)
ANT	−104.6	(67.9)
ANT-question	5.3	(1.0)

Note. SASC-R: Social Anxiety Scale for Children; ACS: Attentional Control Scale; ANT: Attention Network Task, ANT-question: “How do you think you did?”, which is the question that is asked to the participants following the ANT paradigm.

Relations between social anxiety and executive attention control

In order to address the first hypothesis, that social anxiety was negatively related to actual performance executive attention control on the ANT, subjective executive attention control, and evaluation of the ANT performance, we conducted partial correlations, controlled for age and gender. As expected, the SASC-R scores correlated significantly with the ACS scores (r = −.44, p < .001) and the ANT-question scores (r = −.27, p = .002), children with higher self-reported social anxiety reported lower executive attention control and thought that they did worse on the ANT. These two correlations remained significant even after correcting for multiple testing with the Benjamini–Hochberg procedure and the Bonferroni correction. The correlation between ACS scores and ANT-question scores (r = .16, p = .061) did not reach the conventional level of statistical significance. Unexpectedly, the ANT performance scores correlated significantly neither with the SASC-R scores (r = .01, p = .922), the ACS scores (r = .10, p = .234), nor the ANT-question scores (r = −.04, p = .616). This means that how the children actually performed on the ANT was not related to their levels of self-reported social anxiety or self-reported executive attention control, nor was it related to how they perceived their own performance on the ANT (see Table 2).

OPEN IN VIEWER

Table 2.

Hierarchical regression analysis predicting SASC-R scores from age, gender, ACS scores, ANT scores, and ANT-question scores (n = 134).

Criterion variable	Step	R 2	R 2 Change	Predictor	β
SASC-R	1	<0.01
				Age	0.04
				Gender	−0.01
	2	0.21	0.21**
				Age	0.09
				Gender	−0.02
				ACS	−0.42**
				ANT	0.03
				ANT-question	−0.21*

Note. SASC-R: Social Anxiety Scale for Children; ACS: Attentional Control Scale; ANT: Attention Network Task, ANT-question: “How do you think you did?”, which is the question that is asked to the participants following the ANT paradigm.

*p < .05, ** p< .001, standardized β coefficients are reported.

Discussion

According to the Attentional Control Theory, higher levels of anxiety are associated with lower levels of executive attention control (Eysenck & Derakshan, 2011; Eysenck et al., 2007). The goal of the current study was to investigate the role of executive attention control in children with varying levels of social anxiety symptoms. Results with regard to our first hypothesis indicated only partial support for an association between poorer executive attention control and higher social anxiety levels. Self-reported executive attention control (measured with the ACS) was associated with higher levels of social anxiety. However, the experimental objective measure of executive attention control (measured with the ANT) was not related to self-reported social anxiety levels. These findings are similar to findings from the adult (e.g., Moriya & Tanno, 2008; Morrison & Heimberg, 2013; Taylor et al., 2016) and child literature using self-report measures of executive attention control (Pergamin-Hight et al., 2016). In addition, this finding supports the findings by Morrison et al. (2016) that objectively measured executive attention control is not related to social anxiety. One might be tempted to attribute this lack of relations to insufficient reliability of the ANT scores. However, the reliability was actually quite good for a reaction time task, and sufficient for a correlation with age. Taken together, the data of our study in children and the study by Morrison and colleagues (2016) in adults indicate a likely continuity across age in the relation between self-reported, but not experimentally tested, executive attention control and social anxiety symptoms. It may be that patterns of self-reported executive attention control associated with emotional states are established at a relatively early age, and continue into adulthood.

These findings do, however, pose the question whether the executive attention control impairment in social anxiety is an actual deficit or just a perceived deficit. The question, “How do you think you did?”, was related to social anxiety in that lower perception of performance was associated with higher levels of self-reported social anxiety. Moreover, in answer to our second hypothesis, both the self-perceived executive attention control and the subjective evaluation of the ANT predicted unique variance in social anxiety, but not the actual performance on the ANT. This suggests that social anxiety may bias different aspects of the perception of overall performance (both the evaluation of performance and perceived executive attention control). This might indicate that social anxiety is related to a perceived deficit in executive attention control, and not necessarily to a real impairment in executive attention control. Alternatively, this could also be explained by heightened levels of fear of negative evaluation and/or performance anxiety in children with high levels of social anxiety. Clearly more research is needed to test this hypothesis before firm conclusions can be drawn.

Besides the unique relation of social anxiety and the different measures of subjective executive attention control, we also examined the relationship between self-reported executive attention control and experimentally measured executive attention control. The results showed no significant relationship between the two measures. This finding may seem surprising, however, it is similar to previous studies in adults (Reinholdt-Dunne et al., 2013) and children (Muris et al., 2008). Both studies examined self-reported executive attention control and experimentally measured executive attention control in relation to broader levels of psychopathology (i.e., general levels of anxiety, depression, and aggression). These studies highlight the discrepancy between self-report and objectively measured executive attention control in general. Several explanations may be given for this low overlap between self-reported and experimentally measures attention control. First, the fact that the reliability of the ANT scores was lower than the reliability of the questionnaires may have contributed to low correlations of the ANT scores with other variables, but it can hardly explain the total lack of correlations observed here. Second, a more likely explanation states that self-reported attention control measures a different construct than experimentally measured executive control. Self-reports might measure the subjective belief of being able to control attention (see also Muris et al., 2008). Third, it might be that attention control is more complicated, and cannot be measured with a single experimental paradigm. In the future, it may be useful to use tests of executive attention control that address specific components such as the abilities to inhibit, to focusing, or to shift attention. Additionally, future studies could, for example, combine the ANT with other measures such as the Test of Everyday Attention for Children (Teach-Ch; Manly et al., 2001) that was used in the study by Muris et al. (2008). In sum, clearly more research is needed that includes both self-reports and experimental measures. Furthermore, the findings stress the importance of using multiple measures of cognitive control, instead of only relying on either self-reported attention control or a single experimental measure of it.

The current results were observed in children with varying, non-clinical levels of social anxiety. If replicated in children with clinical levels of social anxiety, the findings from the present study, that is, that executive attention control deficits may be perceived rather than real, may be relevant for the development of interventions for social anxiety. Even though clearly more research is needed, our results suggest that treatments may benefit from focusing on perceived deficits, in addition to training actual deficits.

Of course, the present study is not without limitations. First, our sample partly overlapped with samples of previous studies, and it was a selected, but non-clinical sample. Clearly, further research is needed in order to determine if similar findings, for instance, the relation between executive attention control and levels of social anxiety, are also present in children with clinical levels of social anxiety. Some of the literature referred to in the introduction (e.g., Martens & Wyble, 2010; Morrison et al., 2016) is based on clinical samples, which poses a limitation when comparing the findings of these studies to the ones of the current study.

Second, the focus of this paper has been on executive attention control measured with neutral stimuli. Therefore, the findings cannot be compared to results from tasks using emotional materials, such as, for instance, the emotional Stroop paradigm (Williams et al., 1996). This is an important distinction because tasks with emotional materials assess so-called cognitive biases, that is, the preferred processing of some stimuli (usually disorder-related ones) over others (usually neutral ones). In contrast, tasks employing neutral materials like the one used here assess general processing deficits. This difference may be particularly relevant in social anxiety, as it has been related to both cognitive biases and cognitive deficits. Therefore, it would be important for future research and our current theoretical models of social anxiety to investigate this further by including both types of tasks in the same study. This will also have consequences for potential treatments: Depending on whether cognitive biases, cognitive deficits, or self-perceived deficits are most prominent in childhood social anxiety, one or the other approach should be most beneficial.

Third, our findings are limited to social anxiety, and additional research is needed before any conclusions can be drawn about other aspects of anxiety or to any anxiety disorder. Fourth, the ANT task is a very specific task for measuring executive attention control, and other tasks or parent/teacher reports may have higher ecological validity of the executive attention control ability in daily life and the impact of that on daily functioning. Before firm conclusions can be drawn, clearly more studies are needed that use varying experimental measures and observers, such as parents or teachers, besides self-reports and the ANT. Fifth, our study has a narrow focus on the executive attention control of the ANT. Future studies may wish to broaden the scope to also include the orienting and alerting measure of the ANT. Finally, the time of assessment was not standardized and not assessed. Preferably, we would have wanted to test early in the morning for all children. Unfortunately, this was not possible due to time restrictions and practical reasons for the schools we worked with. Variance in attention span could have influenced the reaction times on the ANT.

In conclusion, the present study was the first to investigate objectively measured executive attention control with non-emotional words in relation to social anxiety in children. Findings did not support theoretical notions that there is an actual impairment in executive attention control in social anxiety. Given that our results are similar to empirical findings from the adult literature, it supports the hypothesis that in social anxiety, executive attention control deficits for non-emotional stimuli are perceived rather than actual. This finding is important because innovative treatments of social anxiety disorder in both children and adults are using various ways of training information processing biases as a means of reducing social anxiety symptoms. Knowledge from this study may help researchers better understand the underlying attentional mechanisms present in childhood social anxiety. Even though clearly more studies are needed on the relation between subjectively and objectively measured executive attention control in clinical samples before any firm conclusions can be drawn, our results underscore the importance of studying executive attention control in more detail, using multiple measures of executive attention control when investigating this construct in relation to childhood social anxiety.