The role of attentional shifting in the relation between error monitoring and anxiety in youth

Abstract

The error-related negativity (ERN), a well-established neural marker of anxiety, reflects enhanced attention to internal threat signals. While attention to threat plays a crucial role in the development and maintenance of anxiety, it is unclear how attentional control influences the ERN-anxiety association. To address this, 37 youths (M_age = 10.89 years) completed self-report measures of attentional control and anxiety symptoms. To obtain ERN amplitude, youth completed a flanker task while simultaneous EEG was collected. Attentional control, specifically attentional shifting rather than focusing, moderated the relation between ERN amplitude and anxiety. Youth who displayed smaller neural responses to making an error and higher ability to shift attention experienced lower levels of anxiety, relative to those who exhibited larger neural responses to making an error or lower attention-shifting ability. These findings highlight that response magnitude to internal threat and ability to flexibly shift attention may jointly contribute to anxiety in youth.

1. Introduction

Anxiety disorders affect nearly 20% of children and adolescents (van Schalkwyk & Silverman, 2018). There is growing interest in understanding the roles of attentional processes in the development and maintenance of anxiety. In this study, we examine the interplay between two attention processes, error monitoring and attentional control, in relation to anxiety in children and adolescents.

Anxiety has been shown to negatively affect both attentional control, the ability to voluntarily and strategically focus and shift attention in a goal-directed manner (Derryberry and Reed, 2002), and error monitoring, the detection of and adjustment for inappropriate responses (Crone et al., 2008). Specifically, individuals with anxiety show dysregulated attentional patterns, demonstrating an attentional bias to threat (Cisler and Koster, 2010). This dysregulation seems to inhibit the ability to remain focused on current task demands and shift attention in the presence of salient but task-irrelevant stimuli (i.e., threat; Mogg and Bradley, 2016). Those with anxiety report increased apprehension about the commission of their mistakes, and exhibit enhanced ERN amplitudes in response to errors, suggesting a hypervigilance to internal threat. Therefore, this anxiety-related hypervigilance to the production of errors is likely to be affected by attentional control.

1.1. Error monitoring

Threats tend to be thought of as external events that can cause us harm. However, threats are not limited to the external world, and there is behavioral evidence to suggest that internal threats such as those experienced by the production of erroneous responses can be an extremely aversive variant of response conflict (Botvinick et al., 2001). Making errors leads to subjective distress and physiological changes, such as increased heart rate, dilation of the pupils, as well as skin conductance changes (Proudfit et al., 2013).

A well-established neural marker of error monitoring is the error-related negativity (ERN; Hajcak et al., 2003). The ERN is an event-related potential (ERP) characterized by a fronto-central negative deflection expressed within 100 ms of the commission of an error (Falkenstein et al., 1991; Gehring et al., 1993; Hajcak, 2012; Riesel et al., 2012). The ERN is said to be reflective of a monitoring system which detects for sources of threat, such as conflicts within the environment and making mistakes (Kujawa et al., 2016; Wauthia and Rossignol, 2016), and thought to provide a signal to other brain regions so that top-down control can be adjusted in anticipation of future conflicts. Both youth and adults with anxiety disorders exhibit enhanced response monitoring, expressed as increased (i.e., more negative) ERN amplitudes compared to controls (for a review, see Meyer et al., 2017), suggesting a hypervigilance to internal threat (Weinberg et al., 2010). Additionally, more negative ERN amplitudes precede the onset of anxiety disorders within the pediatric population, such that increased ERN amplitudes at six years of age was predictive of anxiety disorders at nine years of age (Meyer et al., 2015). Previous work has shown that the ERN can be elicited in young children with and without an anxiety disorder (Barker et al., 2015; Meyer et al., 2015; Proudfit et al., 2013; Speed et al., 2017), and has suggested that the network involved in the error monitoring system is relatively stable throughout adolescence and adulthood in typically developing individuals (Buzzell et al., 2017). This can be crucial when considering the ERN as a potential biomarker for anxiety or other psychopathologies as many of these disorders are characterized by either a reduction in or enhanced ERN amplitude (Buzzell et al., 2017; Olvet and Hajcak, 2008). However, little is known about how individual differences in error monitoring and attentional control ability may influence anxiety.

1.2. Attentional control and error monitoring

While several studies have established the role of attentional control on threat in individuals with anxiety, these studies have primarily focused on external rather than internal threat. Anxiety-related hypervigilance to internal threats as measured by the ERN is likely also to be influenced by attentional control. It may be that when top-down cognitive control processes are impaired, the ability to restrain bottom-up attention is reduced thus producing larger responses to these internal threats, leading to heightened levels of anxiety. In contrast, being able to adaptively deploy top-down attentional control may allow youth to strategically shift attention away from internal threats to modulate levels of anxiety.

1.3. Attentional control theory and anxiety

Attentional control theory has been used to explain the role of attentional control processes in the development and maintenance of anxiety disorders (Eysenck et al., 2007). The central hypothesis of this theory is that anxiety impedes attentional control by selectively directing attentional resources to threat-related stimuli (Eysenck et al., 2007; Eysenck and Derakshan, 2011). This theory highlights two attentional systems which may contribute to the dysregulation of attentional processes associated with anxiety: bottom-up or stimulus-driven attention and top-down or goal-driven attention (Coombes et al., 2009). Bottom-up or stimulus-driven attention refers to those instances during which certain, more salient aspects of a stimulus capture attention (Theeuwes, 1991) and are attended to despite intention (Posner et al., 1990). In contrast, top-down or goal-driven processes of attention, including attentional control, are those which are under the control of the individual (Asplund et al., 2010). It is proposed that anxiety disrupts the efficiency of this goal-driven system, allowing increased influence of the stimulus-driven attentional system (Eysenck et al., 2007). It is further proposed that attentional control can be divided into two components: attentional focusing and attentional shifting.

1.4. Attention focusing

Attentional focus can be defined as the ability to maintain attention on task-relevant stimuli, filtering out distracter information (Susa et al., 2012). The attentional control theory suggests that anxiety can be partially attributed to disturbances in the balance of attentional systems, such that bottom-up processes overwhelm top-down processes particularly when there is a perceived presence of threat (Derakshan and Eysenck, 2009; Eysenck et al., 2007; Eysenck and Calvo, 1992). That is, difficulties focusing attention when anxiety is high may be attributed to vigilant monitoring for threatening cues in the environment. Specifically, youth with clinical or self-reported high-trait anxiety exhibit a bias toward threatening stimuli within their environment (Bar-Haim et al., 2007; Dudeney et al., 2015), suggesting that anxiety impedes attentional focusing on tasks with concurrent demands, particularly when threat is present. Previous research has also demonstrated that anxiety can increase attentional focus to salient, threat-related cues indicated by decreased reaction times to probes replacing salient, threatening cues compared to neutral cues (Eldar et al., 2010; Mueller et al., 2009). Findings like this provide further evidence for disrupted top-down attentional focusing by bottom-up attentional processes within anxiety.

1.5. Attentional shifting

Another facet of attentional control, attention shifting, has been suggested to impact anxiety levels (Lonigan and Vasey, 2009). Attention shifting involves the adaptive allocation of attention away from threatening cues in the environment, both internal and external, and the redirection to information in efforts to realign with goal-directed behaviors (Derryberry and Reed, 2002; Perez-Edgar and Fox, 2005). Research has shown that those with higher anxiety may have a harder time shifting their attention away, or disengaging from threat. This is seen in work using the dot probe task, during which, when compared to non-anxious individuals, those with high trait-anxiety exhibit slower reaction times on trials where the probe replaces the neutral cue, suggesting that anxious individuals may be slow to shift attention away from the threat-related cue (Bar-Haim et al., 2005). There is also evidence to suggest that training clinically anxious individuals to disengage, or shift away, from threat reduced both self-reported levels and behavioral indices of anxiety during a speech task (Heeren et al., 2011). Additionally, increased ability to shift attention has been shown to serve as a protective factor against heightened risk for anxiety symptomology (White et al., 2011). Findings such as these suggest attention shifting in the maintenance and treatment of anxiety. To examine the possibility that the ability to adaptively deploy top-down attentional control may allow youth to strategically shift attention away from internal threats to modulate levels of anxiety, the current study aims to examine attentional control as a moderator of the ERN-Anxiety relation. This study is the first to examine whether youth self-reported attentional control, including shifting and focusing ability, moderates the relation between ERN amplitudes and anxiety symptoms. We predicted that attentional control, and more specifically shifting ability, would moderate the relation between internal threat signals and anxiety level, such that lower error monitoring and higher levels of shifting would be associated with lower levels of anxiety.

2. Method

2.1. Participants

Twenty youths with anxiety disorders (ANX; 9 female, M = 11.4 years, 8–16 years) were recruited from a clinic specializing in anxiety treatment. Youth met criteria for a primary Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR; American Psychiatric Association, 2000) anxiety disorder: Generalized Anxiety Disorder (n = 9), Social Phobia/Anxiety Disorder (n = 5), Separation Anxiety Disorder (n = 3), Specific Phobia (n = 2), and Panic Disorder (n = 1). Four youth met criteria for a comorbid (non-primary) diagnosis of Attention-Deficit Hyperactivity-Inattentive (ADHD-I). Four youth were on a stable dose of medication for attention deficits (n = 1) or anxiety (n = 3). To increase variability in anxiety symptomology, 22 healthy youth (CON) who did not meet criteria for and had never been diagnosed with or treated for neuropsychological, emotional, or behavioral disorders were included. Five CON were excluded due to excessive motor artifact (n = 2), equipment malfunction (n = 1), incomplete questionnaire data (n = 1), and identification as a statistical outlier (n = 1) resulting in 17 CON (5 female, M = 10.65 years, 8–14 years).

2.2. Measures

2.2.1. Anxiety symptomology

The Child (C) and Parent (P) versions of the Screen for Child Anxiety Related Disorders (SCARED;Birmaher et al., 1997) were administered to measure anxiety levels using 41-items scored on a three-point Likert scale. Example items include, “I worry about being as good as other kids” and “My child worries that something bad might happen to his/her parents”. Higher scores are indicative of higher levels of anxiety. The SCARED-C and SCARED-P yielded high internal consistency (α = 0.96 and α = 0.91, respectively). A total Anxiety Score for each participant was computed by averaging SCARED-P and SCARED-C total scores (Bechor et al., 2019; Roy et al., 2014).

2.2.2. Attentional control

The Attentional Control Scale for Children (ACS-C; Muris et al., 2004) was administered as a measure of attentional control. This 20-item self-report measure contains two subscales which assess attentional focusing and attentional shifting on a four-point Likert scale (Melendez et al., 2017; Muris et al., 2004; van Son et al., 2021). The focus subscale measures one’s ability to maintain attention on a stimulus, while the shift subscale assesses ability to shift attention from one stimulus to another (Melendez et al., 2017). Questions such as, “It’s very hard for me to concentrate on a difficult lesson, if there is a lot of noise in the class” assess focus, while questions such as, “When I am daydreaming or having distracting thoughts, it is easy for me to switch back to the work I have to do” assess shifting. Higher scores are indicative of better attentional control. Cronbach’s alpha for the total scale was 0.85 (focusing and shifting subscales: α = 0.71 and α = 0.74, respectively).

2.2.3. Error monitoring

Error monitoring was assessed using an arrow version of the flanker task (Eriksen and Eriksen, 1974). Stimuli were presented using E-prime software (Psychology Software Tools, Pittsburgh, PA) on a 14-inch laptop monitor. Youth were instructed to attend to the central arrow and indicate its direction via mouse button press using their dominant hand while ignoring the flanking arrows during equal numbers of congruent (>>>>> or <<<<<) and incongruent (<<><< or >><>>) trials (12 blocks, 32 trials/block, 384 trials total). Each 200 ms trial was preceded by a 300–600 ms fixation cross and followed by a 1700 ms response window. Inter-trial intervals adapted based upon participant speed. Responses during the flanker task were used to calculate mean RTs on both correct and error trials as well as overall accuracy. In addition, post-error slowing was assessed by computing the difference between RTs on correct trials following incorrect responses and RTs on correct trials following correct responses. Scores greater than 0 indicate the occurrence of post-error slowing. Post-error slowing is a well-established index of behavioral adjustment (Ladouceur et al., 2007; Rabbitt, 1966) which is observed as a response slowing after the commission of an error. These behavioral adjustments reflect top-down cognitive control processes which aim to increase the chance of responding correctly in the future (Olvet and Hajcak, 2008). See Supplemental Results for results regarding post-error slowing.

2.3. EEG procedures

Participants were fitted with a 64-channel electrode Waveguard cap (ANT Neuro Corporation, Netherlands), ensuring that Cz was as centrally located as possible by measuring the cross-section between the two preauricular points, and the nasion and inion, and continuous EEG was recorded during the flanker task. The raw signal was amplified by 25,000 using AsaLab high-input impedance amplifier (Advanced Neuro Technology, Enschede, Netherlands). Impedances were <50 kΩ and raw data was sampled at 1024 Hz using a high-pass filter of 0.3 Hz, and referenced to CPz. Using EEGLab (Delorme and Scott Makeig, 2004), data were resampled offline at 512 Hz, and filtered with a low-pass filter of 30 Hz. EEG was segmented into response-locked epochs −400 ms to 800 ms using ERPLab (Lopez-Calderon and Luck, 2014). The −300 to −100 ms time window was used for baseline correction. Epochs with voltage threshold artifact of ±75 μV or observed to be noisy via visual inspection were rejected and excluded from further analysis, and then re-referenced to average of all electrodes. The ERN and CRN were quantified using mean amplitude measures during response-locked error and correct trials, respectively, during the flanker task. In line with methods from previous research in a pediatric population (Hajcak et al., 2007), in the current study we calculated mean amplitudes for both the ERN and CRN for the time interval between −50 and 75 ms after responses at Fz. Fig. 1 and S1 display the scalp topographies representing error- and correct-related activity −50 to 75 ms post-response on error and correct trials, respectively, during the flanker task in youth. A difference score (error minus correct; ΔERN) was also calculated for the analyses.

Fig. 1.

Response-locked grand-averaged waveforms displaying the error-related negativity (ERN) during error trials at Fz (solid black line) as well as correct-related negativity during correct trials at Fz (dashed black line).

2.4. Statistical analysis plan

To test our hypotheses, hierarchical linear regression analyses were conducted. Age, gender, presence of ADHD, and use of medication at the time of collection were entered as covariates in all analyses. The number of error trials were also included as a covariate as the number of error trials can significantly influence ERN amplitudes (Fischer et al., 2017; Lahat et al., 2014). Analyses were conducted on the total Anxiety Score, and the interaction terms were computed using standardized scores (z scores) on the respective attention measures, as well as ERP mean amplitudes. On average, youth committed ~45 errors (range: 9–114), thus reaching the minimum of 6 error trials needed to elicit a reliable ERN response (Meyer et al., 2013; Olvet and Hajcak, 2009). To further explore significant interaction effects, graphical displays were created based on the convention for plotting interactions (see www.jeremydawson.co.uk/slopes.htm). As the ERN is a negative going waveform, in figures plotting significant interactions “Large ERN” refers to a more negative ERN amplitude response, while “Small ERN” refers to a more positive ERN amplitude response. Post-hoc power analysis indicated that the power to detect obtained effects at the 0.05 level was 0.98 for the overall regression in the prediction of anxiety. Power to detect obtained effects at the 0.05 level when examining for effects of attentional shifting and attentional focusing on the ERN/anxiety relation was 0.99 and 0.97, respectively (G*Power 3.1.9.2; Faul, Erdfelder, Buchner, & Lang, 2009). As the focus of the current study was to examine individual rather than group differences, the data presented are continuous. Comparisons between youth with anxiety and control youth can be found in Supplementary Results.

3. Results

Table 1 displays descriptive statistics and correlations between anxiety, attentional control scores, and ERN amplitude controlling for age, gender, medication usage, and presence of ADHD across the whole sample, and table S1 displays the correlations, and means and standard deviations at the group level. A higher Anxiety Score was significantly related to lower levels of attentional control (p = .03). Furthermore, Anxiety Score was significantly and negatively correlated with focus (p = .02), but not shifting (p = .08). No significant correlations were found between ERN, CRN, and ΔERN amplitudes and anxiety or attentional control. Fig. 1 displays the response-locked grand-averaged waveforms displaying the error-related negativity (ERN) during error trials at Fz (solid black line) as well as correct-related negativity during correct trials at Fz (dashed black line).

Table 1.

Partial correlations between anxiety, attentional control, and ERN amplitudes controlling for age, gender, presence of ADHD, and medication use, as well as means and standard deviations for attentional control, and ERN amplitudes.

	Anxiety Score	ACS-C Total	ACS-C Focus	ACS-C Shift	ERN (μV)	CRN (μV)	ΔERN (μV)
ACS-C Total	−.37*
ACS-C Focus	−.40*	.94**
ACS-C Shift	−.31†	.96**	.79**
ERN (μV)	−.14	−0.22	−.21	−.20
CRN (μV)	.25	.13	.03	.19	.36*
ΔERN (μV)	−.02	−.25	−0.23	−.23	.98**	.36*
Mean (SD)	19.50 (11.80)	54.54 (9.87)	24.86 (4.60)	29.68 (5.75)	−2.00 (5.52)	0.86 (4.87)	−1.60 (5.42)

ACS-C = Attentional Control Scale for Children; ERN = error-related negativity; CRN = correct-related negativity; μV = microvolts; SD = standard deviation.

^*p < .05.

^**p < .001.

^†p < .10.

Anxiety Score was regressed on continuous measures of ERN amplitude and total ACS-C score and the two-way interaction between ACS-C scores and ERN amplitude. A significant main effect was seen for ACS-C score on Anxiety Score, β = −0.357, b = −4.21, p = .03, 95% CI [−8.09, −0.34], R² = 0.36, ΔR² = 0.11, however, no main effect was seen for ERN amplitudes on Anxiety Score, β = −0.23, b = −2.65, p = .18, 95% CI [−6.59, 1.28], R² = 0.40, ΔR² = 0.04. A significant interaction effect between ERN amplitude and ACS-C on Anxiety Score, β = −0.30, b = −4.29, p = .05, 95% CI [−8.60, 0.03], R² = 0.47, ΔR² = 0.078, was found (Fig. 3). Specifically, the standardized slope for the effect of ERN was significant (p = .02) when ACS-C scores were one standard deviation (SD) above the mean (β = −1.22) but not one SD below the mean (β = 0.33, p = .53). Youth with higher attentional control showed a stronger association between higher anxiety and a larger (more negative) ERN, whereas youth with lower attentional control had relatively high anxiety regardless of the size of the ERN. Youth with high attentional control and small (less negative) ERN amplitudes had the lowest anxiety levels.

Fig. 3.

Individual differences in self-reported attentional control moderates the relation between ERN mean amplitude and reported levels of anxiety. Note. Large ERN = more negative ERN response; Small ERN = more positive ERN response.

To gain insight into components of attentional control that may explain this interaction effect, the ACS-C subscales were examined separately. Because of the strong association between shifting and focusing abilities, in order to assess how each independently affects the relation between the ERN and anxiety, shifting was included as a covariate when the moderating effect of focusing was examined, and focusing was included as a covariate when the moderating effect of shifting was examined. While there were no significant main effects of either shifting ability, β = 0.07, b = 0.82, p > .80, 95% CI [−5.54, 7.18], R² = 0.38, ΔR² = 0.00 or ERN, β = −0.23, b = −2.72, p = .17, 95% CI [−6.63, 1.20], R² = 0.42, ΔR² = 0.041 on Anxiety Score, the interaction between ERN amplitudes and shifting ability significantly predicted anxiety levels independent of focus ability, β = −0.37, b = −4.69, p = .02, 95% CI [−8.57, −0.81], R² = 0.53, ΔR² = 0.12 (Fig. 4a). The standardized slope for the effect of ERN was significant (p = .01) when ACS-C shifting scores were one SD above the mean (β = −1.33) but not one SD below the mean (β = 0.37, p = .44). Youth with higher attentional shifting showed a stronger association between higher anxiety and a larger, more negative ERN, whereas youth with lower attentional control had relatively high anxiety regardless of the size of the ERN. Youth with high shifting abilities and a smaller, less negative ERN response had lower anxiety levels compared to youth with low shifting abilities or larger, more negative ERN.

Fig. 4.

Individual differences in self-reported (a) shifting abilities moderates the relation between ERN amplitude and reported levels of anxiety but not focusing ability (b). Note. Large ERN = more negative ERN response; Small ERN = more positive ERN response.

When examining focusing ability, no main effects were observed for focusing, β = −0.48, b = −5.59, p = .07, 95% CI [−11.75, 0.57], R² = 0.42, ΔR² = 0.07, or ERN response, β = −0.20, b = −2.40, p = .24, 95% CI [−6.45, 1.66], R² = 0.35, ΔR² = 0.03. Additionally, no interaction effect was seen for focusing abilities, β = −0.23, b = −3.56, p = .14, 95% CI [−8.32, 1.21], R² = 0.47, ΔR² = 0.05 (Fig. 4b). No significant main effects or interactions were found using the ΔERN, ps > 0.05.

4. Discussion

The current study is the first to examine whether attentional control moderates the relation between error monitoring and anxiety in youth. Attentional control interacted with ERN amplitudes to predict anxiety levels, such that youth who displayed attenuated (less negative) ERN amplitudes and high levels of attentional control exhibited lower levels of anxiety compared to youth with low levels of attentional control or enhanced (more negative) ERN amplitudes. When shifting and focusing abilities were examined separately, we found that shifting abilities, rather than focus abilities, moderated the relation between ERN amplitudes and anxiety levels. These results fit within cognitive models of anxiety that suggest that the inability to flexibly shift attention away from threat-related cues contributes to the maintenance of anxiety (Barry et al., 2015; Legerstee et al., 2010), highlighting the interplay between cognitive control processes and error monitoring. Our findings extend these cognitive models to a measure of internal threat, the ERN, by demonstrating that attention shifting abilities moderate the ERN-anxiety relation.

Individuals with clinical or self-reported high trait-anxiety exhibit a hypervigilance to threat-related stimuli compared to non-anxious individuals (Bar-Haim et al., 2007; Bechor et al., 2019; Fox et al., 2002; Mathews and MacLeod, 2002). This hypervigilance toward threat is proposed to be the result of dysregulated patterns of attentional control, such that bottom-up threat processes overwhelm top-down, goal--directed attentional processes (Derakshan and Eysenck, 2009; Derryberry and Reed, 2002). Previous research has demonstrated a link between attentional bias to threat and attentional control, specifically attentional shifting not focusing, such that youth at risk of anxiety disorder with impaired shifting ability displayed higher threat bias scores during a dot-probe task (Susa et al., 2012). Therefore, better control of one’s attention, such as being able to adaptively deploy top-down attentional control and shifting attention, may serve to influence the severity of anxiety symptomology (Melendez et al., 2017). A separate line of research documents that anxiety is associated with an enhanced response-monitoring, specifically, a hypervigilance for error-monitoring (Buzzell et al., 2017; Torpey et al., 2013; Wauthia and Rossignol, 2016; Weinberg et al., 2010). The current study integrated these two lines of research to gain insight into how top-down, attentional control modulates the influence of the bottom-up, enhanced error monitoring typically observed in individuals with anxiety. The current findings indicate that youth who display both a low response to internal threat (i.e., making an error) and higher ability to flexibly shift attention had the lowest levels of anxiety. In contrast, youth who have difficulty shifting attention had relatively high levels of anxiety, even when the error response is small. Further, youth with high shifting ability and enhanced, more negative ERN response also had relatively high levels of anxiety suggesting that higher levels of attentional shifting are not sufficient to modulate anxiety when hypervigilance to error (i.e., threat detection) is high.

Taken together, results are in line with the attentional control theory which postulates that anxiety hinders the ability to shift attention away from distracting, task-irrelevant stimuli (Eysenck et al., 2007). Further, they highlight the interplay between attentional control and error monitoring in youth anxiety. Anxiety levels are low when both attentional shifting is high and error monitoring (ERN) is small. It appears that a specific aspect of attentional control, shifting, is key to moderating the association between error monitoring and anxiety in youth. This key role of attentional shifting in the presence of an internal threat (i.e., an error) is consistent with prior research indicating that anxious individuals have difficulty disengaging attention from external threats (i. e., threatening faces or words) using the dot-probe task (Fox et al., 2002; MacLeod et al., 1986; Mueller et al., 2009). Overall, these results extend application of attentional control theory to youth anxiety by documenting an interplay between top-down attentional control and bottom-up, internal threat monitoring measured using the ERN.

Consistent with previous research, attentional focus was associated with youth self-reported anxiety symptoms (Melendez et al., 2017), however, the ability to flexibly shift attention seems to be important for the regulation of anxiety symptomology in the presence of threat. Evidence from the current study provides an impetus for training flexible deployment of attention to treat anxiety disorders (Bar-Haim et al., 2007; Fox et al., 2002; Hakamata et al., 2010). This type of attention training has been shown to reduce ERN amplitudes in non-referred adults (Nelson et al., 2015, 2017). Findings from the current study suggest that characterizing individual differences in attentional shifting may be useful when tailoring attention training treatments for youth with anxiety, specifically when these individuals also show low responses to errors.

Previous work on the ERN has highlighted its potential to function as a neural biomarker for anxiety, such that greater (more negative) amplitudes are associated with greater risk (Greg Hajcak et al., 2003; Ladouceur et al., 2006; Meyer, 2017). In contrast, it has been suggested that reduced ERN (more positive) amplitudes may function as a neural biomarker for lower risk of anxiety within high-risk populations (Lahat et al., 2014; McDermott et al., 2009), which may be further enhanced if one also has flexible attention shifting. The current study did not find a direct correlation between ERN amplitude and anxiety level; however, although not significant, the correlation between ERN amplitudes and anxiety levels was in the predicted positive direction.

The results for the current study should be considered in light of several limitations. The youth in the current study varied widely in age, ranging from 8 to 16 years of age, with a mean of 10 years old. Because prior research suggests that ERN amplitudes may differ as a function of youth age (Meyer et al., 2012), we included age as a covariate in statistical analyses. While power analysis indicated high power, the sample size of the current study was relatively small. Further, the current study was unable to determine if associations between measured variables differed across the primary anxiety diagnoses represented in our sample. Despite these limitations, the current study shows that attentional control, specifically shifting ability, moderates the relation between error monitoring and anxiety. Understanding the role that shifting ability may play in anxiety may be useful in clinical settings when it comes to assigning treatment for youth with anxiety-related disorders. Our results suggest that those youth who engage in less error-monitoring may benefit from treatment focused on shifting and redirecting their attention when faced with “threat”.

Fig. 2.

Scalp topographies representing error- and correct-related activity −50 to 75 ms post-response on error and correct trials, respectively, during the flanker task in youth.