ARTICLES

Brain Mechanisms of Attention Orienting Following

Frustration: Associations With Irritability and Age
in Youths
Wan-Ling Tseng, Ph.D., Christen M. Deveney, Ph.D., Joel Stoddard, M.D., Katharina Kircanski, Ph.D.,
Anna E. Frackman, M.D., Jennifer Y. Yi, M.A., Derek Hsu, M.D., Elizabeth Moroney, B.A., Laura Machlin, B.A.,
Laura Donahue, B.A., Alexandra Roule, B.A., Gretchen Perhamus, B.A., Richard C. Reynolds, M.S.,
Roxann Roberson-Nay, Ph.D., John M. Hettema, M.D., Ph.D., Kenneth E. Towbin, M.D.,
Argyris Stringaris, M.D., Ph.D., Daniel S. Pine, M.D., Melissa A. Brotman, Ph.D., Ellen Leibenluft, M.D.

Objective: Childhood irritability is a common, impairing Results: Whole-brain activation analyses revealed associa-
problem with changing age-related manifestations that tions with irritability during attention orienting following frus-
predict long-term adverse outcomes. However, more in- tration. Irritability was positively associated with frontal-striatal
vestigation of overall and age-specific neural correlates is activation, specifically in the dorsolateral prefrontal cortex, in-
needed. Because youths with irritability exhibit exaggerated ferior frontal gyrus, and caudate. Age moderated the associa-
responses to frustrating stimuli, the authors used a frustration tion between irritability and activation in some frontal and
functional MRI (fMRI) paradigm to examine associations posterior regions (the anterior cingulate cortex, medial frontal
between irritability and neural activation and tested the gyrus, cuneus, precuneus, and superior parietal lobule [F=
moderating effect of age. 19.04–28.51, df=1, 189, partial eta squared=0.09–0.13]). Spe-
cifically, higher irritability was more strongly related to increased
Method: The authors studied a transdiagnostic sample of activation in younger youths compared with older youths.
195 youths with varying levels of irritability (disruptive mood
dysregulation disorder, N=52; anxiety disorder, N=42; at- Conclusions: Following frustration, levels of irritability cor-
tention deficit hyperactivity disorder, N=40; and healthy related with activity in neural systems mediating attention
volunteers, N=61). Irritability was measured by parent and orienting, top-down regulation of emotions, and motor exe-
child reports on the Affective Reactivity Index. The fMRI cution. Although most associations were independent of age,
paradigm was a cued-attention task differentiating neural dysfunction in the anterior cingulate cortex and posterior re-
activity in response to frustration (rigged feedback) from gions was more pronounced in young children with irritability.
activity during attention orienting in the trial following
frustration. Am J Psychiatry 2019; 176:67–76; doi: 10.1176/appi.ajp.2018.18040491

Irritability can be defined as an increased propensity to ex- evidence-based treatments for irritability (5). A better
perience anger and frustration, compared with peers (1). It is understanding of the pathophysiology of irritability is es-
a serious and common mental health problem among youths sential to guide the development of novel mechanism-based
(1, 2). It predicts adult depressive and anxiety disorders (a treatments for this common and impairing problem.
genetically mediated association) as well as long-term im- Given the increased proneness to frustration associated
pairment (e.g., high suicidality and low educational and in- with irritability, and working from a translational neurosci-
come attainment) (1–3). Irritability is also a core feature of ence perspective, the neural mechanisms mediating irrita-
the new DSM-5 category of disruptive mood dysregulation bility can be captured by studying an organism’s neural
disorder, which is characterized by developmentally in- responses to frustrative nonreward (1, 2). Originally oper-
appropriate, frequent, and severe temper outbursts (2, 4). ationalized in rodents, frustrative nonreward is the psycho-
Youths with severe irritability experience significant im- logical state induced by the failure to receive a reward that
pairment in multiple domains (e.g., at home, in school, and an organism has been conditioned to expect (6). In humans,
with peers) and have high rates of service use, hospitalization, investigators can model frustrative nonreward during func-
and school suspensions (2, 4). However, there are few tional MRI (fMRI) by evoking frustration in real time while
See related feature: CME course (p. 83)

Am J Psychiatry 176:1, January 2019 ajp.psychiatryonline.org 67

ATTENTION ORIENTING AND IRRITABILITY IN YOUTHS

TABLE 1. Demographic and Clinical Characteristics of the sizes of these studies are insufficiently large to generate clear
Study Sample conclusions (23). To address this issue, we recruited a rela-
Characteristic tively large sample of 195 youths. Moreover, no previous
Mean SD fMRI research, to our knowledge, has dissociated neural
responses to a frustrating event from the impact of frustra-
Age (years) 12.87 2.35
IQa 111.82 13.22
tion on the neural mechanisms mediating performance on
Socioeconomic statusb 32.03 17.07 a subsequent cognitive trial. Here, we examined the
Motionc 0.12 0.07 neural response to frustration and the impact of frus-
Children’s Global Assessment Scale 57.58 12.39 tration on attention orienting. We defined the latter as the
(past 6 months)d ability to disengage attention from the current focus, move
Dimensional measures
Affective Reactivity Index 3.07 2.68
attention to a selected alternative target, and direct attention
Screen for Child Anxiety Related 16.55 11.40 to that target (24). We chose an attention-orienting task that
Emotional Disorders engages frontal systems (e.g., the ventral attention network)
Conners’ Parent Rating Scale 59.48 14.05 known to mediate this process (14) and adapted a frustrating
N % task used previously (19) to track adjustments in this brain
Male 98 50.30
system following frustration. The task induces frustration by
Primary diagnosis falsely informing study subjects that they have committed
Disruptive mood dysregulation 52 26.67 errors. Given the critical role of the executive attention sys-
disorder tem in adapting to errors (14), this system may thus be im-
Attention deficit hyperactivity disorder 40 20.51 portant in the context of frustration. It is also noteworthy that
Anxiety 42 21.54
No diagnosis 61 31.28
recent work has implicated attention orienting (in response
to threat) as a potential mechanism of irritability (1, 2, 25).
Medications
Stimulants 45 23.08 We examined associations between irritability, measured
Antidepressants 31 15.90 dimensionally, and neural activations during a frustrating
Antipsychotics 10 5.13 attention-orienting task. Specifically, we studied a trans-
a
Measured with the Wechsler Abbreviated Scale of Intelligence; data are diagnostic clinical sample of 195 youths with varying levels of
missing for one participant. irritability. Based on previous work (19–22), we hypothesized
b
Measured with the Hollingshead Two-Factor Index of Socioeconomic
that irritability would be associated with perturbed activa-
Status; data are missing for 25 participants.
c
Calculated as the mean Euclidean distance of framewise volume shift after tion in frontal-striatal-amygdalar regions during feedback
censoring. processing and during attention orienting following frus-
d
Data were collected for patients only (i.e., children and youths with a diag-
trating feedback. Additionally, given the protracted nature
nosis of disruptive mood dysregulation disorder, anxiety disorder, or atten-
tion deficit hyperactivity disorder); data are missing for 13 patients. of prefrontal cortex development (12, 13), we hypothesized
that associations between irritability and activation in this
assessing its neural correlates (2). In a large, transdiagnostic region would be moderated by age.
sample, we used such methods to examine associations
between irritability and neural activation during an fMRI
METHOD
paradigm that models frustrative nonreward.
Our second goal was to examine age-related variation Participants
in the association between irritability and neural activity. The study sample included 195 children and adolescents ages
Normative responses to frustration change during develop- 8–18 (mean age, 12.9 years [SD=2.3]) with well-distributed
ment (7), yet little research has examined the underlying irritability levels and a mean above the cutoff for severe ir-
neural mechanisms (8). Indeed, while many behavioral ritability (26). Participants had primary diagnoses of dis-
studies document the development of emotion regulation (9, ruptive mood dysregulation disorder (characterized by
10), neuroimaging research has just begun to elucidate the severe, chronic irritability) (N=52), anxiety disorder (N=42),
maturation of brain systems supporting such capacity (11, 12). attention deficit hyperactivity disorder (ADHD) (N=40), or no
For example, the protracted development of the prefrontal disorder (N=61) (Table 1) (for further details, see Tables S1
cortex has been linked to age-related improvements in at- and S2 in the online supplement). They were recruited from
tention shifting and cognitive control, which can modulate National Institute of Mental Health (NIMH) clinics between
affective arousal (11–14). Although age-related brain mech- February 2012 and July 2016. On the Children’s Global As-
anisms have been examined in some clinical populations (e.g., sessment Scale (27), 58% of the sample had a score #60,
in persons with anxiety or depression) (15, 16), little work indicating at least “some noticeable problems” in several
has been done on irritability (8, 17, 18). areas. Most were seeking or receiving treatment; 48% were
Few studies utilize frustration paradigms in youths with being treated with medication, and half of these were taking
irritability, and those that do report neural dysfunction in two or more types of medications. Some patients with dis-
the prefrontal cortex and anterior cingulate cortex, striatum, ruptive mood dysregulation disorder were in an inpatient
amygdala, and parietal cortex (19–22). However, the sample treatment trial. Patients with anxiety disorder were in an

68 ajp.psychiatryonline.org Am J Psychiatry 176:1, January 2019

 TSENG ET AL.

outpatient treatment trial. FIGURE 1. Trial Structure During Frustration Runs of the Affective Posner 2 Taska
This study was approved by
“N” Trial: Rigged vs.
the NIMH institutional re- Positive Feedback
view board, and written con-
sent and assent, respectively,
were obtained from parents TOO SLOW!
and children. Exclusion cri-
teria included an IQ ,70, “N+1” Trial: After Rigged 60% correct
responses
vs. Positive Feedback
pervasive developmental or
neurological disorders, sub- TOTAL: $4.00

stance abuse within the past YOU WIN!

2 months, and lifetime history
40% correct
of psychosis, conduct disor- + responses
der, or unstable or chronic
medical illness. For detailed TOTAL: $5.00
diagnostic and clinical as- jitter ITI 300 ms 200 ms RESPONSE jitter ITI WRONG!
sessments and additional in- (1,000– (1,000– 100%
5,000 ms; 1,260 ms 3,000 ms;
clusion and exclusion criteria incorrect
M=3,000 ms) M=2,000 ms) responses
for the anxiety disorder
group, see the Methods sec- TOTAL: $4.00
tion in the online supplement. 2,000 ms
a
In frustration runs, 60% of correct responses were followed by rigged feedback (“TOO SLOW!”), and 40% of
Measures correct responses were followed by positive feedback (“YOU WIN!”). All incorrect responses were followed
We assessed irritability us- by negative feedback (“WRONG!”). Imaging analyses focused on the N+1 trial (red square) and the “feedback”
(blue square) portions of the task. Neural responses for the N+1 trial were modeled from the onset of the two
ing parent and child reports boxes for 2 seconds; neural responses for the feedback portion were modeled for the whole duration of the
from the Affective Reactivity feedback stimulus (2,000 ms). Significant associations with irritability emerged from the N+1 trial (i.e., the
Index (ARI) (28). To control attentional event following rigged feedback compared with positive feedback [red square]). ISI=interstimulus
interval; ITI=intertrial interval.
for co-occurring anxiety and
ADHD symptoms, we col-
lected parent and child reports from the Screen for Child positive feedback (60% and 40% of correct trials, respectively).
Anxiety Related Emotional Disorders (SCARED) rating scale Each run lasted 8 minutes (Figure 1) (see also Figure S2 in the
(29) and parent reports from the Conners’ Parent Rating online supplement). Imaging analyses focused on the two frus-
Scale–Revised (CPRS-R) long form (30). Total scores from tration runs. Group-level analyses were conducted separately
child and parent reports were averaged for the ARI and for the feedback and attention portions of the task, which
SCARED. The ADHD index T score from the CPRS-R was were separated by jitter (1,000–3,000 ms, with an average of
used to index ADHD symptoms. For participant distribution 2,000 ms) (Figure 1). The feedback portion (Figure 1) probed
(by diagnosis) along these three symptom dimensions, see neural activity during processing of rigged feedback com-
Figure S1 in the online supplement. For more than 90% of the pared with positive feedback. The attention portion (“N+1”
sample, these measures were collected within 3 months of trial) (Figure 1) assessed neural activity during the attentional
scanning. The 3-month window was selected on the basis of event following rigged feedback compared with positive
the stability of the measures and to maximize the sample size. feedback. At the end of each run, participants rated their
These measures have been shown to be highly stable across feelings of unhappiness and frustration on 9-point Likert
periods longer than 3 months (e.g., see Stringaris et al. [28]; scales. For further details regarding task procedures, see
see also the Methods section in the online supplement). the Methods section in the online supplement.

fMRI Paradigm Imaging Acquisition

Participants completed a frustrating attention task, the affec- Neuroimaging data were acquired on a 3-T General Electric
tive Posner 2 paradigm (31), which was adapted from a scanner (General Electric, Boston) using an eight-channel
previous fMRI study (19) and demonstrates good reliability head coil. A high-resolution anatomical scan (1-mm slices,
and validity (31). Participants were asked to identify a target three-dimensional spoiled gradient-echo sequence, 7° flip
following a cue by button press (left or right). The target angle, minimum full echo time, 2563256 matrix, 25.6-cm
appeared in the same location as the cue (valid trials) for 75% field of view) and gradient echo-planar imaging images were
of trials and in the opposite location (invalid trials) for 25%. collected (repetition time=2,300 ms, echo time=25 ms, 24-cm
The task consisted of two non-frustration runs during which field of view, voxel size=2.532.533 mm, 206 volumes per
participants received accurate or positive feedback and two run, flip angle=75° [N=134] or 90° [N=61]). For further details,
frustration runs during which they received rigged or see the Methods section in the online supplement.

Am J Psychiatry 176:1, January 2019 ajp.psychiatryonline.org 69

ATTENTION ORIENTING AND IRRITABILITY IN YOUTHS

Imaging Preprocessing and Individual Analysis simulation. By using methods designed recently to address
Data were analyzed using Analysis of Functional Neuro- concerns regarding inflated false positive rates (33), the
Images (AFNI). Preprocessing included despiking, tempo- cluster size surviving whole-brain correction was set at
ral alignment to the first acquired slice, coregistration, 703 mm3. At this threshold, we observed a large cluster of
smoothing, masking, and intensity scaling. Repetition time 68,000 mm3 for the three-way ARI-by-age-by-condition
pairs with a Euclidean norm motion derivative .1 mm were interaction and a cluster of 86,281 mm3 for the two-way
censored during linear regression. A general linear model ARI-by-condition interaction from the N+1 analyses. To fa-
estimated voxel-wise blood-oxygen-level-dependent signal cilitate interpretation, we extracted clusters by using a more
change (see the Methods section in the online supplement). stringent voxel-wise p value of 0.0001 (clusters $203 mm3)
(Table 2). Using this threshold, we created a conjunction map
Data analyses. Behavioral and post hoc imaging analyses were of the significant three-way and two-way interactions and
conducted with SPSS (IBM, Armonk, N.Y.). For behavioral calculated the shared voxels (i.e., voxels that showed sig-
results (frustration and unhappiness ratings, accuracy, and nificant effects for both the three-way ARI-by-age-by-condition
reaction time), see the Results section in the online sup- interaction and the two-way ARI-by-condition interaction).
plement. Group-level whole-brain activation analyses were Only 8.4% of the voxels were shared, suggesting that most
conducted using AFNI’s 3dMVM, a multivariate model-based voxels that showed a significant ARI-by-condition interaction
analysis of covariance (ANCOVA) appropriate for our fMRI were not moderated by age. To deconstruct significant inter-
data and study design (32). Analyses examined the effects of actions, mean activation across voxels in significant clusters
ARI, age, and the ARI-by-age interaction, with SCARED score, were extracted using AFNI’s 3dROIstat for follow-up anal-
CPRS-R score, and motion (for imaging data) as covariates. All yses in SPSS. For the N trial analyses, no significant clus-
variables were continuous, and they were mean centered to ters were observed (i.e., ARI, age, and their interaction were
reduce multicollinearity due to testing of the interaction term not associated with activation during processing of rigged
(ARI by age). Correlations between dimensional measures feedback compared with positive feedback).
(rs ,0.54) had an acceptable tolerance in the model with ARI, To address concerns regarding threshold-based cluster
age, ARI-by-age interaction, SCARED score, CPRS-R score, forming and replicability of imaging findings, we reanalyzed
and motion (variance inflation factors #1.46) (for further de- our data using the threshold-free cluster enhancement ap-
tails, see Table S2 in the online supplement). proach (34), with a family-wise error rate correction level of
0.05, by permutation testing (for further details, see the Re-
Imaging data. Imaging analyses focused on frustration runs sults section and Figure S9 in the online supplement). Results
because only these runs contained both positive and rigged were largely consistent with the original analysis in AFNI,
feedback. We did not directly compare non-frustration and except that two small clusters for the ARI-by-condition effect
frustration runs (e.g., on positive feedback) because there (in the left caudate and precentral gyrus) became nonsig-
was a fundamental difference in “baselines”: the baselines nificant (the right caudate remained significant).
in frustration runs were likely to be more saturated and el-
evated. However, we conducted a whole-brain analysis for
RESULTS
non-frustration runs only and, compared with the baseline,
found no significant associations between irritability and Imaging Data
neural activation during either positive feedback or atten- Significant findings emerged from the N+1 analysis, in the
tion orienting following positive feedback. comparison of neural activity occurring after rigged feed-
Only valid, correct trials were included, given the insuf- back compared with positive feedback during the frustra-
ficient numbers of other trial types. Separate analyses were tion runs (Figure 1). Specifically, we report interactions involving
conducted for the feedback portion of the N trial and the condition (after rigged compared with positive feedback)
attentional portion of the N+1 trial (Figure 1). For the feedback and ARI (dimensional measure of irritability) (i.e., ARI-by-
portion of the N trial, an ARI-by-age-by-condition (rigged age-by-condition and ARI-by-condition).
compared with positive feedback) ANCOVA was conducted
to assess activation during processing of rigged feedback ARI-by-age-by-condition. During the attentional event im-
compared with positive feedback and how it varied with mediately following rigged feedback compared with positive
irritability and age. For the N+1 trial, an ARI-by-age-by- feedback, activation in several posterior and frontal regions
condition (after rigged compared with positive feedback) varied with levels of irritability and as a function of partic-
ANCOVA was conducted to assess activation during the at- ipant age. These regions include the cuneus, precuneus,
tentional event following rigged feedback compared with superior parietal lobule, medial frontal gyrus, and anterior
positive feedback and how it varied with irritability and age. cingulate cortex (Table 2, Figure 2; see also Figure S3 in the
A whole-brain gray matter mask was used in the analyses, online supplement). Across these regions, higher irritability
including voxels for which data existed for $90% of par- was more strongly related to increased activation in younger
ticipants, voxel-wise p values of 0.001, and multiple-testing compared with older youths. Specifically, higher irritability
correction alphas of 0.05 via Monte Carlo cluster-size was significantly related to increased activation in young

70 ajp.psychiatryonline.org Am J Psychiatry 176:1, January 2019

 TSENG ET AL.

TABLE 2. N+1 Trial: Effect of Affective Reactivity Index (ARI)-by-Age-by-Condition and ARI-by-Condition Interactions From
Whole-Brain Activation Analysisa
Analysisd
Peak Coordinates
b 3
Regions Size (mm ) (x, y, z)c F p hp2 Correlation (r)e
ARI-by-age-by-condition interaction
Right cuneus 2,563 9, –79, 16 25.25 ,0.001 0.12
Right superior parietal lobule 2,563 31, –64, 44 28.51 ,0.001 0.13
Left precuneus and cuneus 1,469 –24, –69, 24 22.91 ,0.001 0.11
Left medial frontal gyrus and anterior 1,469 –9, 44, 26 23.94 ,0.001 0.11
cingulate cortex
Right pre- and post-central gyri 1,359 39, –19, 49 25.62 ,0.001 0.12
Left precuneus 1,047 –14, –66, 41 25.15 ,0.001 0.12
Left middle frontal gyrus 641 –34, 11, 41 20.80 ,0.001 0.10
Right middle occipital gyrus 625 29, –84, 21 25.18 ,0.001 0.12
Right postcentral gyrus 500 19, –29, 66 22.25 ,0.001 0.11
Right superior temporal gyrus 469 51, –51, 21 20.33 ,0.001 0.10
Right superior frontal gyrus 438 11, 54, 29 21.70 ,0.001 0.10
Right lingual and fusiform gyrus 422 21, –61, –4 21.18 ,0.001 0.10
Left precentral gyrus 391 –34, –9, 44 21.34 ,0.001 0.10
Right middle and superior frontal gyrus 375 24, 24, 41 19.23 ,0.001 0.09
Right middle frontal gyrus 313 24, –6, 44 19.75 ,0.001 0.10
Right fusiform gyrus 297 36, –59, –14 19.41 ,0.001 0.09
Left superior parietal lobule 281 –29, –61, 44 20.58 ,0.001 0.10
Left medial frontal gyrus 234 –6, –11, 51 19.04 ,0.001 0.09
Left superior frontal gyrus 203 –9, 14, 54 20.03 ,0.001 0.10
ARI-by-condition interaction
Left and right cingulate gyrus, right 13,594 9, 19, 41 34.81 ,0.001 0.16 0.40
superior frontal gyrus
Right middle frontal gyrus 5,844 36, 16, 41 31.63 ,0.001 0.14 0.38
Left middle frontal gyrus 2,469 –31, 21, 34 25.85 ,0.001 0.12 0.35
Right caudate, thalamus 2,422 11, –19, 19 27.72 ,0.001 0.13 0.36
Right dorsolateral prefrontal cortex 1,422 46, 31, 26 29.23 ,0.001 0.14 0.37
Right cuneus 1,250 9, –76, 19 23.65 ,0.001 0.11 0.33
Right precuneus 1,094 19, –69, 41 25.68 ,0.001 0.12 0.35
Left middle frontal gyrus 1,047 –34, –1, 46 22.33 ,0.001 0.11 0.33
Right inferior frontal gyrus 594 54, 26, 6 28.29 ,0.001 0.13 0.36
Left pre- and post-central gyri 531 –39, –19, 39 20.81 ,0.001 0.10 0.32
Left parahippocampal gyrus 469 –16, –39, –4 23.86 ,0.001 0.11 0.34
Left caudate 453 –9, 6, 21 21.53 ,0.001 0.10 0.32
Right superior temporal gyrus 406 41, –49, 19 21.37 ,0.001 0.10 0.32
Right precentral gyrus 344 56, 6, 6 21.29 ,0.001 0.10 0.32
Left precentral gyrus 266 –61, –1, 14 20.85 ,0.001 0.10 0.32
Left cingulate gyrus 203 –6, –16, 41 20.77 ,0.001 0.10 0.32
Left superior frontal gyrus 203 –6, 11, 59 19.68 ,0.001 0.10 0.31
a
The condition effect refers to the attention portion of the trial immediately after receiving rigged compared with positive feedback (i.e., the N+1 trial).
b
The region comprising the greatest portion of the cluster extent is presented. At a voxel-wise p value of 0.001, the largest significant cluster was 86,281 mm3.
To facilitate interpretation, we extracted clusters using the more stringent voxel-wise p value of 0.0001. At this threshold, clusters $203 mm3 survived whole-
brain correction at an alpha of 0.05.
c
Coordinates are in Talairach space.
d
Post hoc analysis of covariance on the mean blood-oxygen-level-dependent signal for extracted clusters are presented. The df value for all F statistics is 1, 188.
e
Correlations between ARI and the difference in brain activation after receiving rigged compared with positive feedback (rigged minus positive) are presented,
after adjusting for motion and symptoms of anxiety and attention deficit hyperactivity disorder. It is noteworthy that these correlations may be inflated given
that they were computed on the basis of extracted signal change from voxels that survived whole-brain correction (for further details, see Vul et al. [43]).

children (ages 8–11.5) and younger adolescents (ages 11.5–14), precuneus, and inferior frontal gyrus (Table 2, Figure 3; see
with the strongest correlation in young children. However, also Figure S4 in the online supplement). Most (91.6%) sig-
irritability was not related to activation in older adolescents nificant voxels in these regions did not overlap with the
(ages 14–18). Findings in the medial prefrontal cortex and significant voxels for the three-way interaction described
anterior cingulate cortex are summarized in Figure 2. above. The few overlapping regions included the left and
right cingulate gyrus, middle frontal gyrus, cuneus, and
ARI-by-condition. This two-way interaction yielded signifi- precuneus. Given that these regions were qualified by a
cant findings in the left and right cingulate gyrus, middle three-way interaction of ARI by age by condition, we do
frontal gyrus, caudate, dorsolateral prefrontal cortex, cuneus, not interpret them here. Regions showing a significant

Am J Psychiatry 176:1, January 2019 ajp.psychiatryonline.org 71

ATTENTION ORIENTING AND IRRITABILITY IN YOUTHS

FIGURE 2. Age Moderating the Association Between Irritability and Activation in the Medial Prefrontal Cortex and Anterior Cingulate
Cortex During Attention Orienting Following Rigged Compared With Positive Feedbacka
A B
R L F
28.42

Act
ivat
ion
D
iffe
0

ren
ce
Ag

I
AR
e

Z=26 X=–7
Predicted Activation:
After Rigged vs. Positive Feedback

–0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8

C
(After Rigged vs. Positive Feedback)

8–11.5 years 11.5–14 years 14–18 years

0.80 0.80 0.80
r=0.61 r=0.27 r=–0.13
0.60 0.60 0.60
% Signal Change

0.40 0.40 0.40

0.20 0.20 0.20
0 0 0
–0.20 –0.20 –0.20
–0.40 –0.40 –0.40
–0.60 –0.60 –0.60
–0.80 –0.80 –0.80
–6 –4 –2 0 2 4 6 –6 –4 –2 0 2 4 6 –6 –4 –2 0 2 4 6

ARI ARI ARI

a
Panel A shows the left medial prefrontal cortex extending to the anterior cingulate cortex from the whole-brain N+1 trial activation analysis. During the
attentional portion of the trial, activation after receiving rigged compared with positive feedback varied with irritability (i.e., Affective Reactivity
Index [ARI] scores) and age. Panel B shows the associations among ARI, age, and brain activation (all variables were continuous). The blood-
oxygen-level-dependent percent signal change for this cluster was extracted for each condition (the N+1 trial after rigged feedback and after
positive feedback) for each study subject. These values were entered in the same analysis of covariance model (controlling for symptoms of anxiety
and attention deficit hyperactivity disorder [ADHD] and motion) as in the main analysis, and predicted percent signal changes were generated. The
differences between the predicted percent signal changes after rigged compared with positive feedback are plotted. The three-dimensional graph
shows that after receiving rigged compared with positive feedback, younger youths with high irritability exhibited increased activation. Panel
C shows partial regression plots (controlling for symptoms of anxiety and ADHD and motion) by age tertiles (for each age group, N=65) depicting
individual data points and the association between ARI (mean-centered) and the percent signal change difference between trials occurring after
rigged feedback compared with positive feedback. Age was treated as a continuous variable in the analyses. The age tertiles here are for vi-
sualization purposes only. Higher irritability was more strongly related to increased activation on this contrast in early childhood (ages 8–11.5)
compared with early adolescence (ages 11.5–14). Irritability was not related to activation in late adolescence (ages 14–18). It is noteworthy that
these correlations may be inflated given that they were computed on the basis of extracted signal change from voxels that survived whole-brain
correction (43). F=F value; L=left; R=right.

ARI-by-condition effect without a qualifying three-way depression by conducting whole-brain analyses with de-
interaction included the dorsolateral prefrontal cortex pressive symptoms (measured by the self-rated Children’s
(Figure 3), caudate (extending to the thalamus), and inferior Depression Inventory [35]) as the main dimension. There
frontal gyrus, among others (Table 2; see also Figure S4 in the were no significant associations between depressive symp-
online supplement). Across these regions, higher irritability toms and neural activations (threshold voxel-wise p=0.001,
was related to increased activation following rigged feedback cluster extent $703 mm3). Depressive symptoms were
compared with positive feedback. evaluated this way and were not treated as a covariate in the
main analyses (unlike anxiety and ADHD symptoms) because
Post Hoc Analyses: Depressive Symptoms, Frustration 28 participants had missing data.
and Unhappiness Ratings, Medication, DSM-5 Diagnosis, We also conducted analyses to examine the effects of
and Gender “state irritability” measured by self-ratings of frustration
Given the longitudinal link between childhood irritability and unhappiness obtained at the end of each of the two
and depression later in life (3), we evaluated the effect of frustration runs (see the Results section in the online

72 ajp.psychiatryonline.org Am J Psychiatry 176:1, January 2019

 TSENG ET AL.

supplement). The only significant finding was a positive FIGURE 3. Association Between Irritability and Dorsolateral
association between unhappiness and activation in the right Prefrontal Cortex Activation During Attention Orienting
Following Rigged Compared With Positive Feedbacka
superior temporal gyrus during processing of rigged feedback
compared with positive feedback. These analyses have lim- A
R L F
itations because of the problems associated with the state 32.44
measures (see the Results section in the online supplement).
Nonetheless, they suggest that the neural substrates un-
derlying transient, subjective feelings of frustration and
unhappiness may differ from the neural substrates mediating 0
trait irritability.
We evaluated the confounding effect of medications
by iteratively excluding participants by medication class
(stimulants, antidepressants, and antipsychotics) (for fur- Z=26 X=46
ther details, see Tables S3 and S4 in the online supplement).
B
All significant findings remained. Additionally, analyses 1.00
comparing medicated (N=25) with nonmedicated (N=100)

(After Rigged vs. Positive Feedback)

participants on the day of the scan did not reveal significant
between-group differences on the imaging findings. 0.50
Categorical analyses using diagnoses instead of symptom

% Signal Change
dimensions yielded null results in whole-brain activation but
a few findings in functional connectivity (for further details, 0
see the Results section in the online supplement).
We also examined the moderating effect of gender and
found that most of the main imaging results were not mod- –0.50
erated by gender. However, in the inferior parietal lobule,
pre- and post-central gyri, and insula, irritability was related
to increased activation in younger boys and decreased ac- r=0.37
–1.00
tivation in older boys (see also the Results section in the
–5 –3 0 3 5 8
online supplement).
Irritability (ARI)
Additional Region-of-Interest and Functional a
Panel A shows the dorsolateral prefrontal cortex for the whole-brain N+1
Connectivity Analyses trial activation analysis. During the attentional portion of the trial, acti-
Region-of-interest analyses in the amygdala and striatum vation after receiving rigged compared with positive feedback varied
with irritability (i.e., Affective Reactivity Index [ARI] scores). Panel B shows
revealed positive associations between ARI and striatal ac- partial regression plots (controlling for symptoms of anxiety and at-
tivity during the N+1 trial but not during feedback (age tention deficit hyperactivity disorder and motion) depicting individual
moderated some associations) (see the Results section in the data points and the association between ARI (mean-centered) and the
percent signal change difference between trials occurring after rigged
online supplement). We also analyzed functional connec- compared with positive feedback. Higher irritability was related to more
tivity with the inferior frontal gyrus and amygdala seeds (see activation on this contrast. It is noteworthy that these correlations may
the Results section, Tables S5 and S6, and Figures S5–S8 in be inflated given that they were computed on the basis of extracted
signal change from voxels that survived whole-brain correction (for
the online supplement). Notably, we found that higher irri- further details, see Vul et al. [43]). F=F value; L=left; R=right.
tability was related to decreased functional connectivity be-
tween the left inferior frontal gyrus and periaqueductal gray
(extending to the culmen) during the N+1 trial (see Figure S6 event but not during processing of the frustrating feedback
in the online supplement). itself. Two specific findings emerged. First, higher irritabil-
ity was related to increased activation in multiple frontal-
striatal regions independently of age. Second, in other
DISCUSSION
regions, associations between irritability and activation were
This study investigated the neural correlates of attention moderated by age, such that associations were stronger
orienting following frustration in a transdiagnostic sample of in younger children compared with older children. These
195 youths with varying levels of irritability. We used a novel findings suggest that promising treatments for irritability
paradigm to model the frustration that irritable youths are may target frontal-striatal regions and that interventions
prone to experience, particularly when reward is withheld could be prioritized for younger youths with high irritability.
and they are frustrated and asked to adjust their behaviors Irritability was associated with dysfunction in the pre-
(e.g., to stop playing video games and start homework in- frontal cortex, anterior cingulate cortex (20–22, 36, 37), and
stead). We found that irritability was associated with neural striatum (19, 36). This is consistent with previous research
activation during attention orienting following a frustrating in youths and adults with irritability or related phenotypes

Am J Psychiatry 176:1, January 2019 ajp.psychiatryonline.org 73

ATTENTION ORIENTING AND IRRITABILITY IN YOUTHS

(anger and trait aggression) (19–22, 36, 37). Specifically, while strategies that help irritable children allocate attention
engaged in the attentional part of the task after being effectively.
frustrated, highly irritable youths exhibited increased acti- Age moderated the association between irritability and
vation in multiple frontal regions (anterior cingulate cortex, brain activation during attention orienting following frus-
dorsolateral prefrontal cortex, and inferior frontal gyrus) tration in several frontal and posterior regions (anterior
implicated in cognitive control, executive attention, and at- cingulate cortex, medial frontal gyrus, cuneus, and pre-
tention orienting (38, 39). Highly irritable youths also showed cuneus). Specifically, higher irritability was related to in-
increased activation in the striatum, which has a regulatory creased activation in children and younger adolescents but
influence on the cortex and is involved in motor control and not older adolescents; the association was particularly strong
eye movements as well as in set shifting (40). in young children. This finding is inconsistent with the only
Notably, irritability was not associated with performance previous study that examined the interacting effect of age
deficits (e.g., poor accuracy and slower reaction time). Thus, and irritability, which found that as age increased, higher
the observed increased activation may reflect a compensatory irritability was associated with more frontal-striatal-thalamic
mechanism (i.e., compared with healthy youths, irritable activation (17). This discrepancy could be explained by dif-
youths may require more robust recruitment of these regions ferences in sample size (N=30 compared with N=195), sample
following frustration to regulate their negative affect, focus characteristics such as age (4–12 years compared with 8–18
on the task at hand, and meet task demands). Importantly, years), and the nature of the sample (nonclinical compared
our finding is best explained by irritability and not better with clinical) as well as fMRI paradigm (emotionally or
attributed to co-occurring anxiety, ADHD, or depressive neutrally valenced video clips compared with a frustrating
symptoms. Furthermore, we found no associations between cued-attention task). Young and highly irritable children
irritability and activation during non-frustration runs. This may be most susceptible to affectively charged stimuli and
observation suggests that trait-specific neural correlates may have the greatest difficulty disengaging attention from
manifest in irritable youths when they are frustrated. Such negatively arousing stimuli. Although these youths did not
context specificity is consistent with our event-related show performance deficits, their ability to perform the task
analysis contrasting activity during performance of the despite immature neural circuitry may have required pro-
attention-orienting task following frustrating feedback longed and inefficient computational processes that gen-
compared with non-frustrating feedback. Because of the erated increased regional neural activity (41). Indeed, the
insufficient number of incorrect trials, we were unable to increased anterior cingulate cortex activation seen in young
examine the neural responses to errors (in which reward is children, compared with older youths and adults, may re-
omitted and not expected). An interesting and important flect neural inefficiency in cognitive control (41).
question is whether irritability has similar associations Contrary to some previous studies, we did not find an
with responses to errors, responses to “pure” frustrative association between irritability and amygdala activation in
nonreward (i.e., in which reward is expected and omitted be- response to a frustrating event (19, 37). This could be due to
cause of changed contingencies), and responses to frus- differences in frustration paradigms (i.e., a block design [37]
trative nonreward due to deception (rigged feedback). compared with an event-related design, or no jitter [19]
Despite comparable performance on a simple attention compared with jitter between attentional and feedback
task, irritable youths exhibited heightened frontal activation portions of the task [the latter allows for separation of at-
following frustration, suggesting a requirement for greater tention orienting and feedback processing]). Alternatively,
prefrontal cortex engagement to adjust to frustration and the unreliability of amygdala activation may hamper repli-
achieve performance comparable to that of less irritable cation (42). It is noteworthy that our null amygdala finding
youths. Many everyday tasks (e.g., schoolwork, homework, or is consistent with a recent study that also adopted a dimen-
transitions between activities) are much more cognitively sional approach to examine irritability and brain function
demanding than laboratory tasks, such as the one used here. during frustration (20).
Therefore, inefficiency in systems that facilitate post- There are several limitations to our study. First, our
frustration adjustment could lead irritable youths to strug- findings may apply only to the disorders sampled (disruptive
gle in daily life. Moreover, whereas many irritable youths in mood dysregulation disorder, anxiety disorder, and ADHD)
our sample were medicated, the impact of medications on and not to other disorders (e.g., depression and bipolar dis-
their brain function was not evident in our analysis. This order) in which irritability is also common. Second, given the
suggests that current medications may fail to normalize the high correlation between chronological age and puberty sta-
particular neural dysfunction reported here. There is clearly tus, we used age as a proxy for development. Future work is
a need for new treatments, including nonpharmacological needed to directly examine the effect of puberty. Third, al-
approaches (e.g., real-time fMRI neurofeedback or trans- though our cross-sectional design is a helpful starting point
cranial magnetic stimulation targeting frontal-striatal re- to understand brain function over development, only longi-
gions). Additionally, given that neural dysfunction associated tudinal studies can elucidate individual differences in de-
with irritability was found during attentional processes fol- velopmental trajectories. Fourth, as in most studies of youths
lowing frustration, intervention efforts might emphasize with severe impairments, medication may confound the

74 ajp.psychiatryonline.org Am J Psychiatry 176:1, January 2019

 TSENG ET AL.

results; however, our post hoc analyses do not support con- Received April 30, 2018; revision received July 3, 2018; accepted July 30,
founding by medication. Finally, this study used raw, reported 2018; published online Oct. 19, 2018.
scores of irritability symptoms. Other phenotyping ap-
proaches (e.g., latent variable modeling) may provide addi- REFERENCES
tional perspectives. 1. Brotman MA, Kircanski K, Stringaris A, et al: Irritability in youth:
a translational model. Am J Psychiatry 2017; 174:520–532
2. Leibenluft E: Pediatric irritability: a systems neuroscience approach.
CONCLUSIONS Trends Cogn Sci 2017; 21:277–289
3. Vidal-Ribas P, Brotman MA, Valdivieso I, et al: The status of irri-
In a sample characterized by a mix of mood, anxiety, and tability in psychiatry: a conceptual and quantitative review. J Am
ADHD symptoms in the clinically impairing range, we found Acad Child Adolesc Psychiatry 2016; 55:556–570
unique associations of irritability with neural systems me- 4. Roy AK, Lopes V, Klein RG: Disruptive mood dysregulation dis-
order: a new diagnostic approach to chronic irritability in youth.
diating attention orienting, top-down regulation of emotions, Am J Psychiatry 2014; 171:918–924
and motor execution following frustrative nonreward. These 5. Stringaris A, Vidal-Ribas P, Brotman MA, et al: Practitioner review:
associations were not attributed to co-occurring ADHD, definition, recognition, and treatment challenges of irritability in
anxiety, or depressive symptoms. Although most associa- young people. J Child Psychol Psychiatry 2018; 59:721–739
6. Amsel A: The role of frustrative nonreward in noncontinuous re-
tions were independent of age, dysfunction in the anterior
ward situations. Psychol Bull 1958; 55:102–119
cingulate cortex and some posterior regions was more 7. Wakschlag LS, Estabrook R, Petitclerc A, et al: Clinical implications
pronounced in young children with irritability. The neu- of a dimensional approach: the normal:abnormal spectrum of early
ral dysfunction did not seem to be altered by medications, irritability. J Am Acad Child Adolesc Psychiatry 2015; 54:626–634
highlighting a need for new treatments, including non- 8. Pagliaccio D, Pine DS, Barch DM, et al: Irritability trajectories,
pharmacological approaches, that target the common and cortical thickness, and clinical outcomes in a sample enriched for
preschool depression. J Am Acad Child Adolesc Psychiatry 2018;
impairing symptom of irritability. Because associations be- 57:336–342.e6
tween irritability and brain function were found during at- 9. John OP, Gross JJ: Healthy and unhealthy emotion regulation:
tentional processes following frustration, intervention efforts personality processes, individual differences, and life span devel-
could target strategies that help children with irritability opment. J Pers 2004; 72:1301–1333
10. Posner MI, Rothbart MK, Sheese BE, et al: Developing attention:
regulate the negative affect and arousal elicited by frustrating
behavioral and brain mechanisms. Adv Neurosci (Hindawi) 2014; vol
events and flexibly shift their attention to focus on the task at 2014, article ID 405094 (http://dx.doi.org/10.1155/2014/405094)
hand. 11. Cohen AO, Breiner K, Steinberg L, et al: When is an adolescent an
adult? assessing cognitive control in emotional and nonemotional
AUTHOR AND ARTICLE INFORMATION contexts. Psychol Sci 2016; 27:549–562
12. Ahmed SP, Bittencourt-Hewitt A, Sebastian CL: Neurocognitive
From the Department of Health and Human Services, Emotion and De-
bases of emotion regulation development in adolescence. Dev Cogn
velopment Branch and Scientific and Statistical Computing Core, NIMH,
Neurosci 2015; 15:11–25
Bethesda, Md.; the Department of Psychiatry and the Virginia Institute for
13. Ordaz SJ, Foran W, Velanova K, et al: Longitudinal growth curves of
Psychiatric and Behavioral Genetics, Virginia Commonwealth University,
brain function underlying inhibitory control through adolescence.
Richmond; the Department of Psychology, Wellesley College, Wellesley,
J Neurosci 2013; 33:18109–18124
Mass.; the University of Colorado School of Medicine, Aurora, Colo.; the
14. Petersen SE, Posner MI: The attention system of the human brain:
Department of Medicine, Stanford University, Stanford, Calif.; the De-
20 years after. Annu Rev Neurosci 2012; 35:73–89
partments of Psychology and Neuroscience, University of North Carolina
15. Luby JL, Agrawal A, Belden A, et al: Developmental trajectories of
at Chapel Hill; Emory University School of Medicine, Atlanta; the De-
the orbitofrontal cortex and anhedonia in middle childhood and
partment of Psychology, University of California at Los Angeles; and
risk for substance use in adolescence in a longitudinal sample
University of Michigan Medical School, Ann Arbor.
of depressed and healthy preschoolers. Am J Psychiatry 2018;
Send correspondence to Dr. Tseng ([email protected]). 175:1010–1021
The computational resources of the NIH High Performing Computation 16. Tamnes CK, Walhovd KB, Torstveit M, et al: Performance moni-
system’s Biowulf cluster was used in this study (http://hpc.nih.gov). toring in children and adolescents: a review of developmental
Drs. Tseng and Stoddard have received research grant support from
changes in the error-related negativity and brain maturation. Dev
Cogn Neurosci 2013; 6:1–13
NIMH (grant K99MH110570 to Dr. Tseng and grant K23MH113731 to Dr.
Stoddard). The other authors report no financial relationships with com-
17. Karim HT, Perlman SB: Neurodevelopmental maturation as a
mercial interests.
function of irritable temperament: insights from a naturalistic
emotional video viewing paradigm. Hum Brain Mapp 2017; 38:
Supported by the Intramural Research Program of NIMH, conducted 5307–5321
under projects ZIA-MH002778 (clinical protocol NCT00006177), ZIA- 18. Wakschlag LS, Perlman SB, Blair RJ, et al: The neurodevelopmental
MH002781 (clinical protocol NCT00018057), and ZIA-MH002786 (clin- basis of early childhood disruptive behavior: irritable and callous
ical protocol NCT00025935). phenotypes as exemplars. Am J Psychiatry 2018; 175:114–130
The authors thank the children and families who participated in this study 19. Deveney CM, Connolly ME, Haring CT, et al: Neural mechanisms of
as well as the staff of the Emotion and Development Branch and the frustration in chronically irritable children. Am J Psychiatry 2013;
Scientific and Statistical Computing Core at NIMH. The authors also 170:1186–1194
thank Anderson Winkler, M.D., D.Phil. (from the NIMH Emotion and De- 20. Grabell AS, Li Y, Barker JW, et al: Evidence of non-linear associa-
velopment Branch) for assistance on permutation analysis of linear models tions between frustration-related prefrontal cortex activation and
and threshold-free cluster enhancement analysis, Dan Barlow for pro- the normal:abnormal spectrum of irritability in young children.
gramming support, and Gang Chen, Ph.D., for imaging statistics guidance. J Abnorm Child Psychol 2018; 46:137–147

Am J Psychiatry 176:1, January 2019 ajp.psychiatryonline.org 75

ATTENTION ORIENTING AND IRRITABILITY IN YOUTHS

21. Perlman SB, Jones BM, Wakschlag LS, et al: Neural substrates of 32. Chen G, Adleman NE, Saad ZS, et al: Applications of multivariate
child irritability in typically developing and psychiatric populations. modeling to neuroimaging group analysis: a comprehensive alterna-
Dev Cogn Neurosci 2015; 14:71–80 tive to univariate general linear model. Neuroimage 2014; 99:571–588
22. Rich BA, Carver FW, Holroyd T, et al: Different neural pathways to 33. Cox RW, Chen G, Glen DR, et al: fMRI clustering in AFNI: false-
negative affect in youth with pediatric bipolar disorder and severe positive rates redux. Brain Connect 2017; 7:152–171
mood dysregulation. J Psychiatr Res 2011; 45:1283–1294 34. Smith SM, Nichols TE: Threshold-free cluster enhancement: ad-
23. Poldrack RA, Baker CI, Durnez J, et al: Scanning the horizon: to- dressing problems of smoothing, threshold dependence and local-
wards transparent and reproducible neuroimaging research. Nat isation in cluster inference. Neuroimage 2009; 44:83–98
Rev Neurosci 2017; 18:115–126 35. Kovacs M: Children’s Depression Inventory. North Tonawanda, NY,
24. Posner MI, Walker JA, Friedrich FJ, et al: Effects of parietal injury Multi-Health Systems, 1992
on covert orienting of attention. J Neurosci 1984; 4:1863–1874 36. Abler B, Walter H, Erk S: Neural correlates of frustration. Neuro-
25. Kircanski K, White LK, Tseng WL, et al: A latent variable approach report 2005; 16:669–672
to differentiating neural mechanisms of irritability and anxiety in 37. Pawliczek CM, Derntl B, Kellermann T, et al: Anger under control:
youth. JAMA Psychiatry 2018; 75:631–639 neural correlates of frustration as a function of trait aggression.
26. Kircanski K, Zhang S, Stringaris A, et al: Empirically derived patterns PLoS One 2013; 8:e78503
of psychiatric symptoms in youth: a latent profile analysis. J Affect 38. MacDonald AW 3rd, Cohen JD, Stenger VA, et al: Dissociating the
Disord 2017; 216:109–116 role of the dorsolateral prefrontal and anterior cingulate cortex in
27. Shaffer D, Gould MS, Brasic J, et al: A Children’s Global Assessment cognitive control. Science 2000; 288:1835–1838
Scale (CGAS). Arch Gen Psychiatry 1983; 40:1228–1231 39. Corbetta M, Shulman GL: Control of goal-directed and stimulus-
28. Stringaris A, Goodman R, Ferdinando S, et al: The Affective Re- driven attention in the brain. Nat Rev Neurosci 2002; 3:201–215
activity Index: a concise irritability scale for clinical and research 40. Ravizza SM, Ciranni MA: Contributions of the prefrontal cortex and
settings. J Child Psychol Psychiatry 2012; 53:1109–1117 basal ganglia to set shifting. J Cogn Neurosci 2002; 14:472–483
29. Birmaher B, Brent DA, Chiappetta L, et al: Psychometric properties 41. Luna B, Padmanabhan A, O’Hearn K: What has fMRI told us about
of the Screen for Child Anxiety Related Emotional Disorders the development of cognitive control through adolescence? Brain
(SCARED): a replication study. J Am Acad Child Adolesc Psychiatry Cogn 2010; 72:101–113
1999; 38:1230–1236 42. van den Bulk BG, Koolschijn PCMP, Meens PHF, et al: How stable is
30. Conners CK, Sitarenios G, Parker JDA, et al: The revised Conners’ activation in the amygdala and prefrontal cortex in adolescence? a
Parent Rating Scale (CPRS-R): factor structure, reliability, and cri- study of emotional face processing across three measurements. Dev
terion validity. J Abnorm Child Psychol 1998; 26:257–268 Cogn Neurosci 2013; 4:65–76
31. Tseng W-L, Moroney E, Machlin L, et al: Test-retest reliability and 43. Vul E, Harris C, Winkielman P, et al: Puzzlingly high correlations in
validity of a frustration paradigm and irritability measures. J Affect fMRI studies of emotion, personality, and social cognition. Perspect
Disord 2017; 212:38–45 Psychol Sci 2009; 4:274–290

76 ajp.psychiatryonline.org Am J Psychiatry 176:1, January 2019