Neural Mechanisms of Inhibitory Control in Preadolescent Irritability: Insights from the ABCD Study

Abstract

Objective:

Elevated pediatric irritability is a transdiagnostic symptom that predicts multiple mental health problems in adolescence and adulthood. Altered top-down regulatory networks, such as inhibitory control networks that suppress an impulse in favor of goal-directed behavior, are thought to contribute to high levels of youth irritability. Nevertheless, little work has examined links between youth irritability and neural processes supporting inhibitory control in large diverse samples, nor have they focused on the key period ramping up to adolescence (i.e., preadolescence).

Method:

Functional MRI data from 5,380 preadolescents (age M=9.97 years, SD=0.62) in the baseline Adolescent Brain and Cognitive Development (ABCD) Study were analyzed. Parents reported on their preadolescent’s irritability. The stop signal task (SST) was leveraged to probe successful and failed inhibitory control. Activation and functional connectivity with amygdala, ventral striatum, and prefrontal seed regions were calculated during the SST and used in whole brain and region of interest (ROI) group-level analyses evaluating irritability effects.

Results:

Preadolescents with higher levels of irritability displayed decreases in functional connectivity among amygdala, ventral striatum, and prefrontal cortex regions during both successful and failed inhibitory control conditions. These results remained after adjusting for co-occurring anxiety, depression, and attention-deficit/hyperactivity symptoms.

Conclusions:

Findings suggest neural aberrations in inhibitory control play a role in the pathophysiology of preadolescent irritability and associations are not merely due to co-occurring symptoms. Neural mechanisms of inhibitory control associated with irritability may provide novel intervention targets.

Introduction

Elevated irritability in youth is characterized by an increased proneness to frustration and anger relative to peers(Brotman et al., 2017; Leibenluft, 2017). Elevated irritability is concurrently associated with internalizing and externalizing problems, and childhood irritability predicts adolescent and adult mental health problems, functional impairment, and suicidality (Sorcher et al., 2022; Stringaris et al., 2009). Recent work has focused on understanding the pathophysiology of irritability through altered top-down regulatory networks, including inhibitory control networks. Inhibitory control, a cognitive process used to suppress an impulse in favor of more goal-directed behavior (Luna et al., 2015), is a core element of emotion regulation and is necessary to regulate temper outbursts associated with irritability. Indeed, prior research has shown that youths with higher levels of irritability exhibit impaired inhibitory control (Perhamus & Ostrov, 2023)and often have co-occurring impulse control problems (e.g., oppositional defiant, attention-deficit/hyperactivity disorders).

However, despite the theoretical importance of inhibitory control in irritability, relatively little research has examined these associations. While some behavioral studies have suggested impaired inhibitory control in youth with irritability (Cardinale et al., 2023; Perhamus & Ostrov, 2023)or related symptoms (e.g., disruptive behavior or anger; Gagne & Hill Goldsmith, 2011; Martel et al., 2013), other studies have not observed associations between irritability and inhibitory control (Cardinale et al., 2024; Colonna et al., 2022; Karalunas et al., 2021; McKay et al., 2024). These inconsistencies among behavioral studies may be due to differences in sample age: significant associations have been observed in early childhood (Gagne & Hill Goldsmith, 2011; Martel et al., 2013; Perhamus & Ostrov, 2023) and a narrow band of adolescence (ages 15–17, Cardinale et al., 2023), but have not been reported in studies focusing on preadolescence (Karalunas et al., 2021; McKay et al., 2024) or employing wide age bands stretching from preadolescence through adolescence (Cardinale et al., 2024; Colonna et al., 2022), suggesting that behavioral differences in inhibitory control may be sensitive to developmental stages and demonstrating the necessity to evaluate mechanisms of irritability in narrow, developmental periods.

To further investigate these associations, studies have employed functional neuroimaging techniques, which give a window into the background processes involved in inhibitory control and may provide a more sensitive measure of inhibitory control differences (Liuzzi et al., 2020). Studies have largely focused on irritability-related differences in neural patterns during successful inhibition, i.e., when youths successfully inhibit a prepotent response. Examining successful inhibition provides insight into altered brain processes that are required to inhibit responses and may point to neural compensatory mechanisms in irritability. During successful inhibition, studies have observed increased prefrontal cortex activation in irritable youths (n=89, age range: 15–17, Cardinale et al., 2023; n=118, age range: 4–5, Fishburn et al., 2019), suggesting that more prefrontal resources are necessary in order to successfully inhibit a response. However, other studies have reported decreased amygdala (n=19, age range: 11–15, Liuzzi et al., 2020) or superior temporal and ventral pre/post central gyri (n=320, age: 14, Chaarani et al., 2020) activation during successful inhibitory control, which may reflect the necessary allocation of neural resources to regions outside of emotion and sensory processing for successful inhibition of responses. In contrast, additional research has not observed activation differences during successful inhibitory control (n=96, age range: 8–18, Deveney et al., 2012; n=5,948, age range: 9–10, Lee et al., 2022). While activation differences during successful inhibition are relatively mixed across the literature, differences in prefrontal, amygdala, and temporal activation are further supported by altered gray matter volume, cortical thickness, and surface area in those regions in youth with elevated irritability (Ball et al., 2019; Bertocci et al., 2019; Chaarani et al., 2020; Dennis et al., 2019; Jirsaraie et al., 2019; Mulraney et al., 2017; Sammallahti et al., 2023), which may contribute to lower engagement during inhibitory control processes.

To our knowledge, only three studies have examined irritability and neural activation during failed inhibition, i.e., when youths do not inhibit the prepotent response, which may be especially meaningful in irritability as failed inhibition may be a source of an impaired inability to regulate temper outbursts. While no studies observed neural activation differences related to irritability during failed inhibition (Cardinale et al., 2023; Chaarani et al., 2020; Lee et al., 2022), Lee et al. (2022), a recent study using the Adolescent Brain and Cognitive Development (ABCD) baseline sample, examined the combined association of irritability and attention-deficit/hyperactivity disorder (ADHD) symptoms reported decreased region of interest (ROI) coactivation of the left pars orbitalis, supramarginal gyrus, inferior parietal cortex, and lateral occipital cortex in the low ADHD/high irritability group compared to the high ADHD/high irritability and low ADHD/low irritability groups during failed inhibition. This study suggests that links between irritability may be related to differences in neural networks instead of specific regional activation during failed inhibitory control and that these associations are not driven by ADHD, despite their common co-occurrence. However, as few studies have examined neural patterns during failed inhibition in youth with elevated irritability, more research is greatly needed.

While the discrepancies in findings within failed inhibition may be a result of few studies examining this process, the mixed neuroimaging findings concerning successful inhibition may be due to the use of small sample sizes and/or the wide age ranges across studies. Given that the neural networks involved in inhibitory control are developing across childhood and adolescence (Luna et al., 2004), associations between irritability and inhibitory control may also vary across development. Examining these associations in narrower age bands using large samples is critical in identifying how irritability and inhibitory control are linked at specific developmental periods and provides a foundation for research within the greater developmental context. Further, employing both activation and functional connectivity analyses allows for the comprehensive understanding of how networks implicated in inhibitory control work together. Measuring functional connectivity is essential, as neural regions are intertwined, and disruptions in networks may be a more sensitive indicator of functional impairment than activation or coactivation differences alone. To our knowledge, only one study, using data from only 19 subjects ages 11–15, has examined irritability and functional connectivity between brain regions during an inhibitory control task and observed that higher levels of youth irritability were associated with increased amygdala functional connectivity with parietal and temporal regions, as well as the thalamus during successful inhibitory control, although failed inhibitory control was not examined (Liuzzi et al., 2020). Thus, research is sorely needed to examine associations between youth irritability, independent of co-occurring symptoms, and activation and functional connectivity during successful and failed inhibitory control in a large, diverse sample of youth during a narrow developmental range to characterize inhibitory control deficits in youth irritability.

The current study leveraged the ABCD baseline sample to examine concurrent associations between preadolescent irritability and brain networks supporting inhibitory control to address several gaps in the literature. The multi-site ABCD study follows a large, diverse sample from preadolescence to adolescence, assessing both behavioral and neurological functioning (Barch et al., 2018; Casey et al., 2018). Specifically, the ABCD study’s considerable sample size offers greater statistical power to identify reliable neural correlates between inhibitory control and parent-reported preadolescent irritability. It is advantageous that this sample begins in preadolescence (ages 9–10 years), as this period has been identified as an important inflection point in changes in irritability across development, signifying heightened risk (Yu et al., 2023). Importantly, in addition to ROI analyses, we employed whole brain activation and connectivity analyses that examine multiple neural aspects of inhibitory control and allow for both theory- and data-driven approaches, increasing the scope and depth of insights that can be gained from the data. This is the first large scale study to employ a whole brain approach when examining irritability and inhibitory control. Moreover, we examined irritability as a dimensional construct, capturing a wide spectrum of irritability, to determine associations with inhibitory control across varying levels of irritability. We further examined the specificity of irritability in this context, by adjusting for co-occurring symptoms, such as ADHD, anxiety and depression, to determine the effect of irritability over and above related symptoms similarly linked with aberrant inhibitory control (Cai et al., 2021; Crane et al., 2016).

We anticipated that associations between preadolescent irritability and neural processes involved in inhibitory control would be driven by differences in functional connectivity. Given prior work with an ROI-only approach using the ABCD sample did not observe any activation differences related to ADHD or irritability (Lee et al., 2022), we did not expect significant differences in activation in relation to irritability. In contrast, based on previous research suggesting altered connectivity between the amygdala and wide-spread cortical regions in youths with higher levels of irritability during successful inhibition (Liuzzi et al., 2020) and altered coactivation during failed inhibition (Lee et al., 2022), we hypothesized that preadolescent irritability would be associated with altered connectivity patterns during both successful and failed inhibitory control. No direction was hypothesized for the functional connectivity analyses, as only one small previous study with a wide age-range has examined irritability and functional connectivity during a different inhibitory control task (Liuzzi et al., 2020).

Methods

Participants

Data for this study were from the ABCD Study (N=11,878) baseline sample from Release 4 (DOI: 10.15154/1523041). Exclusion criteria were implemented to ensure complete data and satisfactory imaging and behavioral data quality. For detailed exclusion criteria, see Supplementary Material A. A total of 5,380 participants were used in analyses (age M=9.97 years, SD=0.62). For demographic information, see Table 1.

Table 1.

Demographic Information of the Study Sample

	N	%
Sex (N=5,380)
Female	2,663	49.50%
Male	2,716	50.48%
Intersex Male	1	0.02%
Race + Ethnicity (N=5,380)
White	3,141	58.38%
Black	564	10.48%
Asian	113	2.10%
Indigenous/Other	48	0.89%
Biracial or Multiracial	469	8.72%
Hispanic or Latino/a	1,032	19.18%
Refused/Missing/Don’t Know	13	0.24%
Parental education (N=5,373)
No 4 Year Degree	1,882	35.03%
4 Year degree or Higher	3,914	64.97%
	M	SD
Age (N=5,380)	9.97	0.62
Pubertal status (N=5,358)	1.58	0.46
Parent-reported preadolescent irritability (N=5,380)	0.82	1.22

Note. Preadolescent irritability was measured through the CBCL irritability index. Pubertal development was assessed using the Pubertal Development Scale (PDS).

Measures

Stop Signal Task (SST)

The Stop Signal Task (SST) was used to evaluate brain activity during inhibitory control processes. The SST contained two trial types: Go and Stop, 5/6 and 1/6 of the trials, respectively. In the Go trials, a left or right facing arrow (the go signal) was displayed for 1000ms or until participant response, and the participant was instructed to indicate the direction of the arrow. In the Stop trials, an upward facing arrow (the stop signal) replaced the go signal after the stop signal duration and was displayed for 300ms. This led to four SST trials of interest: correct go, incorrect go, correct stop, incorrect stop. For each of these trials, the stop or go refers to the trial type and the correct or incorrect refers to the participant’s performance on those trials. To enable researchers to estimate the speed of the stop process (the stop signal reaction time; SSRT) and achieve approximately 50% success rate on stop trials, the latency between the initial go stimulus and the stop signal is systematically varied as a function of task performance.

The ABCD implementation of the SST has received some criticism and corresponding recommendations for researchers analyzing the data (Bissett et al., 2021; Garavan et al., 2022). Concerns relate to overall task design and coding errors that impact a small subset of subjects. These design issues and coding errors were addressed by applying a set of trial-level and subject-level exclusion criteria to ensure that data included in analyses reflect adequate implementation of the SST; these are detailed in the Supplementary Material A.

Child Behavioral Checklist Irritability Index

The Child Behavioral Checklist (CBCL; Achenbach & Ruffle, 2000) is a caregiver-report measure of youth emotional and behavioral problems. Symptoms are rated on a 3-point Likert scale on the frequency of symptoms, ranging from “0=Not True” to “2=Very True/Often True.” Prior research has quantified an irritability dimension by summing caregiver ratings on CBCL items relating to irritability (Dougherty et al., 2021; Roberson-Nay et al., 2015). The CBCL irritability score was constructed from three items: item 85--Stubborn, sullen, or irritable; item 86--Sudden changes in mood or feelings, and item 95--Temper tantrums or hot temper. Scores ranged from 0–6, with higher scores indicating more child irritability. The CBCL irritability score has good internal consistency and reliability when compared to other irritability measures (Dougherty et al., 2021). The distribution of irritability scores is presented in Supplementary Material B.

Sensitivity Analysis Variables

Mental Health.

To determine the effect of irritability over and above related symptomatology, specificity analyses accounted for depressive, anxious, and ADHD symptoms as well as obsessive compulsive disorder (OCD) and bipolar disorder diagnoses. Depressive, anxious, and ADHD symptoms were measured through the DSM-5 oriented scales of the CBCL (Achenbach & Ruffle, 2000). Items in these scales did not overlap with the CBCL irritability index detailed above. Lifetime (i.e., past or present) OCD and bipolar diagnoses were assessed with the parent-reported Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS; Ambrosini, 2000).

Economic Adversity.

Specificity analyses sought to evaluate the effects of irritability over and above economic adversity, as it is a potent stressor associated with youth psychopathology. ABCD’s neighborhood deprivation index (reshist_addr1_adi_wsum) was used to measure economic adversity.

Cognitive and Social Functioning.

To examine the effect of irritability on inhibitory control processes over and above cognitive and social domains associated with irritability and inhibitory control, specificity analyses controlled for cognitive and social functioning measures. Cognitive functioning was represented by the Cognitive Total Composite score, a compilation of scores from NIH Toolbox cognitive tasks. Social functioning was measured using the parent-reported Strengths and Difficulties Questionnaire’s Prosocial Behavior Subscale (Goodman, 1997), which assesses a child’s considerateness and helpfulness towards others.

Demographics

Demographic variables were used as covariates in the analyses to assess the effects of irritability on neural activity over and above other, common influences on inhibitory control, including child age, sex, pubertal development, and parental education (Brieant et al., 2021; Moilanen et al., 2010; Schulte et al., 2020). Pubertal development was measured through the Pubertal Development Scale (Petersen et al., 1988). Parent report of pubertal development was used unless missing, in which case youth report was used. SES was indicated by whether the participant had at least one parent with a four-year college degree or higher.

Neuroimaging Acquisition and Preprocessing

Images were acquired at 21 sites through 3T Scanners with 32-channel head coils. See Casey et al., (2018) and Hagler et al., (2019) for more detailed information on ABCD neuroimaging acquisition.

Functional MRI data were downloaded from Release 4 after being initially processed by DCAN using their connectivity preprocessing program (Feczko et al., 2021). DCAN’s preprocessing pipeline includes motion correction, coregistration, demeaning and detrending to time, denoising, and applying a bandpass butterworth filter. For full details on data preprocessing, see Feczko et al., (2021).

After downloading the partially processed data from DCAN from the NIH National Data Archive, additional preprocessing steps were applied to complete preprocessing and implement stricter data quality controls. Variance associated with motion and the first derivative of motion were included in regressions to address potential artifacts associated with excessive head motion. TRs were censored with a framewise displacement of .5mm to exclude large movements, and data were smoothed using a 6mm kernel to maximize signal.

Generalized Psychophysiological Interactions

Generalized psychophysiological interaction (gPPI; McLaren et al., 2012)analyses were conducted to evaluate how functional connectivity between specific seed regions and the rest of the brain depend on task conditions. Voxel-wise images are created for each task condition, indicating the strength of the interaction between the BOLD signal of the seed region and the rest of the brain for that condition, above and beyond the baseline BOLD signal. Eight seed regions (bilateral amygdalae, ventral striatum, anterior cingulate cortices, and inferior frontal gyri) were chosen due to their established relationship with irritability and altered irritability-related connectivity in previous studies (Dougherty et al., 2018; Kryza-Lacombe et al., 2021, 2022)or due to their established pattern of activation in SST studies (Cai et al., 2019, 2021; Deng et al., 2017). Masks for the amygdalae and ventral striatum seeds were extracted from the Talairach atlas and converted to MNI space using 3dWarp in AFNI, and spherical masks for the anterior cingulate cortex and inferior frontal gyrus were created from previous studies coordinates (Cai et al., 2019, 2021; Deng et al., 2017; Kryza-Lacombe et al., 2022) using a 5mm radius.

Data Analysis

Whole Brain Analysis

Whole brain analyses were completed using 3dMVM in AFNI (Version 22.3.05; Cox, 1996)to examine the effect of irritability and its interaction with task on activation and connectivity (for gPPI processing, see above). Irritability was included as a quantitative factor. To overcome computational limits given the large dataset size, analyses were completed for each pair of conditions separately. Three pairs of conditions were chosen to represent specific cognitive processes of interest compared to a baseline level of task participation (i.e., correct go): successful inhibitory control (i.e., correct stop and correct go), failed inhibitory control (incorrect stop and correct go), and failed attentional control (i.e., incorrect go and correct go). The ClustSim function in AFNI with a p-threshold of .001 and alpha of .05 was used to calculate a cluster threshold of 35 voxels. Average activation and connectivity values were extracted from each significant cluster. Significant interactions indicate that irritability affects one condition differently than the other condition in the pair, i.e., interactions between irritability and successful inhibitory control, failed inhibitory control, or failed attentional control. For further details on whole brain analyses, see Supplementary Material C.

Region of Interest Analyses

Eighteen regions of interest (ROI) were identified due to their prior associations with inhibitory control and/or irritability (Cai et al., 2019; Deng et al., 2017; Kryza-Lacombe et al., 2021; Liuzzi et al., 2020). ROIs included the left and right inferior, middle, and superior frontal gyri, dorsolateral prefrontal cortex, anterior cingulate cortex, insula, caudate, ventral striatum, and amygdala. For details on ROI creation, see Supplementary Material D.

The interaction between irritability, condition, and performance on ROI activation and connectivity was evaluated through multilevel linear mixed effects models in R (Version 4.2; R Core Team, 2023)using the lme4 (Bates et al., 2015)and the lmerTest (Kuznetsova et al., 2017)packages, to examine the degree to which ROI activation and connectivity are related to irritability depending on task condition and performance. These models additionally controlled for child age, sex, pubertal development, and parental education, and nested by family and MRI scanner. Participants were excluded from all ROI analyses if they were missing any data for the covariates (n=30).

Participants were also excluded from individual ROI analyses if they had no non-zero voxels in the mask (n=0–86), or their activation or connectivity weights were considered outliers (>3 standard deviations from the mean; n=1–85). To reduce the possibility of Type I error due to multiple comparisons, a Bonferroni correction was implemented across ROI analyses; to account for the 18 ROIs, an alpha threshold was set at .0028.

Specificity Analyses

Analyses were repeated to evaluate the interaction between irritability and inhibitory control on brain activation and connectivity over and above cooccurring symptomatology and potential confounding environmental and individual functioning factors. Symptomatology measures included the CBCL DSM-5 symptom scales for depression, anxiety disorders, and ADHD as well as parent-reported KSADS past or present diagnosis of obsessive compulsive disorder (OCD) and bipolar disorder. Environmental and individual functioning measures included economic adversity, as measured through the Neighborhood Deprivation Index, social functioning, as measured through the parent-reported Prosocial Behavior Scale, and cognitive functioning, as measured by a composite score of the NIH toolbox cognitive tasks.

Specificity models were run for each significant whole brain cluster or ROI by repeating analyses including both the control variable and its interaction with task condition (for whole brain clusters) or its interaction with task condition and performance (for ROIs) as predictors. Each control variable was evaluated individually.

Moderation Analysis

As biological sex differences in irritability symptomatology and brain development are common, additional analyses were performed to evaluate whether biological sex moderated the associations between irritability and neural activation/connectivity during inhibitory control. For each significant result, analyses were repeated with biological sex as a moderator of the interaction between irritability and task condition (for whole brain clusters) or task condition and performance (for ROIs).

Results

Increased irritability was significantly associated with worse accuracy during the SST. For full behavioral results, see Supplementary Material E.

Whole Brain Results

Several clusters exhibited main effects of irritability across pairs of conditions that were not additionally involved in higher-order interactions: higher irritability was associated with decreased connectivity between the amygdala, ventral striatum, and left inferior frontal gyrus seeds and frontal, temporal, parietal, and lateral regions. There were no significant main effects of irritability on whole brain activation. Full results are presented in Supplementary Material F.

Significant interactions of irritability with pairs of conditions (correct stop and correct go, incorrect stop and correct go, and incorrect go and correct go) are described below and presented in Table 2.

Table 2.

Whole Brain Analysis Interaction of Irritability and Task Condition Results

CORRECT STOP AND CORRECT GO
Activation	Region (R/L)	F	df 1	df 2	k	x	y	z
	R Rolandic Operculum	20.83	2	5207	49	−58	−6	12
Connectivity	Region (R/L)	F	df 1	df 2	k	x	y	z
Seed: Right Amygdala
	R Supplementary Motor Area	22.21	2	5339	55	−12	−4	68
	L Superior Medial Gyrus	17.84	2	5339	37	−2	−34	36
	R Inferior Frontal Gyrus	16.03	2	5339	35	−56	−12	2
INCORRECT STOP AND CORRECT GO
Connectivity	Region (R/L)	F	df 1	df 2	k	x	y	z
Seed: Right Ventral
Striatum
	R Supplementary Motor Area	22.79	2	5333	38	−12	−14	48
	R Thalamus	23.00	2	5333	36	−8	6	−2
Seed: Left Inferior
Frontal Gyrus
	L Insula	18.76	2	5344	41	32	−8	−12
INCORRECT GO AND CORRECT GO
Connectivity	Region (R/L)	F	df 1	df 2	k	x	y	z
Seed: Left Amygdala
	L Precentral Gyrus	15.87	2	5339	60	14	12	76
	R Putamen	16.93	2	5339	47	−36	14	−6
	R Superior Temporal Gyrus	17.25	2	5339	38	−64	12	−4

Note. All clusters were significant at the corrected cluster-threshold of p < .0001. One cluster was excluded for being 100% white matter. Coordinates are in MNI space.

Interaction of Irritability and Correct Stop (Successful Inhibitory Control) versus Correct Go Trials

Activation.

There was a significant interaction between irritability and the correct stop and correct go pair of conditions on activation in the right rolandic operculum, such that as irritability increased, activation decreased for correct stop (i.e., successful inhibitory control) but was similar across irritability scores for correct go trials (Figure 1A).

Figure 1.

Whole brain activation and gPPI results A. Whole brain activation in the right rolandic operculum (i, ii) during correct stop and correct go trials depending on irritability.

B. Whole brain gPPI connectivity during correct stop and correct go trials depending on irritability between the right amygdala and the right inferior frontal gyrus (i, ii). The same pattern was observed in the left superior medial gyrus (iii) and the right supplementary motor area.

C. Whole brain gPPI connectivity during incorrect stop and correct go trials depending on irritability between the right ventral striatum and right thalamus (i, ii). The same pattern was observed in the right supplementary motor area. Interaction between irritability and incorrect stop and correct go trials on left inferior frontal gyrus and left insula connectivity (iii, iv).

D. Whole Brain gPPI connectivity during incorrect go and correct go trials depending on irritability between the left amygdala and right putamen (i, ii - left cluster). The same pattern was observed in the right superior temporal gyrus (ii - right cluster) and the left precentral gyrus.

Note. Vertical bars indicate standard error

gPPI: Amygdala.

There was a significant interaction between irritability and the correct stop and correct go pair of conditions on right amygdala connectivity with frontal regions (right supplementary motor area, left superior middle and right inferior frontal gyri): as irritability increased, connectivity decreased during correct stop trials but was similar across irritability scores for correct go trials (Figure 1B). There were no significant whole brain results with remaining gPPI seeds for the successful inhibitory control contrast.

Interaction of Irritability and Incorrect Stop (Failed Inhibitory Control) versus Correct Go Trials

gPPI: Ventral Striatum.

There was a significant interaction between irritability and the incorrect stop and correct go pair of conditions on connectivity between the right ventral striatum and ipsilateral supplementary motor area and thalamus, such that as irritability increased, connectivity decreased during incorrect stop trials but was similar across irritability scores for correct go trials (Figure 1C).

gPPI: Inferior Frontal Gyrus.

There was a significant interaction between irritability and the incorrect stop and correct go pair of conditions on connectivity between the left inferior frontal gyrus and left insula: as irritability increased, connectivity decreased during incorrect stop trials but was similar across irritability scores for correct go trials (Figure 1C).

There were no significant whole brain activation results or whole brain connectivity results with remaining gPPI seeds for the failed inhibitory control contrast.

Interaction of Irritability and Incorrect Go (Failed Attentional Control) versus Correct Go Trials

gPPI: Amygdala.

There was a significant interaction between irritability and the incorrect go and correct go pair of conditions on left amygdala connectivity with contralateral putamen, precentral and superior temporal gyri, such that as irritability increased, connectivity decreased during incorrect go but was similar across irritability scores for correct go trials (Figure 1D). There were no significant whole brain activation results or whole brain connectivity results with remaining gPPI seeds for the failed attentional control contrast.

ROI Results

There were no significant main effects of irritability in any ROI analyses, or any significant interactions between irritability and condition and/or performance on ROI activation.

Interactions between Irritability, Condition, and Performance

gPPI: Amygdala.

There was a significant interaction between irritability, condition, and performance on connectivity between the left amygdala and the left inferior, β=.03, t(21109.60)=3.22, p=.001, and superior, β=−.03, t(14855.85)=−3.49, p<.001, frontal gyri, and ventral striatum, β=.03, t(14719.26)=3.19, p=.001, driven by differences in incorrect go trials. For left amygdala connectivity with the left inferior frontal gyrus and left ventral striatum ROIs, as irritability increased, connectivity decreased during incorrect go trials but was similar across irritability scores for other types of trials (correct go, incorrect stop, correct stop). However, a different pattern was observed for left amygdala connectivity with the left superior frontal gyrus: as irritability increased, connectivity increased during incorrect go trials, but was similar across irritability scores for other types of trials (Figure 2).

Figure 2.

ROI gPPI connectivity during incorrect go and correct go trials depending on irritability. (A) Interaction between irritability, trial type, and performance on left amygdala and left inferior frontal gyrus (B) connectivity. As indicated by the arrow, the same pattern was observed for the left ventral striatum (C).

(D) Interaction between irritability, trial type, and performance on left amygdala and left superior frontal gyrus (E) connectivity.

Note. Images B, C, and E show a priori ROI Masks. Vertical bars indicate standard error

There were no significant ROI results with the remaining gPPI seeds.

Specificity Analyses

Specificity of the Interactions between Irritability and Inhibitory Control Processes

Additional analyses to adjust for cooccurring youth mental health symptoms, including anxiety, depression, and ADHD symptoms as well as OCD and bipolar diagnoses, and their interactions with inhibitory control processes indicated that the interaction results between irritability and neural functioning during inhibitory control were not primarily driven by these factors, as all interaction results remained significant, ps < .022. Further, analyses adjusting for economic adversity as well as social and cognitive functioning also indicated that interaction results between irritability and neural functioning during inhibitory control were not primarily driven by these factors, as the irritability-related interactions remained significant, ps < .009.

Specificity of the Main Effects of Irritability

After adjusting for youth symptomatology (anxiety, depression, ADHD, OCD, and bipolar disorder), economic adversity, and social and cognitive functioning, all main effects of irritability remained significant with one exception: after controlling for ADHD symptoms, the main effect of irritability on right ventral striatum connectivity with the right superior frontal gyrus whole brain cluster during incorrect stop and correct go trials was no longer significant, p=.101.

Moderation Analysis

Biological sex significantly moderated the whole brain interactions between irritability and neural patterns during failed inhibition (i.e., during incorrect stop trials) in one cluster (right ventral striatum connectivity with the supplementary motor area) and failed attentional control (i.e., during incorrect go trials) in four clusters (left amygdala connectivity with the left precentral gyrus, right putamen, and right superior temporal gyrus), such that there were stronger effects in female than male preadolescents. Biological sex moderated the ROI interactions among irritability, condition, and performance in two ROIs (left amygdala connectivity with the left inferior and superior frontal gyri), such that there were stronger effects in male than female participants. For full moderation results, see Supplementary Material G.

Discussion

Irritability is an important transdiagnostic symptom that is party to many disorders, yet little is known about how irritability interacts with altered inhibitory control, a prevalent transdiagnostic cognitive impairment. However, to truly be impactful, understanding how these pieces fit together in order to create detailed models of transdiagnostic constructs is necessary. Moreover, examining these transdiagnostic neural mechanisms at timely inflection points in development such as preadolescence, prior to the well-documented uptick in diagnoses in adolescence, is crucial to harness and translate scientific findings to the clinic. This study observed that higher levels of preadolescent irritability was associated with decreased task accuracy as well as aberrant neural patterns during multiple inhibitory control processes (i.e., successful and failed inhibitory control) and failed attentional control, setting the stage for transdiagnostic impact.

Using the ABCD baseline sample, we observed associations between multiple inhibitory control processes and preadolescent irritability at both the behavioral and neural levels. At the neural level, we examined both activation and connectivity: while there was only one significant association between irritability and inhibitory control in activation, alterations in connectivity during both successful and failed inhibitory control were discovered. First, alterations in activation and connectivity in the context of successful inhibitory control (i.e., decreased right rolandic operculum activation and decreased right amygdala and prefrontal cortex connectivity during correct stop trials) were observed in preadolescents with higher levels of irritability. This corroborates previous research demonstrating reduced prefrontal cortex activation in youth with high levels of irritability during successful inhibition (Cardinale et al., 2023; Fishburn et al., 2019). Further, the altered connectivity pattern is consistent with previous research in a small sample of preadolescents and adolescents observing aberrant amygdala connectivity in relation to successful inhibitory control in irritability (Liuzzi et al., 2020). Though activation and connectivity patterns were present in both studies, Liuzzi et al. (2020) observed aberrant amygdala activation and connectivity with temporal and parietal regions, while this study instead observed alterations in prefrontal activation and amygdala connectivity with the prefrontal cortex. However, in Liuzzi et al (2020), associations were only evaluated with 19 subjects across a wide age range starting at age 11. Thus, it is not unexpected that the patterns observed in this study would differ, as the present study encompasses preadolescent youth ages 9–10 and the larger ABCD sample allows for more statistical power to observe neural activation and connectivity patterns.

Second, we observed altered connectivity in the context of failed inhibitory control (i.e., decreased right ventral striatum and thalamus connectivity, and left inferior gyrus and insula connectivity during incorrect stop trials). Few studies have directly probed failed inhibitory control (i.e., incorrect stop trials in our paradigm), and prior work that has examined this process with irritability has failed to find activation differences and, further, has not investigated functional connectivity (Cardinale et al., 2023; Chaarani et al., 2020; Lee et al., 2022). Nevertheless, altered coactivation patterns have been observed during failed inhibitory control (Lee et al., 2022), and thus, together, our findings suggest that associations between failed inhibitory control and irritability may present as altered network patterns. While the observed coactivation findings in Lee et al., (2022) support the presence of altered network patterns, it is notable their results were observed in different regions than those in the present study. However, this is not unexpected due to differences in analysis methodology: while Lee et al., (2022) evaluated coactivation among cortical regions, we employed whole brain functional connectivity analyses, which is a more direct measure of how neural regions may be working in concert throughout the task and allows for the evaluation of networks involving the amygdala and ventral striatum. Finally, we observed neural alterations in the context of a more general failure of attentional control in relation to irritability (i.e., left amygdala connectivity with striatal and prefrontal cortex regions during incorrect go trials), suggesting aberrations in attentional processes in addition to those in inhibitory control processes. This is a novel finding, as few studies examine incorrect go trials during the stop signal task, and no previous studies have examined this in irritability using functional connectivity. Future work should incorporate incorrect go trials into analyses to gain further insight into irritability and altered attentional processes.

Overall, these findings suggest that irritability is associated with altered neural connectivity across successful and failed inhibitory control, as well as impaired attentional control, demonstrating the broad involvement of inhibitory control processes in irritability. These neural alterations are further supported at the behavioral level. Preadolescents with greater levels of irritability exhibited lower accuracy in the SST, consistent with prior behavioral findings suggesting impaired inhibitory control in youths with higher levels of irritability (Gagne & Hill Goldsmith, 2011; Perhamus & Ostrov, 2023). The amygdala connectivity alterations observed in youth with elevated levels of irritability during incorrect go trials may be a neural manifestation of the difficulty in maintaining attentional control behaviorally exhibited by youth with higher versus lower irritability. Further, though only significant in the full ABCD baseline sample, the stop signal duration, representing the latency to inhibit a prepotent response, was higher in youth with elevated irritability, suggesting that even when able to inhibit a response, youth with elevated irritability may still have greater difficulty in doing so. This difficulty in inhibiting a response could be due to aberrant neural patterns during successful inhibition, such as the decreased prefrontal activation and amygdala-prefrontal connectivity in youth with higher versus lower levels of irritability. Importantly, all interactions persisted above and beyond co-occurring symptoms and diagnoses, i.e., anxiety, depressive, ADHD, OCD, and bipolar disorder, as well as economic adversity and social and cognitive functioning, demonstrating that these results are not primarily due to co-occurrence with related constructs, ps < .012. This is consistent with previous studies observing associations between irritability and inhibitory control when accounting for other symptoms (Cardinale et al., 2023; Chaarani et al., 2020; Lee et al., 2022; Liuzzi et al., 2020). Thus, these neural findings support a role of inhibitory control deficits in mechanistic models of youth irritability and indicate that multiple aspects of inhibitory control, i.e., both successful and failed, are implicated in irritability.

Notably, we observed differences in neural connectivity, but only one finding related to neural activation during inhibitory control and youth irritability. The largely absence of significant associations in activation is in line with previous work using the ABCD sample (Lee et al., 2022). However, other studies have reported activation differences in multiple regions, though these studies were in different developmental periods than the ABCD study (e.g., adolescent samples: Cardinale et al., 2023; Chaarani et al., 2020; early childhood: Fishburn et al., 2019; or across preadolescence and adolescence: Liuzzi et al., 2020). Indeed, developmental changes may account for such variability in results among studies. The mechanisms through which brain regions work together shift across development: as preadolescents transition to adolescents, there is a normative disjoint in the speed in which regions mature (Gabard-Durnam et al., 2014; Gogtay et al., 2004). Indeed, preadolescents with past or present irritability have exhibited reduced gray matter volume in the amygdala and other limbic regions, prefrontal cortex and striatal regions (Ball et al., 2019; Bertocci et al., 2019; Dennis et al., 2019; Jirsaraie et al., 2019; Mulraney et al., 2017; Sammallahti et al., 2023), suggesting differential neural development during preadolescence. The results from this study suggest that altered functional connectivity may be a sensitive early indicator of how regions may be shifting as irritability symptoms develop.

Furthermore, although this probe was designed for inhibitory control, functional connectivity aberrations were present in several regions associated with other neural processes. Regions from emotion processing networks, including the amygdala and insula, as well as from reward processing networks, including the ventral striatum and putamen, exhibited altered connectivity patterns in youth with higher versus lower irritability. These networks have additionally been implicated in youth with elevated irritability during reward (Deveney et al., 2013; Dougherty et al., 2018; Kryza-Lacombe et al., 2021; Perlman et al., 2015)and threat processing tasks (Deveney et al., 2020; Hommer et al., 2014; Kircanski et al., 2018; Salum et al., 2017; Stoddard et al., 2017; Zhang et al., 2021). Moreover, structural differences in inhibitory control, emotion processing, and reward regions, including reduced prefrontal cortex, limbic, insula, and striatal volumes have been observed in youth with elevated irritability (Ball et al, 2019; Chaarani et al., 2020; Dennis et al., 2019; Jirsaraie et al., 2019; Mulraney et al., 2021; Sammallahti et al., 2023), further supporting the involvement of multiple functional networks in youth irritability. Thus, though speculative, the aberrant connectivity among these regions may indicate that altered inhibitory control and reward/emotion processes are intertwined in youth with elevated irritability.

Despite our strengths of employing whole brain activation and connectivity analyses in this large and diverse sample of preadolescents, the current report has several limitations. First, the use of a very large collaborative dataset introduces inherent trade-offs. Specifically, computational limits restricted omnibus tests in the whole brain analyses. Instead, theoretically driven pairs of conditions were selected and conservative corrections for multiple comparisons were applied (Cox et al., 2017). With these pairs of conditions, the substantial dataset enabled us to address key research questions with robust statistical power and employ a whole brain approach with both activation and functional connectivity analyses, allowing for a more comprehensive test of the links between irritability and inhibitory control. Second, our irritability measure, though validated for sensitivity, reliability, and psychometric properties, relies on a limited number of CBCL items, potentially limiting its ability to capture the full spectrum of irritability (Dougherty et al., 2021). Future research should consider more comprehensive youth- and parent-reported irritability measures. Lastly, our study examined concurrent associations only at baseline, providing an initial exploration of the neural association between irritability and inhibitory control; thus, we cannot make assumptions about temporal associations or causality. Nevertheless, we are preparing subsequent longitudinal analyses to probe longitudinal associations between inhibitory control-related brain processes and changes in irritability across the transition to adolescence as the ABCD data collection is ongoing.

In conclusion, in the ABCD preadolescent baseline sample, we observed unique associations between preadolescent irritability and altered neural connectivity during successful and failed inhibitory control and during failed attentional control. This study demonstrates the importance of incorporating aberrant inhibitory control into pathophysiological models of irritability that have largely focused on altered reward and emotion processing. Future work is needed to examine these interconnected neural processes within a developmental context and how they contribute to changes in irritability across development. Though preliminary, this basic science work with neural measures corroborates inhibitory control as a potential cognitive mechanism to be leveraged for interventions for preadolescent irritability, and indeed, many early childhood interventions have successfully promoted cognitive control as a treatment target for irritability related disorders, such as ADHD and depression (Edwards et al., 2022; Meyer et al., 2020). Developing timely interventions for preadolescents is essential to tackle irritability concerns before the transition to adolescence, when youth are at elevated risk for the development of more significant mental health challenges.

Highlights.

• Preadolescent irritability and inhibition-related neural patterns in ABCD sample.

• Used whole brain and region of interest, activation and connectivity analyses.

• As irritability increased, connectivity decreased during inhibition-related trials.

• Effects during both successful and failed inhibition.

• Effects persisted after adjusting for co-occurring symptoms.