Irritability as a transdiagnostic risk factor for functional impairment in autistic and non-autistic toddlers and preschoolers

Abstract

Trait irritability in toddlerhood is a powerful risk factor for later internalizing and externalizing challenges in non-autistic children, but the predictive clinical utility of irritability is unknown in autism. Irritability is a trait-level emotional response (i.e., frustration) to a blocked goal and is one source of disruptive behavior. Irritability has two facets: Frustration is the degree to which emotion is elevated after a blocked goal, while soothability is the rate of recovery from peak distress. We aimed to: (1) compare and describe the two facets of irritability in non-autistic and young autistic children, and (2) assess whether children’s reward sensitivity and executive function moderate the relation between irritability and clinical symptoms. Participants were 90 autistic (n-43) and non-autistic (n=47) 2- and 4-year-olds. Autistic children did not have different levels of frustration but were more difficult to soothe compared to non-autistic children, according to parents. Further, frustration and soothability were less strongly correlated for autistic compared to non-autistic children. For all children, executive function (specifically, inhibition) moderated, or ameliorated the strength of, the relation between irritability (both soothability and frustration) and externalizing challenges. This study provides evidence for irritability as a transdiagnostic risk factor for clinically significant emotion regulation challenges. Further, the effect of trait irritability may be ameliorated by children’s executive function in a transdiagnostic manner. Future work should examine the unique aspects of soothability to how irritability presents within autism, as well as evaluate and modify emotion regulation interventions for autistic toddlers and preschoolers.

Identifying the earliest possible risk factors of later functional impairment can inform preventative intervention efforts. The Research Domain Criteria (RDoC) perspective encourages a focus on risk factors that span traditional diagnostic categories (Beauchaine & Hinshaw, 2020), such as irritability. Irritability is often defined as a stable, trait-level emotional response (i.e., frustration, anger) to a blocked goal (Leibenluft & Stoddard, 2013). Irritability can be reliably identified in the first few years of life and, as a type of emotion regulation, it is a meaningful predictor of later psychopathology, including both internalizing and externalizing challenges (Dougherty et al., 2013; Kessel et al., 2017). The difference between patterns and levels of irritability that are developmentally normative and those that indicate risk for later psychopathology are poorly understood (Wakschlag et al., 2019).

Irritability has two temporal facets: the degree to which an emotion is elevated after a blocked goal, and the rate of recovery from peak distress (e.g., Rothbart et al., 2001; Karalunas et al., 2019; Nigg, 2022). While traditional measures often do not distinguish between these two facets or only index the first facet, temperament questionnaires such as the Early Childhood Behavior Questionnaire (Putnam et al., 2006; Rothbart et al., 2001) have separate scales for each, termed “Frustration” and “Soothability”.

Irritability in Autism

Very little is known about the development of irritability in very young children with autism spectrum disorder (ASD). ASD has a prevalence rate of 1 in 36 and supporting autistic children has a high public and private health cost (Maenner et al., 2023). At least 50% of children with ASD exhibit challenges with emotion regulation, aggression, or other significant behavioral dysregulation, which is double the rate of non-autistic children (Kaat & Lecavalier 2013). Aggressive or unsafe behavior can limit autistic children’s social opportunities and inclusion.

Few studies have examined the extent to which irritability in autism is qualitatively different from irritability in non-autistic children. One study of 42 3- to 10-year-old children found that autistic children had worse soothability, but not frustration, than non-autistic children (Kostantareas & Stewart, 2006), while another study of 111 2- to 8-year-old autistic children found that they had worse frustration, but not soothability, compared to non-autistic norms (Adamek et al., 2011). Irritability may contribute to mental health-related functional impairment for autistic children just as for non-autistic children (Adamek et al., 2011), but few studies have examined this. Autistic toddlers and preschoolers with higher levels of irritability may be less likely to access and benefit from early intervention and community supports. High levels of irritability in autistic children may amplify parent stress, contribute to parent mood challenges, and reduce parenting efficacy.

In addition to trait irritability, many factors may contribute to irritable presentations in autism: lack of functional communication, medical comorbidities, psychosocial stressors, task demands beyond a child’s abilities, and maladaptive reinforcement patterns (McGuire et al., 2016). Because so many other factors may interact with an autistic child’s initial frustration response to impact the calming process, the relation between the frustration and soothability facets of irritability may be less strong for autistic compared to neurotypical children. However, previous studies have not investigated this.

The Role of Emotion Regulation in Mitigating Irritability

Emotion regulation, which involves the dual development of top-down arousal and bottom-up intentional control, is responsible for overall and contextual reductions in irritability as children grow (Wakschlag et al., 2018). Both reward sensitivity (a bottom-up process) and executive function (a top-down process) contribute to emotion regulation (Figure 1). Challenges with the top-down facilitation of emotion regulation, in particular, are responsible for a broad range of diagnostic conditions (Beauchaine & Cicchetti, 2019) and are a compelling process to study from an RDoC or transdiagnostic perspective.

Figure 1.

Top-Down and Bottom-Up Modulation of Emotion Regulation in Early Childhood

Differences in reward sensitivity may moderate the extent to which irritability relates to later psychopathology. Children with high irritability may have high reward sensitivity (Vidal-Ribas et al., 2016). Reward sensitivity can be studied by examining neurological measures of the behavioral activation system (BAS)/ behavioral inhibition system (BIS). EEG frontal alpha asymmetry (FAA) is one way to index reward sensitivity (Schiltz et al., 2018). Further, examining FAA as relative left and right hemispheric activation may index approach and avoidance tendencies independently, which differentially relate to executive function in an emotionally valanced context (Edmunds et al., 2023) and to internalizing/externalizing challenges (Grimshaw & Carmel, 2014). Indeed, theory suggests that observable challenges in rewarding contexts (e.g., impulsivity, motivation) could be due to either overactive approach tendencies or underactive avoidance tendencies, or a combination of both (Beauchaine & Hinshaw, 2020). As such, it is necessary to measure each separately.

Executive function (EF) abilities mature rapidly throughout the toddler and preschool years and may be less developed in autistic children compared to non-autistic children (McEvoy et al., 1993; Pellicano et al., 2006; Faja & Dawson, 2013; Edmunds et al., 2021). EF is comprised of many skills, including “cold” EF like working memory, shifting, and inhibition, and “hot” EF like delay of gratification, which is an EF skill that occurs in the context of reward. Separating EF skills is particularly helpful to understand their developmental onset and interaction with emotional functioning in the early years (Carlson, 2005). The ability to inhibit a dominant response and to delay gratification, in particular, are aspects of EF that differ in autism (Cremone-Caira et al., 2021; Faja & Dawson, 2013) and may temper children’s expression of irritability and broader functional outcomes (Wakschlag et al., 2018; Faja & Dawson, 2013). For school-aged children with ADHD, an “irritable” subgroup characterized by a combination of high frustration and low soothability, had more challenges inhibiting unhelpful actions than other subgroups (Karalunas et al., 2019). Inhibition was found to be a strong predictor of irritable behavior in a sample of similarly-aged autistic children (Ademak et al., 2011). Relations between irritability and inhibition have yet to be examined in younger autistic children. The ability to delay gratification (i.e., a “hot” EF) has been specifically linked to reduced expression of irritable emotional response in autistic children (Mills et al., 2022). Theoretically, inhibition may be the most helpful type of “cold” EF for modulating children’s irritable responses, while examining a “hot” EF skill like delay of gratification may be more ecologically valid, as so much of children’s learning and early experiences occur in the context of reward.

Children’s “bottom-up” reward sensitivity and their “top-down” executive function may both interact to affect the way that young children’s irritability evolves (or does not evolve) into externalizing and internalizing challenges (Figure 2). For example, during a Go/No-Go conflict monitoring task, non-autistic irritable children with an enhanced error related negativity (ERN), an event related potential (ERP) computable from EEG, had higher levels of later internalizing symptomatology (Grabell et al., 2017). In addition, activation of the dorsolateral prefrontal cortex (PFC), an area implicated in both reward processing and executive function, predicted irritability in preschoolers during a delay of gratification task (Grabell et al., 2018). Up to 15-20% of autistic children may have a mood disorder by preschool or elementary school (Salazar et al., 2015). For school-aged autistic children, approach-based reward sensitivity as indexed by left FAA predicted “hot” executive function during a reward processing task, but not later emotion dysregulation (Edmunds et al., 2023). Given these findings, younger autistic children with high trait irritability may benefit from consideration of both their reward sensitivity and executive function to help them interact successfully with their surroundings.

Figure 2.

A Two-Facet Model of Irritability

Understanding individual differences in irritability as a function of reward sensitivity and executive function would enable more precise characterization of the types of emotion regulation and behavioral challenges which autistic children experience as well as facilitate more targeted intervention recommendations.

The Current Study

Very little is known about the levels of and relation between the facets of irritability, frustration and soothability, in autistic toddlers and preschoolers. Early neurophysiological correlates of irritability, which may elucidate the mechanisms by which irritability contributes to functional impairment, are also unknown. Exploring their relation to executive function and internalizing/externalizing challenges in an integrative manner may provide insight into unique aspects of emotion regulation in autism as well as potential treatment targets. The goals of this study were to:

Aim 1: Examine the relation between the two facets of irritability in non-autistic and autistic toddlers and preschoolers

We hypothesized that toddler- and preschool-aged autistic children will have higher levels of both facets of irritability, frustration, and soothability, consistent with elevated levels of disruptive behavior compared to neurotypical young children.

Aim 2: Examine the concurrent predictors of irritability in non-autistic and autistic toddlers and preschoolers

We hypothesized that, consistent with previous literature, irritability would be positively concurrently related to internalizing and externalizing behaviors, as well as reward sensitivity. We hypothesized that irritability would be negatively related to executive function, such that young children with higher levels of frustration would have more difficulty with executive function (specifically, with inhibiting their prepotent response and delaying gratification).

Aim 3: Assess whether children’s reward sensitivity and executive function moderate the relation between irritability and mental health challenges.

We hypothesized that lower reward sensitivity and more developed executive function would buffer the strength of the relation between irritability and later internalizing and externalizing symptoms.

Methods

Participants

Participants were 90 autistic (n = 43) and non-autistic (n = 47) 2- and 4-year-olds (Table 1). Participants were recruited into an ongoing longitudinal study of executive function (EF) development in early childhood; secondary data from one visit was used for this study. Families were recruited from a hospital database of parents interested in participating in developmental research and from the community via advertising (fliers and websites) and community groups (parent support groups, infant play groups). There were more male children and more Latine children in the autistic group compared to the neurotypical group, X²(1, N = 90) = 4.60, p = .03 and X² (1, N = 90) = 4.82, p = .03, respectively. There were no group differences by age, race, parental education, or parental income (Table 1).

Table 1.

Participant Characteristics

	All	Non-Autistic	Autistic
		N (%) or M (SD)
Number of Participants	90	47	43
Gender
Male	65 (71)	29 (62)	36 (84)
Female	25 (27)	18 (38)	7 (16)
Race
American Indian/Alaskan Native	0 (0)	0 (0)	0 (0)
Asian	3 (3)	2 (4)	1 (3)
Hawaiian/Pacific Islander	0 (0)	0 (0)	0 (0)
Black/African American	3 (3)	2 (4)	1 (3)
White/Caucasian	58 (64)	36 (71)	22 (69)
More than one Race	15 (17)	10 (20)	5 (16)
Unknown	10 (10)	1 (2)	0 (0)
Other	3 (3)	0 (0)	3 (5)
Ethnicity
Non-Hispanic/Latine	74 (81)	48 (94)	26 (81)
Hispanic/Latine	8 (9)	2 (4)	6 (19)
Parental Education Level
Some High School (No Diploma)	0 (0)	0 (0)	0 (0)
GED Diploma	0 (0)	0 (0)	0 (0)
High School Graduate	1 (1)	0 (0)	1 (1)
Associate Degree (2yr Degree)	1 (1)	0 (0)	1 (1)
Some College	3 (3)	0 (0)	3 (9)
BA/BS Degree (4yr Degree)	23 (25)	12 (24)	8 (25)
Graduate/Professional Degree)	38 (42)	23 (45)	13 (41)
Unknown/Missing	24 (26)	16 (31)	6 (19)
Age (Months)
	44.93 (11.52)	43.62 (11.39)	46.36 (11.62)
Average Household Income
< $20,000	1 (1)	1 (1)	1 (1)
$20,000 - $40,000	1 (1)	0 (0)	0 (0)
$40,000 - $60,000	1 (1)	0 (0)	0 (0)
$60,000 - $80,000	5 (6)	1 (1)	4 (13)
$80,000 - $100,000	9 (10)	4 (8)	4 (13)
$100,000 - $120,000	4 (4)	3 (6)	1 (3)
$120,000 - $140,000	3 (3)	0 (0)	3 (9)
$140,000 - $180,000	12 (13)	8 (16)	4 (13)
$180,000 - $200,000	6 (7)	3 (6)	2 (6)
> $200,000	12 (13)	7 (14)	4 (13)
Unknown/Missing	35 (39)	24 (47)	9 (28)

All participants had an age equivalent of 12 months or above as assessed by the Visual Reception domain of the Mullen Scales of Early Learning (MSEL; Mullen, 1995). Participants in the autism group entered with existing or suspected autism diagnoses. Diagnostic status for both groups were confirmed using the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) (Lord et al., 2012). Children with autism or autism concerns completed the Autism Diagnostic Interview-Revised (ADI-R) (Kim & Lord, 2012; Rutter et al., 2003). DSM-5 criteria were applied based on expert clinical judgment.

Procedures

All procedures were approved by the Institutional Review Board at Boston Children’s Hospital (Boston, USA). Caregiver consent was obtained before data collection. For the time point from which these concurrent data were taken, caregivers and their children completed three lab visits within 6 months of the first lab visit. The first visit included an ASD diagnostic evaluation and cognitive assessment using the MSEL. Children completed executive function inhibition (tricky box), and delay (gift delay) tasks, and a 4-minute eyes open resting EEG recording across the second and third visits. Parents were then invited to complete the CBQ or ECBQ and BASC-P questionnaires.

Measures

Irritability

The two facets of irritability were measured using the Frustration and Soothability subscales of the Early Childhood Behavior Questionnaire (EBCQ) and Childhood Behavior Questionnaire (CBQ). The EBCQ and CBQ are parent-report measures of temperament for 18- to 36-month-olds and 3- to 7-year-olds, respectively. The Frustration subscale indexes negative affect during interruption or goal blocking. The Soothability subscale assesses the rate of return to baseline from distress. Items are rated by parents on a 7-point Likert scale, with higher scores indicating greater frustration and greater soothability, respectively.

The Frustration subscale from the EBCQ and the Anger/Frustration subscale from the CBQ were combined to create a Frustration metric that spanned the age range of the sample. Similarly, the Soothability subscale from the EBCQ and the Falling Reactivity & Soothability subscale from the CBQ were combined to create a Soothability metric that spanned the age range of the sample.

Reward Sensitivity

Reward Sensitivity was operationalized as the relative differences in resting left (vs. right) and right (vs. Left) frontal alpha (8-13 Hz) resting EEG power.

Resting EEG Data Collection

Participants completed 4 minutes of eyes open resting EEG data collection. The resting paradigm included a set of silent screen-saver videos with non-social content which played twice. Participants sat either on a chair or their parent’s lap, and one research assistant sat in the room with the participant to provide behavioral support. Neural responses were continuously recorded via a Net Amps 400 (Electrical Geodesics, Inc.) using the 128-channel HydroCel Geodesic sensor net 2.0. Impedances were below 50 kΩ prior to each recording. Data were referenced to the vertex and sampled at 500 Hz using a 4 KHz antialiasing hardware filter.

Resting EEG Data Pre-processing & Analysis

The Batch EEG Automated Processing Platform was used to pre-process and conduct power analyses on the resting state data (Levin et al., 2018, Gabard-Durnam et al., 2018). Pre-processing methods were similar to those implemented in Edmunds et al. 2023, with the following differences noted below. For artifact detection, the following channels, in addition to the 10–20 electrodes, were used for ICA with MARA: 32, 27, 23, 34, 28, 19, 16, 29, 41, 46, 44, 47, 51, 57, 64, 60, 67, 72, 37, 30, 13, 6, 3, 4, 117, 112, 111, 105, 1, 123, 116, 87, 103, 114, 102, 98, 97, 85, 77, 100, 95, 75. Electrodes were spread evenly across the scalp, and the number of electrodes was chosen relative to our recording length to maximize ICA performance (Särelä and Vigário, 2003). Data were then segmented into 1 second windows. Segments were inspected again for artifact and segments with an amplitude greater than 40 μV were not included in the final data analysis. A 40 μV cutoff reflects the smaller amplitude that results from the wavelet thresholding and ICA steps during artifact detection (Gabard-Durnam et al., 2018). Using a fast Fourier transform with a Hanning Window, we computed the power spectrum for each segment for all electrodes. For each electrode, power across all segments was averaged for the alpha frequency band (8-13hz). Power within the alpha frequency band (8-13 Hz) from electrodes indexing left and right frontal activation (F3 and F4) was extracted and natural log-transformed.

Reward Sensitivity Variable Derivation

Reward sensitivity was operationalized as relative left and right frontal alpha activation. Specifically, asymmetry in children’s FAA was calculated using residual values (Coan & Allen, 2003; 2004; Reznik & Allen, 2018) as residual values are more sensitive than traditional difference scores (Meyer et al., 2017). Residual values were computed for each hemisphere to capture unique frontal hemispheric activation. Specifically, left residual power values were residual values from a model of right power predicting left power (i.e., left power, controlling for right), and right residual power values were the opposite. Because alpha power reflects the inverse of activation, the inverse of each value was taken.

Executive Function

Executive function was assessed using a standard battery of tasks adapted for 2- to 5-year-old autistic children (Carlson, 2005; Garen et al., 2014; Faja & Dawson, 2013) and used successfully in this sample. Two aspects of executive function were measured for the current study: inhibition and delay.

Inhibition was indexed by the Preschool Executive Function Battery (PEFB) Tricky box (Garon et al., 2014), which involves 2 boxes – one box with handles on each side that open transparent doors on the same sides as the corresponding handles and another box with handles that open doors on the opposite sides. A toy was placed behind one door, and children were tasked with retrieving the toy using the correct handle. Correct responses, where children immediately chose the correct handle, and spontaneous self-corrections, where children initially reached for the incorrect handle but ultimately chose the correct handle, were recorded. Inhibition performance was operationalized as the number of correct responses and spontaneous self-corrections on contralateral trials.

Delay was indexed by the gift delay task (bow or wrap for 2- and 4-year-olds, respectively). During gift delay, children were told to wait while the experimenter noisily wrapped a prize. Peeking was the dependent variable, measured ordinally to capture children who do not peek, who peek over their shoulder, and who turn their whole body around to peek.

Internalizing and Externalizing Challenges

Internalizing and externalizing challenges were operationalized as the Internalizing Problems and Externalizing Problems Composite T-scores, respectively, of the Behavior Assessment System for Children-Parent Report Scale, Third Edition, Preschool Version (BASC-P; Kamphaus & Reynolds, 2015). The BASC-P is a standardized Likert rating scale of behavioral and emotional functioning, designed for children 2 to 5 years of age.

Analytic Plan

Multiple linear regression models were used for all models across all three aims. Nonverbal developmental quotient (DQ, calculated using MSEL Visual Reception subscale) differed by group and was controlled for in all models. For Aim 3 (moderation by inhibition and reward sensitivity of the relation between irritability and mental health challenges), all predictors were centered prior to the calculation of interaction terms. All variables of interest were normally distributed (i.e., were within skewness < .8 and kurtosis < 3). Please see Supplemental Table 1 for a bivariate correlation matrix between variables of interest. Examination of correlation matrix as well as diagnostics on primary regression analyses do not indicate concern about multicollinearity. Given regression models with two to four predictors planned for each aim, the sample size of 90 from the primary dataset was powered to detect medium to large effects. Standardized estimates of interest from each regression model are reported along with the associated p value.

Results

Table 2 reports diagnostic group differences in children’s irritability, executive function, and mental health challenges. Compared to neurotypical children, autistic children had greater delays in nonverbal cognition as well as greater challenges with soothability and delay of gratification. There were no significant group differences in levels of frustration, inhibition, frontal alpha asymmetry, or internalizing and externalizing challenges.

Table 2.

Descriptive Statistics

	All		Non-Autistic		Autistic
	Range	M (SD)	Range	M (SD)	Range	M (SD)
Nonverbal DQ ***	32.74–138.00	87.08 (22.42)	32.75–138.00	100.68 (19.62)	34.91–116.36	79.74 (18.86)
Frustration	0.67–6.62	4.24 (1.19)	1.83–6.62	4.03 (1.27)	0.67–6.00	4.45 (1.06)
Soothability ***	2.43–6.33	4.88 (1.00)	2.86–6.33	5.32 (0.75)	2.78–6.33	4.50 (0.97)
Right FAA	−1.07–0.75	0.00 (0.40)	−1.07–0.75	0.01 (0.42)	−0.91–0.57	−0.01 (0.35)
Left FAA	−0.88–0.81	0.00 (0.34)	−0.88–0.76	−0.01 (0.34)	−0.82–0.81	0.03 (0.36)
Delay ***	0–2	0.99 (.90)	0–2	1.32 (.86)	0–2	0.45 (.67)
Inhibition	0–1	0.39 (.34)	0–1	0.47 (.34)	0–1	0.31 (.30)
Externalizing Problems	37–81	53.33 (10.00)	37–81	52.04 (9.36)	40–80	54.69 (10.72)
Internalizing Problems	33–77	52.79 (9.89)	40–75	53.69 (9.87)	33–77	51.11 (9.93)

Notes: ***Group difference at p<.001, (controlling for nonverbal cognitive ability for all non-nonverbal DQ comparisons). DQ = developmental quotient. FAA = frontal alpha activation

Aim 1: Examine the relation between the two facets of irritability in non-autistic and autistic toddlers and preschoolers

All models control for nonverbal cognition. Frustration did not differ by diagnostic group, β = .16, p = .18 (Table 2). Autistic children were more difficult to soothe than were non-autistic children, β = −.44, p < .001 (Figure 3). An exploratory post-hoc analysis revealed that, across the entire sample, soothability was correlated with autism characteristics measured in a continuous manner (ADOS-2 calibrated severity score), β = −.31, p = .01

Figure 3.

While Non-Autistic and Autistic Children Had Similar Levels of Frustration, Autistic Children Were More Difficult to Soothe Than Non-Autistic Children

Note: *** p<.001, controlling for nonverbal cognitive ability.

The relation between frustration and soothability was moderated by diagnostic group, R² change = .10, p < .001, such that frustration and soothability were more strongly correlated for non-autistic children, interaction β = 1.22, p < .001. Specifically, frustration and soothability were strongly correlated for non-autistic children such that those who had a greater frustration response were more difficult to soothe, r(47) = −.73, p < .001. Frustration and soothability were non-significantly correlated for autistic children, r(43) = −.30, p = .06 (Figure 4).

Figure 4.

Frustration and Soothability Are Less Related for Autistic Children than They Are for Non-Autistic Children

Aim 2: Examine the concurrent predictors of irritability in non-autistic and autistic children

As facets of the same construct, frustration and soothability were added in the same step to all models to assess their relation to reward sensitivity, executive function (inhibition, delay), and mental health (internalizing and externalizing) challenges. Neither frustration nor soothability were significant predictors of right or left FAA, −.07 < βs < .01, .63 < ps < .97. In an exploratory post-hoc analysis for autistic children only, left FAA was correlated with inhibition, r = .44 p = .02.

Frustration was a value-added predictor of inhibition (i.e., controlling for soothability), β = .51, p < .001, but soothability was not, β = .13, p = .28. There were no differences by diagnostic group in the degree to which frustration, soothability, and inhibition were related, β = −.14, p = .25. In contrast, for delay, neither frustration nor soothability were value-added predictors, β = .19, p = .19, and β = .10, p = .50.

Soothability was a value-added predictor of both externalizing problems, β = −.28, p = .03, and internalizing problems, β = −.34, p = .01. In contrast, frustration was not significantly related to externalizing problems, β = .14, p = .25, or internalizing problems, β = .23, p = .06.

Aim 3: Assess whether children’s reward sensitivity and executive function moderate the relation between irritability and mental health challenges.

Reward Sensitivity

Reward sensitivity did not significantly moderate the relation between irritability (frustration or soothability) or mental health challenges (internalizing or externalizing problems), .64 < ps < .80. These relations did not differ by diagnostic group, ps > .05.

Executive Function

For all children across diagnostic groups, inhibition skills significantly moderated, or affected the strength of, the relation between soothability and externalizing problems, controlling for frustration, interaction β = .28, p = .007. For children with better inhibition skills, the relation between challenges with soothability and externalizing problems was mitigated, compared to children with more challenges with inhibition (Table 3, Figure 5).

Table 3.

Inhibition Skills Mitigate the Relation between Soothability and Externalizing Challenges
	B	S.E.	β	t (p)
Nonverbal DQ	.13	0.06	.25	2.21 (.03)
Frustration	2.76	1.08	.34	2.57 (.01)
Soothability	−2.74	1.24	−.27	−2.20 (.03)
Inhibition	−10.91	3.61	−.36	−3.02 (.004)
Inhibition X Soothability	8.84	3.16	.28	2.80 (.007)
Dependent variable: Externalizing Problems T-Score (BASC-P)
Inhibition Skills Mitigate the Relation between Frustration and Externalizing Challenges
	B	S.E.	β	t (p)
Nonverbal DQ	.15	0.06	.30	2.51 (.01)
Frustration	2.57	1.08	.32	2.38 (.02)
Soothability	−3.15	1.27	−.31	−2.48 (.02)
Inhibition	−9.28	3.65	−.31	−2.54 (.01)
Inhibition X Frustration	−7.54	3.04	−.27	2.48 (.015)
Dependent variable: Externalizing Problems T-Score (BASC-P)

Figure 5.

For children with better inhibition skills, soothability is less related to externalizing challenges

Notes: Panel showing the relation between soothability and externalizing problems at inhibition = −2 SDs has been removed because this sample has no participants with inhibition data within that range. Figure created with InterActive R shinyapp (McCabe et al., 2018).

Similarly, for all children, inhibition skills also significantly moderated the relation between frustration and externalizing problems, controlling for soothability, interaction β = −.27, p = .02. For children with better inhibition skills, the positive correlation between frustration and externalizing challenges was ameliorated (Table 3, Figure 6).

Figure 6.

For children with better inhibition skills, frustration is less related to externalizing challenges

Notes: Panel showing the relation between soothability and externalizing problems at inhibition = −2 SDs has been removed because this sample has no participants with inhibition data within that range. Figure created with InterActive R shinyapp (McCabe et al., 2018).

Inhibition did not significantly moderate the relation between soothability and internalizing problems, β = .04, p = .72, or the relation between frustration and internalizing problems, β = −.04, p = −.76.

Children’s delay of gratification did not significantly moderate the relation between either facet of irritability (frustration and soothability) and either aspect of mental health challenges (internalizing and externalizing problems), ps > .05.

Discussion

This study examined group differences, internal coherence, and emotion regulation-related correlates of two facets of irritability—frustration and soothability—in 90 autistic and non-autistic 2- to 4-year-olds. Autistic children did not have different levels of frustration but were more difficult to soothe compared to non-autistic children, according to parents. Further, frustration and soothability were less strongly correlated for autistic compared to non-autistic young children. For all children, inhibition skills moderated, or ameliorated the strength of, the relation between irritability (both soothability and frustration) and externalizing challenges. However, children’s delay (a top-down “hot” executive function skill) and bottom-up reward sensitivity did not moderate the relation between irritability and mental health challenges.

Irritability: Especially Multifaceted in Autism

Our findings suggest that levels of soothability, more than frustration, are impacted for autistic children, which is in line with some prior work (e.g., Konstantareas & Stewart, 2006). However, previous findings are mixed. Another study found that 2- to 8-year-old autistic children had greater frustration, but no difference in soothability, although they did have greater variability in soothability scores (Adamek et al., 2011). While this study used the same temperament-based measure of frustration and soothability as the current study, Adamek and colleagues compared autistic children to compared to published norms rather than a newly recruited neurotypical control group. In addition, neither study controlled for cognitive ability; both methodological differences may have contributed to these mixed findings.

Given that frustration and soothability are less related in autistic toddlers and preschoolers, these facets of irritability may have different or additional etiologies for autistic compared to non-autistic children. For example, challenges with soothability may be more related to communication difficulties or core autism characteristics (e.g., sensory sensitivities or need for sameness) rather than trait irritability for autistic young children. The group difference in soothability likely has clinical implications for autistic toddlers and preschoolers; other predictors of soothability in autism should be explored, such as sensory, communication, or autonomic nervous system differences (e.g., Hogan et al., 2021). Future research should examine this. Understanding the transdiagnostic individual differences in irritability helps us better understand the heterogeneity in autism specifically. While soothability may predict outcomes transdiagnostically, future work exploring supports for the challenges with soothability that are specific to autistic individuals may protect against negative mental health outcomes.

Soothability may function more strongly as a transdiagnostic predictor of later mental health challenges than as an autism-specific difference. Although we found that autistic toddlers and preschoolers had more challenges with soothability than those who are non-autistic, it should be noted that greater difficulties with soothability was also related to greater autism characteristics in a continuous manner across groups, suggesting that, whatever the antecedents of difficulties with soothing, that soothability challenges are present across many young children regardless of diagnosis. This is supported by prior literature. In a sample of young children with Fragile X syndrome, soothability was found to predict later anxiety but not to distinguish diagnostic groups (Tonnsen et al., 2013). More importantly, the relation between soothability and externalizing challenges, and its moderation by children’s inhibition skills, was found across diagnostic groups. This suggests that despite differences in amounts of difficulty with soothing, these difficulties relate to broader impairment in a similar manner for autistic and non-autistic children.

Bottom-Up Pathway for Emotion Regulation: Reward Sensitivity

We found no correlations between reward sensitivity and other factors—irritability, inhibition, or mental health challenges—in our sample of autistic and non-autistic toddlers and preschoolers. For school-age autistic children, reward sensitivity, indexed with the same metric, was correlated with executive function in an emotionally valanced context, but not emotion-regulation (Edmunds et al., 2023). The strongest findings examining left and right frontal alpha activation (FAA) have found relations between left FAA and ADHD characteristics in children and adults, and right FAA and depressive symptoms in adults. In an exploratory post-hoc analysis for autistic children only, we found that left FAA was correlated with inhibition, but not delay. FAA might relate to inhibition more meaningfully (or mechanistically) for autistic compared to non-autistic participants. Alternatively, FAA may be potentially useful for indexing skills that inherently recruit both executive function and emotion regulation processes, in line with Carmel and Grimshaw’s asymmetric inhibition model (2014). More work is needed to determine the specific neurobiological processes indexed by FAA. While EEG is ideal for autistic toddlers and preschoolers who are active and distractible, work using fMRI, which can better determine cortical versus limbic activation and better localize prefrontal cortical activation, may be essential in understanding how executive function affects emotion regulation as children grow, both within autism and transdiagnostically.

Top-Down Pathway for Emotion Regulation: Executive Function

Regarding top-down emotion regulation processes involving executive function, frustration and soothability seem to behave similarly across autistic and non-autistic children. Across our sample, the relations between both facets of irritability and externalizing challenges were ameliorated by inhibition skills. The mitigating effect of children’s soothability on externalizing challenges is less potent for children who are most skilled at inhibition. However, inhibition also ameliorates the degree to which trait frustration contributes to externalizing challenges. In other words, for preschoolers and toddlers with higher trait frustration responses, better inhibition skills may keep their frustration responses from escalating into clinically significant levels of externalizing challenges. This relation appears to be transdiagnostic, in that trait frustration did not differ based on whether these children were autistic or non-autistic and that, for all children, frustration related to externalizing challenges but was mitigated by inhibition skills. Despite potentially different levels of executive function skills across diagnostic categories (e.g., autism, ADHD, depression, etc.), more work should examine whether executive function skills may exert the same protective effect on early mental health processes transdiagnostically.

Internalizing and Externalizing Challenges in Autism

While autistic children may show greater transdiagnostic irritability from toddlerhood to school age (Jahromi et al., 2012; Edmunds et al., 2023), it appears that prior studies are mixed with regard to whether autistic children have challenges with clinically significant internalizing or externalizing symptoms. We found no significant diagnostic group differences for parent-reported internalizing and externalizing challenges. Previous work found that 2- to 6-year-old autistic boys were reported to have greater externalizing, but not internalizing challenges than neurotypical boys (Li et al., 2019). In adolescence, autistic individuals had higher internalizing challenges than non-autistic teens, but externalizing challenges were not assessed (Solomon et al., 2012).

Limitations

This study had several limitations. First, all variables of interest were measured concurrently, while the emotion regulation model assessed is longitudinal in nature. Future work should examine whether temperament-based irritability, measured earliest in development, predicts later externalizing and internalizing challenges, as moderated by reward sensitivity and executive function measured at a midpoint. Second, there are measurement limitations. Irritability was measured using a parent report questionnaire. Future work should examine irritability measured using an interactive, standardized measure that better accounts for the developmentally appropriate levels of irritability in toddlerhood in a developmental manner (e.g., Wakschlag et al., 2018). Reward sensitivity was measured using only a physiological metric. The delay variable in this study, while behavioral/task-based as opposed to physiological, indexed a mixture of EF and reward sensitivity. Delay tasks are included under the Positive Valence System for NIMH's Research Domain Criteria matrix, and this may have contributed to the non-significant findings for the relation between children’s ability to delay gratification, irritability, and mental health challenges. Future work should attempt to measure each construct of interest using multiple methods. Finally, the sample was predominantly White, non-Hispanic/Latine, highly educated, and high income. Research that continues to report on families with greater relative privilege ignores the ways that structural inequities may contribute to the development of emotion regulation for minoritized autistic children and their families. Externalizing behavior is often misdiagnosed and over-pathologized in children of color, especially Black children (Fadus et al., 2020). While inequities are intersectional in nature for most families, racism may have an outsized effect on both the development and societal interpretation of behavioral dysregulation.

Implications and Conclusions

These findings align with developmental psychopathology theory (Beauchaine & Cicchetti, 2019). Even though we examine brain-to-behavior processes across a specific clinical group and a comparison group, it provides a useful opportunity to demonstrate that the same processes (irritability, inhibition) may affect the development of mental health challenges in both populations.

Neurobiological theories of irritability suggest that irritability may be due to challenges either with context-sensitive regulation (i.e., reacting differentially to environmental cues), or with difficulties with attention modulation (Leibenluft & Stoddard, 2013). Our finding that inhibition skills modulate the degree to which children’s irritable temperament becomes impairing may provide some evidence for the latter theory, but the most interesting aspect of our results from a transdiagnostic perspective is that this same process is at play for all children regardless of autism diagnostic status. Although it is best to start by exploring dimensional differences rather than categorical disorders (Beauchaine & Hinshaw, 2020), autistic individuals have additional differences in social and environmental attention beyond those theorized for children with high irritable temperament. A developmental psychopathology framework would support future research that explores the interaction between autistic brain-based differences and other transdiagnostic differences such as irritability.

Clinicians should consider adding interventions to help families and autistic children with emotion regulation, as these interventions may make an outsized impact on the well-being of autistic children and families. Although treatment of overt aggression in older children with ASD could include psychotropic medication or intensive applied behavior analysis (ABA), treatment options for children younger than 5 years are limited (McGuire et al., 2016). One autism-specialized parent training protocol for challenging behavior has demonstrated efficacy for children aged 3-7 years (Bearss et al., 2015) but should be interpreted with caution given the use of the antecedent, behavior, and consequence (ABC), which may both under- and overestimate different aspects of irritability in children under 5 years (Schmidt et al., 2014).

Further, autistic children with higher levels of irritability may be less likely to access and benefit from early intervention for core autism-related challenges such as social communication. Although children who receive Naturalistic Developmental Behavioral Interventions (NDBIs) may display fewer in-session dysregulated behaviors than those who receive discrete trial ABA, (Mohammadzaheri et al., 2015), NDBIs do not explicitly target behaviors related to negative emotion (Schreibman et al., 2015). Interventions for toddlers with ASD are often parent-mediated; high child irritability may amplify parents’ stress and reduce parenting efficacy (Burrell et al., 2020). Thus, child irritability may affect parents’ ability to deliver intervention strategies effectively, parents’ motivation to engage in intervention, or interventionists’ ability to work directly with children. Exploring how irritability may exacerbate risk for later internalizing and externalizing challenges as moderated by executive function skills may provide unique insight into early supports for autistic toddlers and preschoolers.