Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Child Psychol Psychiatry. Author manuscript; available in PMC 2013 Jun 1.
Published in final edited form as:
PMCID: PMC3354023

Isolating neural components of threat bias in pediatric anxiety



Attention biases towards threat are often detected in individuals with anxiety disorders. Threat biases can be measured experimentally through dot-probe paradigms, in which individuals detect a probe following a stimulus pair including a threat. On these tasks, individuals with anxiety tend to detect probes that occur in a location previously occupied by a threat (i.e., congruent) faster than when opposite threats (i.e., incongruent). In pediatric anxiety disorders, dot-probe paradigms detect abnormal attention biases towards threat and abnormal ventrolateral prefrontal cortex (vlPFC) function. However, it remains unclear if this aberrant vlPFC activation occurs while subjects process threats (e.g., angry faces) or, alternatively, while they process and respond to probes. This magnetoencephalography (MEG) study was designed to answer this question.


Adolescents with either generalized anxiety disorder (GAD, n=17) or no psychiatric diagnosis (n=25) performed a dot-probe task involving angry and neutral faces while MEG data were collected. Synthetic Aperture Magnetometry (SAM) beamformer technique was used to determine whether there were group differences in power ratios while subjects processed threats (i.e., angry vs. neutral faces) or when subjects responded to incongruent vs. congruent probes.


Group differences in vlPFC activation during the response period emerged with a 1-30 Hz frequency band. No group differences in vlPFC activation were detected in response to angry-face cues.


In the dot-probe task, anxiety-related perturbations in vlPFC activation reflect abnormal attention control when responding to behaviorally-relevant probes, but not to angry faces. Given that motor responses to these probes are used to calculate threat bias, this study provides insight into the pathophysiology reflected in this commonly-used marker of anxiety. In addition, this finding may inform the development of novel anxiety-disorder treatments targeting the vlPFC to enhance attention control to task-relevant demands.

Keywords: magnetoencephalography, ventrolateral prefrontal cortex, attention


Cognitive neuroscience research implicate threat-related attention biases in the etiology and maintenance of anxiety (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van, 2007). However, many gaps remain in our knowledge about the pathophysiology of such biases. Indeed, abnormal biases towards threat may result from dysfunction in many processes, including directing attention toward or away from threat and/or directing attention to task demands (Fox, Russo, Bowles, & Dutton, 2001). Attention-bias-modification treatment (ABMT), a treatment that targets threat biases, appears to both reduce anxiety and increase attention control, perhaps via recruitment of frontal cortical regions (Browning, Holmes, Murphy, Goodwin, & Harmer, 2010). As the neural mechanisms underlying abnormal attention biases are elucidated with increasing detail, novel behavioral treatments such as ABMT can be tailored towards these underlying perturbations more effectively, potentially improving therapeutic response.

Dot-probe paradigms are used commonly to measure attention biases by presenting stimulus pairs followed by a probe that requires a response (Figure 1). Threat biases are measured using the reaction time difference between incongruent trials, where the probe replaces a neutral stimulus, and congruent trials, where the probe replaces a threatening stimulus. During dot-probe paradigms, adolescents with generalized anxiety disorder (GAD) exhibited greater ventrolateral prefrontal cortex (vlPFC) activation than healthy adolescents (Monk et al., 2006; Pine & Monk, 2008). Moreover, anxiety levels and vlPFC activation detected with functional magnetic resonance imaging (fMRI) exhibit negative correlations in anxious adolescents (Monk, et al., 2006) and positive correlations in healthy adolescents (Telzer et al., 2008). Together, these findings suggest that vlPFC, through its effects on attention control, facilitates emotion regulation when confronting threats (Corbetta, Patel, & Shulman, 2008; Monk, et al., 2006; Monk et al., 2008).

Figure 1
Dot-probe paradigm and time period definitions.

However, while the vlPFC may play a central role in mediating threat bias and anxiety, the precise role of aberrant vlPFC activation in pediatric anxiety remains unknown. For example, on dot-probe tasks, aberrant vlPFC activation could be associated with difficulties directing attention towards/away from threats (i.e., angry faces vs. neutral faces) or, alternatively, towards the probes that follow these threats (i.e., incongruent vs. congruent probes) (Bar-Haim, et al., 2007). Detecting between-group differences in only one of these contrasts would isolate the aberrant activation to a particular time period. Since motor responses to incongruent vs. congruent probes are used to calculate threat bias, aberrant vlPFC activation in response to probes may demonstrate a central role for this brain region in mediating threat bias in patients with anxiety disorders.

Magnetoencephalography (MEG) can be used to differentiate vlPFC dysfunction to the face presentation vs. to the probes that appear during the response period. Unlike fMRI, which indexes blood flow, MEG directly indexes the brain’s electrical activity, with sensitivity to changes occurring over a few milliseconds (Hari, Parkkonen, & Nangini, 2010). This technique is also free of scalp distortions found in event-related potentials (ERPs). Available neuroimaging studies on dot-probe performance in anxious individuals rely on ERP or fMRI (Eldar & Bar-Haim, 2010; Monk, et al., 2006). ERP studies using dot-probe paradigms have identified frontal recruitment following probe onset (Eldar & Bar-Haim, 2010); however, with limited spatial resolution, it is unclear whether the vlPFC is specifically recruited. fMRI studies depend on a slow hemodynamic-response function, limiting temporal resolution; therefore, aberrant activation found in pediatric anxiety could reflect perturbations to many aspects of a dot-probe trial (Monk, et al., 2006). However, MEG uniquely combines excellent temporal and spatial resolutions. Thus, it is ideally suited to test whether aberrant vlPFC activation occurs when processing threats or responding to the probes used to measure attention biases. Of note, the relationship between fMRI and MEG signals is complex (Winterer et al., 2007), with some data suggesting that fMRI activation correlates with event-related desynchronization (i.e., negative oscillatory power) at frequencies below 25 Hz (Singh, Barnes, Hillebrand, Forde, & Williams, 2002).

Few MEG studies have investigated pediatric populations (e.g., Rich et al., 2011), and to our knowledge, none examine between-group differences in pediatric anxiety disorders or investigate attention biases using the dot-probe task in any age group. The current MEG study compares vlPFC oscillatory power ratios in GAD and healthy adolescents performing a dotprobe task in order to test whether the aberrant vlPFC activation manifests specifically during the response period capturing attention bias following threat cues.



This study recruited physically-healthy, medication-free, 8-to-18-year olds with IQ>70 (Wechsler, 1999). Written informed assent/consent was obtained from all participants and parents. All procedures were conducted in accordance with the National Institute of Mental Health (NIMH) Institutional Review Board.

Psychiatric history was assessed by trained clinicians via the Kiddie-Schizophrenia-and-Affective-Disorders-Schedule (K-SADS) (Kaufman et al., 1997). Patients met full criteria for DSM-IV GAD (n=15) or for social phobia (SoPh) (n=2) accompanied by prominent GAD symptoms not sufficiently distinct from social worries to qualify unequivocally for GAD. Children with secondary anxiety diagnoses or major depressive disorder (MDD) were included because these disorders often co-occur, and prior fMRI studies of the dot probe task include them. Healthy comparison (HC) individuals were free of any current or past Axis I psychiatric disorder. Group characteristics are outlined in Table 1.

Table 1
Group characteristics and behavioral results (mean and standard deviation) of the study patients and health comparison groups

The final sample included 17 GAD and 25 HC adolescents, matched for age, gender, and IQ. Co-morbid diagnoses included: SoPh secondary to GAD=4, separation anxiety disorder=2, panic disorder=1, specific phobia=3, and MDD=3. This final sample was obtained after excluding individuals for technical reasons (GAD=2, HC=6), excessive movement (GAD=10, HC=8) or poor behavioral performance (<65% accuracy, GAD=2, HC=2). The excluded groups were younger [12.1±2.7, t(70)=2.98, p<0.004] than the individuals included in the analysis [14.0±2.6], but did not differ with respect to sex, IQ, and trait anxiety [all p>0.1].


The dot-probe task

In dot-probe trials (Figure 1), participants viewed a fixation cross. Then, two faces appeared at the sides of the fixation cross. After the face pair disappeared, a dot-probe appeared in one of the previously-occupied face locations. Using left or right button press, participants identified the probe orientation, (:) or (..). The response mapping of the probe orientation was counterbalanced across participants.


Face pairs consisted of two black-and-white images of the same individual presented side-by-side. The face identities included six males and six females from the NIMstim face set (Tottenham, 2009). Emotional-neutral trials were classified based on the relation between the locations of the emotional face and the probe. In congruent trials, the probe was presented in the location previously occupied by the emotional face. In incongruent trials, the probe was presented in the location opposite the emotional face. The location of the emotional face was counterbalanced across all trials. In neutral trials, two identical neutral facial expressions appeared and the probe location was insignificant.

Angry and happy blocks alternated twice with order counterbalanced across participants. In each block, 24 congruent and 24 incongruent trials according to the emotion of the block (either angry-neutral or happy-neutral trials) and 24 neutral trials were randomly presented. Each 2500 ms trial consisted of a 500 ms fixation, a 500 ms face pair, a 200 ms probe, and a 1300 ms inter-stimulus interval. Based on prior work implicating vlPFC in threat biases (Monk, et al., 2006), analyses only focused on data collected during angry blocks.

Apparatus and Acquisition

A whole-head 275-sensor MEG system (CTF Systems, Inc., kurage.nimh.nih.gov/meglab/Main/MegFacility) was located in a magnetically shielded room. Each sensor was configured as a first-order axial gradiometer, with two 18-mm radius coils separated by 50 mm, spaced 22 mm apart. MEG data were sampled at 600 Hz (bandwidth: 0-150 Hz). Three localization coils were attached to the nasion, left pre-auricular and right pre-auricular sites to measure head position. Using E-Prime 1.1, task stimuli were displayed on a screen approximately 85 mm from the seated subject.

On a separate visit, a T1-weighted magnetic resonance image using a high resolution 3D MPRAGE MRI sequence (TI/TE/flip angle=725 ms/min/6°, 1.2mm in-plane resolution, and 1.2 mm slice thickness) was collected. Vitamin-E capsules served as fiducial points for co-registration. Using this anatomical image, MEG data were co-registered and normalized into Talaraich space in AFNI software (Cox, 1996). Two patients did not complete a structural MRI; therefore, a similarly aged and gendered participant’s brain was used for co-registration. A multi-sphere head model, based on brain shape was derived for source localization procedures described below.

Data Analysis


Trials with incorrect responses and trials with reaction times less than 150 ms or greater than 2.5 standard deviations outside each individual’s mean reaction time for each event type were excluded from both behavioral and MEG analyses.

Attention bias scores were calculated by subtracting the mean reaction times on angry congruent trials from angry incongruent trials. Positive bias scores reflect a bias towards threat and negative bias scores reflect a bias away from threat.

Group differences in accuracy rates, reaction times, and bias scores were examined and attention biases were tested within each group separately using t-tests with an alpha=0.05.

Synthetic Aperture Magnetometry (SAM) beamformer MEG analysis

Raw MEG data were filtered using 3rd gradient reference coils with fixed weights. DC offset was removed, and data were filtered with high-pass (0.50 Hz) and 60 Hz notch filters.

In MEG analyses, frequency bands and time periods must be selected. Traditionally, for MEG studies which represent the first attempt to address a particular research question, a broad frequency band has been used (Cornwell et al., 2008; Rich, et al., 2011); therefore, in the current study, a 1-30 Hz frequency band was selected for our primary analyses. This approach facilitates comparisons with previous ERP studies using this frequency band (Eldar, Yankelevitch, Lamy, & Bar-Haim, 2010). Moreover, this approach effectively balances Type I and Type II errors that can accrue when conducting multiple analyses in many narrow frequency bands. Nevertheless, secondary analyses examined theta (4-8 Hz), alpha (8-14 Hz), beta (14-30 Hz), and gamma (30-90 Hz) bands. Time periods were defined to capture the periods of dot-probe trials and balance the number of times the non-phased locked signal was sampled for each frequency band. The face-presentation period was defined as the 500 ms following face onset and the response period corresponds to the 1000 ms following probe onset, including the dot-probe and subsequent motor response. For 30 Hz signals, the 500 ms time period was sampled 15 times. For signals lower in frequency, fewer samples were obtained.

A three-way group (GAD, HC)-by-period-(face-presentation, response)-by-condition interaction examined differences between anxious and healthy groups in oscillatory power activity (i.e., non-phase locked activity). The condition term in this interaction varies based on period because congruent and incongruent trials are not defined until the probe appears. Specifically, the face-presentation period included angry (i.e., angry congruent and angry incongruent trials) and neutral trials to capture emotion processing. The response period included the two trial types used to calculate behavioral threat bias, angry-incongruent and angry-congruent (see Figure 1). This comparison captures differential attention shifts towards or away from a spatial location previously occupied by an angry-threat cue.

Separate group-by-condition interactions within each period, face-presentation and response, were used to decompose the 3-way interaction. To fully characterize the between-group difference, subsequent analyses examined the face presentation period, relative to baseline fixation (i.e., the 500 ms preceding the face). In addition, angry-congruent and angry-incongruent trials were examined relative to neutral trials and baseline (i.e., 1000 ms immediately preceding the face onset).

The synthetic aperture magnetometry (SAM) beamformer technique was used to calculate the electromagnetic source power distribution for individual voxels (Vrba & Robinson, 2001). SAM produces a 3D representation of brain activity by using the recorded magnetic field across the sensor array to compute a set of beamformer weights to estimate activation at individual voxels. At each voxel, an optimal spatial filter is determined through a minimum variance procedure that reduces extraneous power sources but preserves signal from that particular voxel; therefore, it minimizes potential artifacts (e.g., eye blinks or muscle contractions). The orientation of the source dipole is estimated by a vector-based approach. This SAM analysis requires no a priori specification of the number of active sources and provides source power images in 3D space within specific frequency bands.

For each individual participant, SAM analyses were performed on the MEG data using 7 mm voxel spacing. The SAM procedure calculates pseudo F-values, ratio of power differences (i.e., activation). Positive power ratios indicate greater power in the active state (e.g., angry incongruent) than the comparison state (e.g., angry congruent). (Vrba & Robinson, 2001). After estimating pseudo F-ratios at each voxel, SAM volumes were normalized to Z-scores. The normalized data were transformed into common Talairach coordinate space.

Group analyses of the whole-brain normalized pseudo F-ratios were conducted using AFNI software. The vlPFC was chosen as the a priori region for our main hypothesis, as tested in a three-way group-by-period-by-condition interaction. For completeness, between-group differences in other cortical regions are also reported. To correct for multiple comparisons in fMRI studies, a whole-brain-derived false discovery rate (FDR) threshold of 0.05 is the current standard and can be approximated using a p<0.005 intensity threshold and 20 3.5×3.5×5 mm voxel threshold (Lieberman & Cunningham, 2009). Given the larger voxel size (7mm)3 in this MEG study, we report regions that survived a p<0.005 uncorrected threshold with a minimum of 4 voxels to approximate 0.05 FDR.

Correlational analysis

Within each group, Pearson correlations examined the association between anxiety, age and measures of threat bias. To examine the association between anxiety measures and MEG data, a functionally-defined region-of-interest (ROI) was created using an 8-mm radius sphere centered at the right vlPFC peak (35,34,−8) identified by the group-by-condition interaction during the response period. Significant correlations between the extracted pseudo-F values from this ROI and anxiety measures were tested. In addition, group differences in correlation were examined using a Fisher’s r-to-Z-transformation. A significance level of alpha=0.05 was used.


Behavioral Results

Table 1 summarizes the behavioral data. There were no between-group differences in accuracy, bias, or reaction-time indices [all p>0.1]. No significant attention biases towards or away from angry faces were detected in either group [all p>0.5].

Across both groups, STAI trait anxiety scores were negatively correlated with threat bias [R=−0.35, p<0.03] (Figure 2A). In the HC group, individuals with higher trait anxiety had greater negative threat bias scores (i.e., greater avoidance) [R=−0.49, p<0.02]. A non-significant negative correlation manifested in the patient group [R=−0.30, p>0.3], with no significant between-group difference [p>0.5]. No correlation between age and threat bias was detected across both groups or within each group separately [all p>0.7].

Figure 2
Correlations between trait anxiety measures, threat bias and vlPFC power.

SAM beamformer analysis

The primary analysis compared the un-averaged MEG activity (i.e., oscillatory power ratios) in the 1-30 Hz frequency band between groups, focusing on the three-way group-by-period-by-condition interaction. A significant three-way interaction was detected in the right vlPFC BA47, [(35,41,−8), F=15.00, k=15]. To decompose this interaction, group differences between conditions were examined in the face-presentation and response period separately. No group differences were found between conditions during the face-presentation period. However, as shown in Figure 3, during the response period between angry-incongruent and angry-congruent trials, a group-by-condition interaction was detected in the right vlPFC BA47, [(28 20,−8), t=3.17, k=10, (35,34,−8), t=3.13, k=5]. Compared to the HC group, the patient group exhibited a greater negative power ratio when contrasting the response periods assessing attention bias (i.e., angry incongruent vs. angry congruent). Specifically, in patients, a significant negative power ratio manifested, suggesting greater power in the angry-congruent, relative to the angry-incongruent condition [(42,41,−8), t=3.77, k=6]. In contrast, the HC group showed similar activity in the two conditions (Table 2). This pattern is also illustrated using extracted data from the vlPFC ROI (Figure 3).

Figure 3
Between-group difference in oscillatory power during response period of threat bias
Table 2
Differences in 1-30 Hz oscillatory power in ventrolateral prefrontal cortex (vlPFC) during the response period.

To characterize this finding further, group differences in each condition were examined relative to baseline fixation. Patients exhibited greater right vlPFC activation (i.e., positive power ratio) [(35,48,−8), t=3.10, k=14] during angry-congruent trials, relative to baseline, than the HC group. No between-group differences were noted for the angry-incongruent trials. Both groups exhibited vlPFC activation in both the angry-incongruent [GAD: (−21,55,−8), t=5.65, k=921, HC: (28,48,−15), t=5.33, k=2128] and angry-congruent trials relative to baseline [GAD: (28,55,0), t=5.64, k=771, HC: (28,48,−15), t=5.88, k=1897]. Although the peak was lateralized, the activation extended bilaterally in all cases.

Correlational analysis

Across both groups, a negative correlation was found between STAI trait anxiety scores and vlPFC activation during the response period for the threat bias contrast [R=−0.40, p<0.01] (Figure 2B). However, the correlations differed significantly between groups [Fisher’s r-to-Z=−2.05, p<0.05]. In the HC group, trait anxiety scores and vlPFC activation were negatively correlated [R=−0.44, p<0.04]; HC individuals with greater trait anxiety scores had greater negative power. This relation was explained by a negative correlation between trait anxiety and the vlPFC activation in response to angry-incongruent trials relative to baseline [R=−0.42, p<0.05]. However, in patients, vlPFC activation to threat bias was not significantly correlated with trait anxiety [R=0.23, p>0.4] or PARS scores [p>0.9].

No correlations between age, threat bias, and vlPFC activation during the response period for threat bias were noted across both groups or within each group separately [all p>0.2].

Additional analyses

For completeness, secondary analyses considered narrower frequency bands. Although the HC group showed no difference in power between angry-incongruent and congruent conditions in the 1-30 Hz frequency band, the HC group did exhibit vlPFC activation [(35,20,−8), t=4.01, k=14] when analyses were restricted to the 14-30 Hz frequency band.

Group differences were also detected in the other regions in narrower frequency bands for the angry bias contrast. In the 30-90 Hz band, the patient group exhibited greater negative power in the fusiform gyrus than the HC group [(−21,−50,−8), t=−3.33, k=7]. In the 8-12 Hz band, compared to the patient group, the HC group had greater activation in the putamen [(21,13,13), t=−3.20, k=11] and greater negative power in the parahippocampus [(−14,−29,−15, t=3.25, k=9].

During the face presentation period, no significant group differences in vlPFC were noted for 1-30 Hz; however, a group difference was detected in the precuneus for this broad frequency band. Compared to the HC group, the patients had greater negative power in response to angry relative to neutral faces [(−14,−78,49), t=3.84, k=7]. In addition, in the 14-30 Hz frequency band, the HC group exhibited greater activation in the parahippocampus than the patients [(21,−36,0), t=3.57, k=9].


To our knowledge, this is the first MEG study in pediatric anxiety. Previous dot-probe work using fMRI found group differences between GAD and healthy adolescents in vlPFC activation (Monk, et al., 2006). However, the limited temporal resolution of fMRI prevents conclusions regarding whether processing of threat faces or responses to the dot-probes induces abnormal vlPFC engagement in GAD. Using MEG, with both excellent temporal and spatial resolutions, a group difference in right vlPFC activation on the dot-probe task was isolated to the response period. This difference manifested for the same contrast used to calculate threat bias (angry-incongruent vs. angry-congruent). Patients exhibited greater negative oscillatory power in right vlPFC during this period compared to the HC group; trait anxiety scores in the HC group negatively correlated with this vlPFC activation. No group differences in vlPFC activation were detected in response to the angry face stimuli. Isolating the aberrant activation to the response period, rather than the face presentation period, clarifies the nature of attention-related perturbations in pediatric anxiety disorder. Novel treatments, such as attention training regimens or psychopharmacologic treatments, might be developed based on their specific ability to shape vlPFC function and direct attention toward task-relevant behaviors, rather than more general approaches.

The group difference in the threat bias contrast during the response period was explained by greater activation during the angry-congruent condition in patients than HC (Figure 3). In this condition, the probe occurs in the location previously occupied by the angry face. In the current study, individuals with high anxiety levels tended to avoid the threat cue (i.e., negative threat bias). Thus, greater power in response to congruent probes in the patient group may reflect difficulties disengaging attention from the location used to avoid the threat (i.e., neutral cue). Because activation in the current study was isolated to the response period, the vlPFC activation may serve to redirect attention, facilitating shifts of attentional resources away from threats in the service of probe identification. Therefore, vlPFC is needed to mobilize attention away from a stimulus pair that includes a threat to enable a response to the task-relevant probe.

The current dot-probe study revealed no between-group differences in behavior based on diagnosis. This raises questions on the functional significance of vlPFC findings, occurring in the absence of behavioral differences. Nevertheless, the significant correlation between threat bias and levels of anxiety on the STAI generally replicates the patterns found in previous research. Moreover, a lack of behavioral differences between anxious and healthy adolescents also clarifies interpretations because between-group differences in vlPFC engagement cannot be attributed to so-called performance confounds.

Typically, anxious individuals studied in low-stress environments exhibit behavioral responses biased towards threat on the dot-probe task (Bar-Haim, et al., 2007; Waters, Mogg, Bradley, & Pine, 2008). However, as in the current study, (Bar-Haim, et al., 2007) threat biases are altered in stressful environments. Anxious adults in mildly stressful (Constans, McCloskey, Vasterling, Brailey, & Mathews, 2004) or life-threatening situations often show avoidance, rather than vigilance towards threat. Moreover, in these situations, the level of anxiety often shows an association with threat avoidance as opposed to vigilance (Bar-Haim et al., 2010), as was found in the current study with anxiety on the STAI. Consistent with prior work, imaging environments may be mildly stressful and may influence attention bias (Monk, et al., 2006). Finally, it is common to observe between-group differences in neural activation in the absence of behavioral differences, possibly because imaging measures are more sensitive to group differences compared to behavioral indices. Larger studies may clarify these possibilities.

Caution is needed when considering the lateralization of group difference. Such findings could reflect Type II errors. However, in subsequent analyses using more liberal statistical thresholds, findings remained unilateral. Lateralization of group differences may indicate a distinction of function in the right and left vlPFC. Group differences in the right vlPFC activation during the response period for angry-congruent conditions may reflect response inhibition or reflexive orienting (Levy & Wagner, 2011). In other words, the right vlPFC activation may either reflect withholding a response until the dot-probe appears or reorienting attention to the appearance of task-relevant dot-probes. More research is needed to clarify these possibilities.

Several study limitations should be noted. First, patients included individuals with MDD and other comorbid illnesses. This approach was adopted to most directly extend prior fMRI research, which had used similar inclusion criteria. Moreover, after excluding MDD cases, the vlPFC group difference remained significant [t(37)=2.62,p<0.01]. Second, since the frontal cortex matures throughout adolescence, the broad age range may be a limitation. Again, we considered this age range to directly extend prior work (Monk, et al., 2006). Moreover, anxious and HC groups were age-matched to control for these developmental effects. Regardless, future longitudinal and cross-sectional studies examining more refined age-ranges may be able to identify developmentally-sensitive changes in the vlPFC and their impact on anxiety disorders. Third, like other MEG studies attempting to balance Type I and Type II errors, a broad frequency band, 1-30 Hz, was used to detect group differences in vlPFC activation. When individual frequency bands were examined, few yielded clear results, possibly due to a relatively small number of experimental trials. Future studies need to isolate the contributions from individual frequency bands. Finally, while our hypotheses were generated from fMRI studies, it is difficult to compare directly findings based on BOLD signal and MEG power due to complex relations between these two signals. To do so, the same group of subjects should be studied using both imaging modalities. Although the current study cannot answer questions that arise due to differences between MEG and fMRI, group differences in vlPFC activation during the response period were detected, suggesting the role of the vlPFC in orienting to task demands.

Despite these limitations, connecting aberrant vlPFC activation to aberrant attention control in individuals with pediatric anxiety is an important step toward improving treatments. Attention biases are often detected in individuals with anxiety and there is increasing interest in novel behavioral treatments (e.g., ABMT). Understanding the neural architecture engaged on attention orienting tasks, like the dot-probe task, may inform the further development of ABMT and other novel treatments. The current and prior imaging work suggests that an appropriate treatment target would be vlPFC activation when anxious patients must direct their attention to task-relevant behaviors following a distracting threat cue. In contrast, a less promising treatment target would be vlPFC function during initial attention orienting to threat; other structures (e.g., amygdala) may play a larger role in such abnormalities. Future neuroimaging work may provide further evidence that ABMT enhances both vlPFC function and attention control. Identifying vlPFC-mediated changes in attentional bias with training may support the use of ABMT as a prevention strategy in adolescents at-risk for developing anxiety disorders.


  • Pediatric anxiety is associated with abnormal ventrolateral prefrontal cortex (vlPFC) function.
  • Attention biases towards threat are often detected in individuals with anxiety using dotprobe paradigm.
  • Group differences in vlPFC activation were isolated to the response period using magnetoencephalography (MEG).
  • Aberrant ventrolateral PFC activation may reflect difficulties directing attention to task-relevant behaviors following threats.
  • Targeting treatment to behavioral and neural dysfunction may yield better long-term outcome.


This research was supported by the Intramural Research Program of the National Institute of Mental Health, National Institutes of Health (K99-MH091183-01 awarded to J.C.B.) and the National Alliance of Research on Schizophrenia and Depression (NARSAD) Young Investigator Award (J.C.B.). A previous version of these findings was presented at the Society of Biological Psychiatry annual conference in May 2009.


Conflicts of interest statement: No conflicts declared

All authors declare no potential conflicts of interest or financial interests related to this manuscript. The current affiliation of A.D. is with the Psychology Department, Duke University.

We thank Nina Shiffrin and Michelle Goldwin for their assistance in data acquisition.


  • Bar-Haim Y, Holoshitz Y, Eldar S, Frenkel TI, Muller D, Charney DS, et al. Life-threatening danger and suppression of attention bias to threat. American Journal of Psychiatry. 2010;167(6):694–698. [PubMed]
  • Bar-Haim Y, Lamy D, Pergamin L, Bakermans-Kranenburg MJ, van IMH. Threat-related attentional bias in anxious and nonanxious individuals: a meta-analytic study. Psychological Bulletin. 2007;133(1):1–24. [PubMed]
  • Birmaher B, Khetarpal S, Brent D, Cully M, Balach L, Kaufman J, et al. The Screen for Child Anxiety Related Emotional Disorders (SCARED): scale construction and psychometric characteristics. J Am Acad Child Adolesc Psychiatry. 1997;36(4):545–553. [PubMed]
  • Browning M, Holmes EA, Murphy SE, Goodwin GM, Harmer CJ. Lateral prefrontal cortex mediates the cognitive modification of attentional bias. Biological Psychiatry. 2010;67(10):919–925. [PMC free article] [PubMed]
  • Constans JI, McCloskey MS, Vasterling JJ, Brailey K, Mathews A. Suppression of attentional bias in PTSD. Journal of Abnormal Psychology. 2004;113(2):315–323. [PubMed]
  • Corbetta M, Patel G, Shulman GL. The reorienting system of the human brain: from environment to theory of mind. Neuron. 2008;58(3):306–324. [PMC free article] [PubMed]
  • Cornwell BR, Carver FW, Coppola R, Johnson L, Alvarez R, Grillon C. Evoked amygdala responses to negative faces revealed by adaptive MEG beamformers. Brain Research. 2008;1244:103–112. [PMC free article] [PubMed]
  • Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical Research. 1996;29(3):162–173. [PubMed]
  • Eldar S, Bar-Haim Y. Neural plasticity in response to attention training in anxiety. Psychological Medicine. 2010;40(4):667–677. [PubMed]
  • Eldar S, Yankelevitch R, Lamy D, Bar-Haim Y. Enhanced neural reactivity and selective attention to threat in anxiety. Biological Psychology. 2010;85(2):252–257. [PubMed]
  • Fox E, Russo R, Bowles R, Dutton K. Do threatening stimuli draw or hold visual attention in subclinical anxiety? Journal of Experimental Psychology: General. 2001;130(4):681–700. [PMC free article] [PubMed]
  • Hari R, Parkkonen L, Nangini C. The brain in time: insights from neuromagnetic recordings. Ann N Y Acad Sci. 2010;1191:89–109. [PubMed]
  • Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, et al. Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL): initial reliability and validity data. Journal of American Academy of Child & Adolescent Psychiatry. 1997;36(7):980–988. [PubMed]
  • Kovacs M. Child Depression Inventory. 2009. 1992 from http://www.pearsonassessments.com/HAIWEB/Cultures/enus/Productdetail.htm?Pid=015-8044-762&Mode=summary.
  • Levy BJ, Wagner AD. Cognitive control and right ventrolateral prefrontal cortex: reflexive reorienting, motor inhibition, and action updating. Annuals of the New York Academy of Sciences. 2011;1224(1):40–62. [PMC free article] [PubMed]
  • Lieberman MD, Cunningham WA. Type I and Type II error concerns in fMRI research: re-balancing the scale. Social & Cognitive Affective Neuroscience. 2009;4(4):423–428. [PMC free article] [PubMed]
  • Monk CS, Nelson EE, McClure EB, Mogg K, Bradley BP, Leibenluft E, et al. Ventrolateral prefrontal cortex activation and attentional bias in response to angry faces in adolescents with generalized anxiety disorder. American Journal of Psychiatry. 2006;163(6):1091–1097. [PubMed]
  • Monk CS, Telzer EH, Mogg K, Bradley BP, Mai X, Louro HM, et al. Amygdala and ventrolateral prefrontal cortex activation to masked angry faces in children and adolescents with generalized anxiety disorder. Arch Gen Psychiatry. 2008;65(5):568–576. [PMC free article] [PubMed]
  • Pine DS, Monk CS. The Development and Cognitive Neuroscience of Anxiety. In: Nelson CA, Luciana M, editors. Handbook of Developmental Cognitive Neuroscience. The MIT Press; Boston, Massachusetts: 2008. pp. 764–765.
  • Research Units on Pediatric Psychopharmacology Anxiety Study Group The Pediatric Anxiety Rating Scale (PARS): development and psychometric properties. J Am Acad Child Adolesc Psychiatry. 2002;41(9):1061–1069. [PubMed]
  • Rich BA, Carver FW, Holroyd T, Rosen HR, Mendoza JK, Cornwell BR, et al. Different neural pathways to negative affect in youth with pediatric bipolar disorder and severe mood dysregulation. Journal of Psychiatry Research. 2011;45(10):1283–94. [PMC free article] [PubMed]
  • Singh KD, Barnes GR, Hillebrand A, Forde EM, Williams AL. Task-related changes in cortical synchronization are spatially coincident with the hemodynamic response. Neuroimage. 2002;16(1):103–114. [PubMed]
  • Spielberger CD, Edwards CD, Montuori J, Lushene R. STAI-C. 2009 from http://www.mindgarden.com/products/staisch.htm#about.
  • Telzer EH, Mogg K, Bradley BP, Mai X, Ernst M, Pine DS, et al. Relationship between trait anxiety, prefrontal cortex, and attention bias to angry faces in children and adolescents. Biological Psychology. 2008;79(2):216–222. [PMC free article] [PubMed]
  • Tottenham N. NimStim Face Stimulus Set. 2009 from http://www.macbrain.org/resources.htm.
  • Vrba J, Robinson SE. Signal processing in magnetoencephalography. Methods. 2001;25(2):249–271. [PubMed]
  • Waters AM, Mogg K, Bradley BP, Pine DS. Attentional bias for emotional faces in children with generalized anxiety disorder. Journal of American Academy of Child & Adolescent Psychiatry. 2008;47(4):435–442. [PubMed]
  • Wechsler D. Wechsler Abbreviated Scale of Intelligence (WASI) Harcourt Assessment, Inc; San Antonio, TX: 1999.
  • Winterer G, Carver FW, Musso F, Mattay V, Weinberger DR, Coppola R. Complex relationship between BOLD signal and synchronization/desynchronization of human brain MEG oscillations. Human Brain Mapping. 2007;28(9):805–816. [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...