Parenting and Peer Victimization in the Development of Callous-Unemotional Behaviors: Moderation by Irritability and Basal Cortisol

Irritability, operationalized as a propensity to experience anger and frustration (Brotman et al., 2017; Leibenluft et al., 2024), consists of both a persistent mood component (i.e., tonic irritability) and emotional and behavioral temper outbursts (i.e., phasic irritability) and is a facet of negative emotionality (Beauchaine & Tackett, 2020; Leibenluft et al., 2024). Although often measured as a dimension of oppositional defiant disorder (ODD), irritability is also a feature of normative temperament variation, a symptom in over 20 DSM-5 diagnoses, and predictive of high levels of both concurrent comorbidity and heterotypic continuity across the spectra of internalizing and externalizing problems (Beauchaine & Tackett, 2020; Klein et al., 2021; Leibenluft et al., 2024). Therefore, within this study, and consistent with the ESCAPE-AL model (Kimonis, 2023), we focus on irritability as a transdiagnostic factor related to emotional sensitivity, rather than specifically as a dimension of ODD.

Some theoretical models posit that irritability and callousness are characteristic of two distinct phenotypes of disruptive behaviors, including in early childhood (Frick & Morris, 2004; Wakschlag et al., 2018). However, recent work has highlighted the developmental overlap among irritability and CU behaviors. For instance, among preschoolers oversampled for disruptive behavior problems, the angry/irritable mood domain of ODD (and, independently, the argumentative/defiant domain of ODD) was closely associated with the callous domain of CU behaviors (Bansal et al., 2021). Further, high levels of co-occurring irritability and CU behaviors were related to especially high levels of impairment across numerous socioemotional domains in a clinical sample of elementary-aged children (Waschbusch et al., 2020) and (with co-occurring headstrong/defiant symptoms) a community sample from Spain (Ezpeleta et al., 2022).

The ESCAPE-AL model has specifically called for work examining irritability as an emotional sensitivity factor and examining whether these effects may extend to threatening or stressful interpersonal situations outside of the caregiving environment, such as peer victimization (Kimonis, 2023). This study aims to respond to this call.

CU behaviors may also develop in response to environmental and interpersonal stressors specifically for individuals who exhibit high HPA axis activity. The HPA axis is a component of the mammalian stress response system and is crucial for maintaining allostasis across several physiological systems (Lightman & Conway-Campbell, 2010). HPA axis activity is typically indexed using salivary assays of the hormone cortisol, and can be examined in regard to basal levels, diurnal patterns, or reactivity to stressors (Gunnar & Quevedo, 2007; Klimes-Dougan et al., 2001). Higher HPA axis activity, particularly basal levels, have been hypothesized to reflect greater biological sensitivity to environmental inputs (Boyce & Ellis, 2005). In fact studies have pointed to a role of hyperactive HPA axis functioning in the development of CU behaviors, especially at younger ages and when considering interactions with environmental or interpersonal stressors. For instance, among a high-poverty birth cohort sample, children with higher levels of conduct problems and CU behaviors in first grade demonstrated higher levels of basal cortisol and cortisol output during a reactivity task at 15 months old (Mills-Koonce et al., 2015). Furthermore, in the same sample, maternal harsh intrusion in infancy predicted low empathic and prosocial behaviors (a component of CU behaviors) in middle childhood specifically for children who also demonstrated high basal cortisol in infancy (Waller & Wagner, 2019). The ESCAPE-AL model (Kimonis, 2023) proposes that extreme or chronic interpersonal stress would place greater strain on the systems of individuals who are biologically sensitive to their environment; in turn, this combination of stressors and sensitivity would “overload” their psychophysiological systems, ultimately resulting in hypoactivity and the development of callousness (Kimonis, 2023; McEwen, 1998; Mills-Koonce et al., 2015).

Of note, hypoactive HPA axis functioning is theorized to be a genetically-mediated predisposing factor for the development of CU behaviors which may operate regardless of environmental risk (Hawes et al., 2009). HPA axis hypoactivity is thought to reflect fearlessness and sensation-seeking due to a lack of responsivity to the environment (Blair, 1999; Coren, 1999; Quay, 1965). These lower levels of fear reactivity are theorized to underlie disruptions in learning processes which would typically socialize youth away from engaging in CU behaviors (Waller & Wagner, 2019), consistent with the “primary” pathway to CU behaviors (Craig et al., 2021; Yildirim & Derksen, 2015).

Punitive or hostile parenting practices were assessed at T1 using parent report on the Authoritarian scale (8 items) of the Parenting Styles and Dimensions Questionnaires (PSDQ; Robinson et al., 2001). This scale includes non-reasoning/punitive (4 items; e.g., “I punish by taking privileges away from our child with little to no explanation”) and verbal hostility (4 items; e.g., “I yell or shout when my child misbehaves”) subscales. Items responses were scored from 1 (Never) to 5 (Always) and averaged. The scale has strong reliability in the past (e.g., Olivari et al., 2013) and demonstrated lower than convention, but adequate, internal consistency in the present study (Cronbach’s α = 0.65, McDonald’s ω = 0.63).

At T1, teachers completed the Preschool Peer Victimization measure– Teacher Report– Revised (PPVM-TR-R; Crick et al., 1999; Godleski et al., 2015) to assess relational (4 items; e.g., “This child gets left out of the group when someone is mad at them or wants to get back at them”) and physical (4 items; e.g., “This child gets hit, kicked, or pinched by peers”) victimization. Items were scored from 1 (Never or almost never true) to 5 (Always or almost always true) and were averaged within subscale. Subscales demonstrated adequate internal consistency in the present study (Relational victimization Cronbach’s α = 0.87, McDonald’s ω = 0.87; Physical victimization Cronbach’s α = 0.84, McDonald’s ω = 0.84).

Irritability was assessed at T1 using a composite of teacher report on two questionnaires assessing children’s tendency to experience or exhibit anger and frustration– the Anger/Frustration subscale of the Child Behavior Questionnaire– Short From (CBQ-SF; Putnam & Rothbart, 2006; Rothbart et al., 2001; 6 items; e.g., “Gets quite frustrated when prevented from doing something they want to do”) and 4 items adapted from an observational method assessing behavioral displays of anger (e.g., “Uses toys or classroom materials roughly"; Hubbard et al., 2004). These items are similar to recently-developed measures capturing tonic and phasic irritability (Silver et al., 2025). CBQ-SF items were scored on a 1 (Extremely untrue) to 7 (Extremely true) scale; anger display items were scored on a 1 (Never) to 4 (Almost always) scale. Items were averaged within scale, then scales were standardized and averaged to create an irritability composite, consistent with prior work with this scale which has supported a unidimensional factor structure and convergence across parent and teacher report (Perhamus & Ostrov, 2021, 2023). In this study, both subscales demonstrated adequate internal consistency (Cronbach’s αs =.76–.87, McDonald’s ω =.75–.87) and were moderately correlated (r =.59, p <.001).

At T1 and T2, parents completed the Inventory of Callous Unemotional Traits– Preschool Version (ICU; Frick, 2004), which is a 24-item measure assessing callousness (e.g., “Does not care who they hurt to get what they want”), uncaring (e.g., reverse coded “Feels bad or guilty when they have done something wrong”) and unemotionality (e.g., “Does not show emotions”). Items are rated on a 0 (Not at all true) to 3 (Definitely true) scale and averaged. Internal consistency was robust (Cronbach’s αs = 0.81– 0.83, McDonald’s ωs = 0.83– 0.91).

Saliva samples were collected at T1 on three consecutive days in the morning (between approximately 9:00am– 11:00am) using SalivaBio Children’s Swabs. Collection was completed at home (Cohorts 2 and 3) or school (Cohorts 1 and 4) depending on Covid-19 restrictions. Cohort was included as a covariate in all models to account for Covid-19 impacts. Swabs were held under children’s tongues for 90 seconds while they played a “food game” in which they viewed pictures of food and indicated if they liked the food with a thumbs up or down. This procedure was designed to stimulate saliva and maintain children’s attention. At school, children were assessed in a small group outside of their classrooms with one RA attending to one or two children at a time. Children refrained from eating, drinking, engaging in vigorous exercise (e.g., running), sleeping, and brushing teeth within the prior hour. Children must not have been experiencing symptoms of illness (e.g., fever) within the last 24 hours per their caregivers or teachers. Efforts were made to schedule school-based saliva collection on days that represented typical school days and after all children had been at the childcare center for at least 30 minutes. For home-based collection, caregivers were provided with training materials, detailed instructions, and completed daily check-ins with project staff. Caregivers were instructed to place samples in the back of their freezers until picked up by project staff.

Samples were stored at -27 °C in a laboratory freezer for up to 6 months and then packaged with dry ice and sent to Salimetrics SalivaLab (Carlsbad, CA) for assay. Samples were assayed using the Salimetrics Salivary Cortisol Assay Kit (Cat. No. 1-3002). Samples were thawed, vortexed, and centrifuged prior to assay. The assay involved a high sensitivity enzyme immunoassay, during which 25 µl of saliva were extracted from each sample and tested using a lower sensitivity limit of 0.007 µg/dL, a 0.012–3.00 µg/dL standard curve range, a 4.6% average intra-assay coefficient of variation (CV), and a 6% inter-assay CV. These criteria have been determined to be accurate by Salimetrics SalivaLab and exceed guidelines set by the National Institutes of Health. Samples with insufficient saliva or that returned biologically implausible values were excluded. Each sample was assayed in duplicate and values were averaged to create a daily index. Daily values were then averaged to create a time point value. Samples were moderately correlated across days (rs = 0.30 − 0.54, ps < 0.001; McDonald’s ω = 0.66).

To account for diurnal changes in basal cortisol levels, the amount of time that passed between the child’s typical awakening time and cortisol collection time was included as a covariate in models examining cortisol. Parents reported on their child’s typical wake time during the week. Saliva collection time was recorded by undergraduate research assistants for school-based collection, and by parents for home-based collection, and the collection time was averaged across samples to create an average collection time for the time point. Both variables were then converted to a continuous numeric variable, and the child’s typical wake time was subtracted from the average collection time to create an index of time since awakening. As expected, this variable was moderately negatively correlated with basal cortisol values (r = −.31, p <.001), consistent with typical diurnal cortisol patterns which would be expected to decrease across the morning (Klimes-Dougan et al., 2001).

The model examining main effects of authoritarian parenting and physical and relational victimization on CU behaviors was just-identified. Contrary to predictions, none of these variables directly predicted change in CU behaviors over the transition to kindergarten.

Irritability’s main effect on CU behaviors was then added to the main effects model. This model was also just-identified, and there was no significant direct effect of irritability on CU behaviors. Interaction terms between irritability and physical victimization, irritability and relational victimization, and irritability and authoritarian parenting were then added. This model was also just-identified. A significant interaction between irritability and authoritarian parenting emerged. The RoS analysis found that the effect of authoritarian parenting on CU behaviors was significant at less than − 5.74 SDs and above 2.12 SDs from the irritability mean. As the lower bound of the RoS was below the observed range of the data, only the upper bound was considered further. Specifically, at 2.12 SDs above the irritability mean, authoritarian parenting predicted increases in CU behavior over the transition to kindergarten (B = 0.26, SE = 0.13, p =.05). In other words, consistent with hypotheses, authoritarian parenting predicted increases in CU behaviors specifically at high levels of irritability. Simple slopes are depicted in Fig. 1.

Next, all effects of irritability were removed from the model, and the main effect of cortisol was added. Cortisol was also regressed onto time since awakening. As T2 CU behaviors and predictor variables other than cortisol were not regressed onto the time since awakening covariate, this model was over-identified and provided good fit to the data [χ²(5) = 5.60, p =.35; RMSEA = 0.02; CFI = 1.00; SRMR = 0.01]. Contrary to predictions, there was a significant negative main effect of cortisol on CU behaviors, suggesting that lower levels of cortisol predicted increases in CU behaviors across the transition to kindergarten. Finally, interaction terms were added. The model continued to provide good fit to the data [χ²(5) = 5.42, p =.37; RMSEA = 0.02; CFI = 1.00; SRMR = 0.01]. No interaction terms were significant.