Applying new RDoC dimensions to the development of emotion regulation: Examining the influence of maternal emotion regulation on within-individual change in child emotion regulation

Abstract

While the Research Domain Criteria (RDoC) acknowledges that environmental and developmental influences represent important elements of the RDoC framework, there is little specificity regarding how and when to systematically examine the impact of these dimensions on domains of function. The primary aims of this paper are to demonstrate the ways in which the RDoC can be expanded to include an explicit emphasis on 1) examining within-individual change in developmental processes over time and 2) evaluating the extent to which selective and measurable environmental influences drive meaningful change during key developmental periods. We provide data from an ongoing randomized control trial as a proof of concept to highlight how repeated assessments within an experimental intervention design affords the unique opportunity to test the impact of environmental influences on within-individual change. Using preliminary data from 77 mother-child dyads repeatedly assessed across 12 months during the preschool period, we demonstrate the influence of changes in maternal emotion regulation (ER) on within-individual growth in child ER and link that growth to fewer teacher-reported externalizing problems. In line with this Special Issue, findings are discussed within the context of expanding and clarifying the existing RDoC framework to explicitly incorporate environmental and developmental dimensions.

While the Research Domain Criteria (RDoC) now openly acknowledges that the environment and development are “equally important elements of the RDoC framework” (Garvey, Avenevoli, & Anderson, 2016), there is little specificity regarding how and when to systematically examine the impact of these dimensions on particular domains of function. Although this allows investigators the freedom to define which environmental influences and developmental periods may be most relevant for their research (Cuthbert, 2014; Garber & Bradshaw, 2020), it jeopardizes rigor and reproducibility, and risks recapitulating field-wide norms that are driven by tradition or researcher preference. Moreover, the RDoC does not explicitly delineate development as a dynamic process, a central tenet of developmental psychopathology that necessitates repeated assessments over time (Cicchetti & Rogosch, 1999). In line with recent commentaries calling for an expansion of the RDoC framework, we highlight the need to incorporate explicit consideration of both environmental and developmental influences on within-individual change. The primary aim of this paper is to demonstrate the ways in which the RDoC can be expanded to reflect the ongoing impact of environmental and developmental dimensions on emotion regulation (ER). Specifically, we propose including an explicit emphasis on 1) examining within-individual change in this developmental process over time and 2) evaluating the extent to which selective and measurable environmental influences drive that change during key developmental periods. We focus specifically on the influence of maternal ER on the development of child ER during the preschool period, a key developmental window known for rapid emotional development (Carlson & Wang, 2007; Denham et al., 2003).

We first provide a brief overview of the RDoC and recent critiques calling for the necessary incorporation of environmental and developmental dimensions. We then summarize literature pointing to the importance of examining ER as a within-individual process that changes over time as a function of environmental and developmental influences. Next, we provide data from our own work as a proof of concept to highlight how repeated assessments within an experimental intervention design affords the unique opportunity to test mechanisms of within-individual change in developmental processes. Specifically, we examine the influence of changes in maternal ER on within-individual change in child ER during the preschool period and assess how these within-individual changes are linked to internalizing and externalizing problems. Findings are discussed within the context of expanding and clarifying the existing RDoC framework to explicitly incorporate environmental and developmental dimensions. We also consider how this approach can be translated to other key periods of development and extended to other relevant developmental processes.

Widening the RDoC Lens to Include Multiple Dimensions

For the last decade, the RDoC has provided an alternative framework for research on mental disorders. The goal of the RDoC is to characterize the pathophysiology of psychopathology, with a specific emphasis on genomics and neuroscience, and identify fundamental circuit-behavior relationships that exist transdiagnostically (Insel et al., 2010). The RDoC aims to create a unified structure to investigate individual differences across multiple domains of functioning and multiple levels of analysis. The overarching goal of this work is to advance our understanding of behavior across the full range of functioning, and ultimately identify and refine targets for intervention (Cuthbert & Insel, 2013). While the RDoC has undoubtedly contributed to advancements in our understanding of psychopathology (Carcone & Ruocco, 2017), recent research has highlighted ways in which its utility and applicability may be enhanced, ultimately increasing traction in pursuit of its intended goals.

Two of the most common critiques have focused on the importance of incorporating dimensions of the environment and development into the RDoC framework. This call for a widening of the RDoC lens stems from decades of research highlighting the ways in which environmental and developmental influences inextricably shape domains of functioning across all levels of analysis (Franklin, Jamieson, Glenn, & Nock, 2015; Garber & Bradshaw, 2020). Accordingly, researchers have proposed variations of a four-dimensional model, including one that retains the original two dimensions and adds two new dimensions that capture critical environmental and developmental influences (Mittal & Wakschlag, 2017; Woody & Gibb, 2015). Along these lines, Garber and Bradshaw (2020) suggest that the inclusion of environmental and developmental dimensions is not only “compatible and complementary” to the current RDoC perspective, but necessary to “inform future research and interventions…with children and adolescents, [and] with adults across the lifespan” (p. 342). These sentiments have been echoed in recent empirical work (e.g., Ip, Jester, Sameroff, & Olson, 2019) and conceptual papers focused on a wide range of psychopathology (e.g., Beauchaine & Hinshaw, 2020; Glenn, Cha, Kleiman, & Nock, 2017), as well as in other special issues of journals including the Journal of Clinical Child & Adolescent Psychology (Garber & Bradshaw, 2020) and Development & Psychopathology (Beauchaine & Cicchetti, 2016). Taken together, this highlights a growing emphasis on better characterizing change that occurs within-individuals across development and quantifying the extent to which that change is influenced by environmental factors.

ER as a Developmental Process: Investigating Within-Individual Change

Emotion regulation (ER) is a dynamic process by which individuals modify the intensity and/or duration of their emotional experience in response to situational circumstances (Cicchetti, Ackerman, & Izard, 1995; Gross, 1998; Thompson, 1994). Effective regulation of emotion aids in organizing adaptive responses to shifting environmental demands (Thompson, 1994) and reflects concomitant changes in cognitive, emotional, and social domains (Fox, 1994; Saarni, Campos, Camras, & Witherington, 2007). Although competing conceptualizations of ER exist, researchers generally agree that it is a multifaceted process that is continually modified across development (Aldao, 2013; Cole, Martin, & Dennis, 2004; Crowell, Vlisides-Henry, Kaliush, & Beauchaine, 2020; Morris, Criss, Silk, & Houltberg, 2017; Walden & Smith, 1997).

Regardless of the conceptual model, understanding developmental changes in ER represents a central tenet ubiquitous to all ER theory. However, there is a dearth of research describing how this process develops within-individuals over time and, at present, the RDoC offers little guidance on how to achieve this aim. To date, our understanding of the maturation of ER rests almost entirely on evidence obtained from between-person differences utilizing either age-based cohort studies (e.g., comparing delay of gratification in a cohort of 3-year-olds and a cohort of 4-year-olds; Hongwanishkul, Happaney, Lee, & Zelazo, 2005) or prospective longitudinal studies that focus on changes in rank-order stability (for exceptions see Blandon, Calkins, Keane, & O’Brien, 2008; Bocknek, Brophy-Herb, & Banerjee, 2009; Lengua et al., 2015; Moilanen, Shaw, Dishion, Gardner, & Wilson, 2010). While this research has been critical in identifying key development periods of rapid ER growth, continuing to pursue between-person difference designs is unlikely to lead to any new or substantial discoveries (Cole & Jacobs, 2018). Advances in our methodological and analytic approach are needed to more accurately reflect and capture theoretical conceptualizations of ER as a within-individual process that develops over time (Cicchetti & Rogosch, 1999; Kim-Spoon & Grimm, 2016; Wood, 2011).

Here, we aim to illustrate the potential utility of using repeated assessments to model within-individual change in ER using the RDoC framework. Given that ER is not currently defined by a specific RDoC domain (see Fernandez, Jazaieri, & Gross, 2016; Sun, Vinograd, Miller, & Craske, 2017 for discussions), we build on previous work that utilized the RDoC to assess functioning across related cognitive, emotional, and social domains (Ip et al., 2019). In line with Ip and colleagues (2019), we focus on within-individual changes in four validated constructs across four RDoC domains during the preschool period: executive control (cognitive system), delay of gratification (positive valence system), regulation of frustration (negative valence system), and emotion knowledge (social processes system). While each of these constructs contribute to the development of ER, they are also conceptualized under the larger umbrella of “self-regulation” (Posner & Rothbart, 2000). The extent to which these constructs constitute ER and/or self-regulation is under ongoing investigation.

Executive control, also termed effortful control, is defined as the ability to regulate attentional and behavioral impulses (Rothbart et. al 2006). Executive control can be conceptualized in terms of “hot” and “cool” effortful control, with the latter referring to attentional and inhibitory control in response to neutral, non-rewarding stimuli (e.g., Stroop-like tasks) (Kim, Nordling, Yoon, Boldt, & Kochanska, 2013; Sturge-Apple, Davies, Cicchetti, Hentges, & Coe, 2017). “Hot” executive control, referred to here as delay of gratification, denotes inhibition in the face of heightened emotion elicited by a rewarding stimulus (e.g., a toy; Kochanska, Murray, Jacques, Koenig, & Vandegeest, 1996). Regulation of frustration is another key domain of ER, and it refers to the ability to modify the intensity and/or valence of emotional responses during stressful situations to achieve a desired goal (Cole, Martin, & Dennis, 2004). Finally, emotion knowledge refers to one’s understanding of emotion as well as the ability to perceive and understand others’ emotional states and reactions (Izard et al., 2011). While many studies have examined these constructs in isolation (e.g., Bendezú et al., 2018; Carlson & Wang, 2007; Moilanen et al., 2010; Supplee, Skuban, Trentacosta, Shaw, & Stoltz, 2011), it has become increasingly clear that the simultaneous consideration of constructs across multiple domains may yield important information about the development of ER (see Campos, Frankel, & Camras, 2004; Ip et al., 2019; Saarni et al., 2007; Zelazo & Cunningham, 2007). Moreover, enhancing our understanding of within-individual changes in these domains can sharpen the precision with which we can evaluate the impact of known environmental influences on developmental trajectories of ER.

Environmental Influences on ER: The Role of Maternal ER

Decades of research and theory highlight the influential role of the environment, particularly parental influences, in shaping individual differences in the development of ER (Calkins & Hill, 2007; Hajal & Paley, 2020; Morris et al., 2017). A parent’s response to their child’s emotion provides immediate feedback about the acceptability of emotions, shaping the way in which emotions are understood, experienced, expressed and regulated in the future (Eisenberg, Cumberland, & Spinrad, 1998; Morris, Silk, Steinberg, Myers, & Robinson, 2007). Mounting research suggests that parents with ER difficulties may find it especially challenging to respond in a supportive manner to their child’s expression of emotion (e.g., Buckholdt, Parra, & Jobe-Shields, 2014). For example, parents with ER difficulties may feel overwhelmed by their own internal emotional experience when attempting to navigate challenging emotional interactions with their child, making it more difficult to attend and respond effectively (Rutherford, Wallace, Laurent, & Mayes, 2015). Along these lines, research suggests that parents with ER difficulties are much more likely to ignore, dismiss, magnify, or punish their child’s emotional experience, even if unintentionally (Morelen, Shaffer, & Suveg, 2016). This combined with parental modeling of poor ER strategies places children at increased risk for ER difficulties (Li, Li, Wu, & Wang, 2019; Morris et al., 2017; Morris et al., 2007), and in turn, increased risk for internalizing and externalizing psychopathology (Crespo, Trentacosta, Aikins, & Wargo-Aikins, 2017; Ip et al., 2019; Kaufman et al., 2017; Röll, Koglin, & Petermann, 2012).

Taken together, these findings underscore the importance of more precisely characterizing the impact of known environmental influences, like maternal ER, on within-individual change in child ER. While previous work in this area has undoubtedly advanced our understanding of links between environmental influences and ER development (Bariola, Gullone, & Hughes, 2011), traditional study designs are naturalistic and often focus on between-person differences over time, hindering our ability to precisely pinpoint environmental influences as causal predictors of within-individual change in ER (Rutter, Pickles, Murray, & Eaves, 2001). At present, the RDoC provides little guidance regarding how to incorporate environmental influences into study designs, impeding collective efforts to capture important explanatory variance in ER as a developmental process. The implementation of prospective, experimental designs would afford opportunities to empirically test environmental influences as causal mechanisms of change in ER and enhance progress toward the identification of potentially malleable intervention targets. For example, employing experimental interventions that target maternal ER would allow us to more precisely quantify the causal impact of this environmental influence on the within-individual change in child ER. Given that intervention studies suggest that parent factors like maternal ER may moderate the effectiveness of traditional parenting interventions (e.g., Shelleby & Shaw, 2013) and that targeting maternal ER can improve child outcomes (Havighurst et al., 2013; Kehoe, Havighurst, & Harley, 2020), there is strong impetus for utilizing experimental manipulation of known environmental influences to examine effects on potential changes in ER over time.

Developmental Influences on ER: Preschool as a Key Period

Prominent theory and empirical research point to key developmental periods across childhood and adolescence during which ER development accelerates (Carlson & Wang, 2007; Denham et al., 2003), highlighting when we might aim to examine change in this developmental process. Incorporating more explicit guidance in the RDoC about the timing and tempo with which these processes unfold across time offers a roadmap for optimal intervals of assessment (e.g., frequency and duration of assessment) organized by developmental period. For example, the preschool period is characterized by marked neurobiological changes that underlie notable shifts in cognitive, social, and emotional domains of functioning (Brown & Jernigan, 2012; Garon, Bryson, & Smith, 2008), making it a prime target for frequent assessments of within-individual change in ER. Moreover, the sensitivity of this developmental period makes it particularly vulnerable to environmental influences. Indeed, parental influences on ER development appear to be particularly salient during the preschool period (Cole, Lougheed, & Ram, 2018), highlighting the potential significance of clarifying causal mechanisms during this time. Given that prospective work has linked ER problems during the preschool period to the subsequent emergence of internalizing and externalizing problems (e.g., Ip et al., 2019; Röll et al., 2012), focusing our efforts on the development of ER during this developmental period may yield important implications for etiological and intervention models of psychopathology (Beauchaine & Cicchetti, 2019).

Current Study

Here, we leverage data from the STEADY Study, an ongoing, multisite, randomized control trial (RCT), which utilizes Dialectical Behavior Therapy (DBT), a robust and effective method for improving ER (Linehan, 1993; Neacsiu, Bohus, & Linehan, 2014). This prospective study design includes repeated assessments of child ER domains over 12 months, allowing for an examination of within-individual change during the preschool period (3–4 years), as well as the opportunity to test 1) how within-individual changes are impacted by changes in maternal ER and 2) whether within-individual changes are related to internalizing and externalizing problems. We focus on a sample of 77 mother-child dyads recruited from a single site between October of 2017 and January of 2020. This includes 39 mothers with ER difficulties (i.e., mothers with borderline personality disorder; BPD) and 38 non-disordered, control mothers (healthy controls; HC). Half of mothers with BPD were randomly assigned to one year of standard DBT Skills training, and the other half to “Waitlist Control” (WLC), which allows us to assess the impact of changes in maternal ER on changes in child ER.

We first examine within-individual change in four domains of child ER (i.e., executive function, delay of gratification, regulation of frustration, and emotion knowledge), assessed repeatedly across 12 months (i.e., baseline, 4 months, 8 months, 12 months). To do so, we utilize a free curve slope intercept (FCSI) latent growth curve model, which makes the fewest assumptions about the form or rate of growth (Wood, 2011; Wood & Jackson, 2013). While this approach to modeling within-individual variability is only one of many, the free curve slope intercept (FCSI) model allows us to distinguish between the relative amounts of between-person (intercept) and within-individual (slope) differences for each of the four ER domains. Next, we examined the impact of changes in maternal ER on between-person differences (intercept) and within-individual change (slope) for each of the four ER domains. Finally, we assessed whether the effect of changes in maternal ER on change in child ER domains impact teacher-reported internalizing and externalizing problems. We expect that there will be significant between-person differences and within-individual changes in each ER domain across the 12-month period. We also hypothesize that improvements in maternal ER will predict within-individual changes in each of the child ER domains. Finally, we predict that within-individual changes in ER domains will be associated with internalizing and externalizing problems at 12-month follow-up.

Method

Sample

Participants were 77 mothers (M_age=33.17 years, SD=4.83; 35% racial/ethnic minority) and their children (M_age=42.48 months; SD=3.78; 56% female; 47% racial/ethnic minority) enrolled in an ongoing RCT. Two groups of mother-child dyads were recruited using a university-based Research Recruitment Program which employed a targeted multimedia and digital messaging program for mothers of children between 36 and 47 months old (i.e., one advertisement targeting BPD and another targeting HC). Mothers with BPD were also recruited from ambulatory psychiatric treatment clinics in the same geographic region.

Inclusion and Exclusion Criteria

Initial phone screening was conducted to ensure that all mothers were biological mothers with at least 50% physical custody of the target child since birth. During this call, all mothers also completed the 10 item McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD; Zanarini et al., 2003): scores ≥ 7 were required for the BPD groups and scores ≤ 2 were required for the HC group. Additionally, mothers in the BPD group were asked about current treatment engagement and any mothers currently participating in DBT services were deemed ineligible.

Eligibility was further determined during an in-person clinical intake, which included semi-structured diagnostic interviews for mothers. Trained research personnel administered the Structured Clinical Interview for DSM-5 (SCID-5; First, Williams, Karg, & Spitzer, 2015) and the Structured Interview for DSM-IV Personality (SIDP-IV; Pfohl, Blum, & Zimmerman, 1995) to determine eligibility. Approximately, 20% of intake interviews were double-coded from digital recordings and showed strong inter-rater reliability (Krippendorff’s α = .91). Mothers in the BPD group met at least 3 diagnostic criteria for BPD, with one of these three symptoms being affective instability or uncontrolled anger (symptom range=3–9; mean=5.13; SD=1.82). HC mothers reported no history of psychiatric illness (e.g., depression, anxiety) currently or since their child’s conception, and showed no evidence of clinically significant emotion regulation difficulties as determined by scoring 0 on the affective instability and uncontrolled anger criteria on the SID-P. Additionally, any mothers in a current psychotic or manic episode were deemed ineligible.

All mothers and children also completed the Peabody Picture Vocabulary Test Fourth Edition (PPVT-IV; Dunn & Dunn) and the Expressive Vocabulary Test Second Edition (EVT-2; Williams, 1997) to provide an estimate of IQ. All eligible participants demonstrated standard scores ≥ 70. All mothers also completed the Ages and Stages Questionnaire (ASQ; Squires, Bricker, & Twombly, 2009), a brief developmental screening instrument designed to detect developmental delays across five domains of development (i.e., communication, gross motor, fine motor, problem-solving, and personal-social skills); only children without developmental delays were eligible.

Procedure

Those who were eligible were enrolled to complete four laboratory assessments (baseline, 4-month, 8-month, and 12-month follow-ups). This assessment schedule paralleled the 12-month DBT treatment schedule (described below). Assessments were conducted by research assistants blinded to mother’s group status and included questionnaires to assess maternal ER and behavioral tasks that assessed each of the four child ER domains (described below). Mothers also provided contact information for their child’s preschool teacher or daycare provider. Following the baseline and 12-month laboratory assessment, teachers were contacted and asked to complete an online questionnaire assessing internalizing and externalizing problems observed in the preschool/daycare setting. All mother-child dyads completed the in-person baseline assessment, 81.8% completed the 4-month assessment, 71.4% completed the 8-month assessment, and 45.9% completed the 12-month assessment.¹

Randomization and Treatment

Mothers in the BPD group were randomized to receive DBT Skills (n=14) or WLC (n=25).² DBT Skills training followed the treatment protocol described by Linehan and colleagues (2015). This included participation in a weekly standard 2-hour skills group for 12 months, which consisted of teaching skills across 4-modules: mindfulness, emotion regulation, interpersonal effectiveness, and distress tolerance (Linehan, 2014). DBT mothers were assigned a study therapist who met with them at least monthly for 30 minutes and were available to meet weekly if needed. Study therapists were also available for coaching calls. Additionally, all study therapists attended weekly consultation team meetings for 1 hour. Study therapists were master and doctoral level clinicians, all of whom had significant training in DBT. The DBT Adherence Scale (Harned, Korslund, Schmidt, & Gallop, 2021) was used to code randomly selected DBT Skills group sessions for adherence and all scored sessions were deemed adherent. Participants randomized to the DBT condition completed, on average, 30 weekly sessions over the course of a year (63% of the 48-week treatment protocol), consistent with treatment compliance in seminal RCTs demonstrating the efficacy of DBT (Linehan et al., 2006; Lynch, Trost, Salsman, & Linehan, 2007). The current study implemented an intent-to-treat design, and all participants were retained in analyses.

Measures

Maternal Emotion Regulation Difficulties

Maternal ER difficulties were measured using the Difficulties with Emotion Regulation Scale (DERS; Gratz & Roemer, 2004). The DERS is a 36-item self-report measure of perceived emotion regulation abilities. Mothers indicated how often the items apply to them on a 5-point Likert-type scale ranging from 1 (almost never) to 5 (almost always). The DERS measures 6 subscales of emotion regulation, including lack of emotional awareness, lack of emotional clarity, impulse-control difficulties, difficulties engaging in goal-directed behaviors, non-acceptance of emotional responses, and limited access to emotion regulation strategies. All items were summed to create a total score, with higher scores representing greater ER difficulties.

Child Emotion Regulation Domains

Executive Control was measured using the Dimensional Change Card Sort (DCCS; Diamond, Carlson, & Beck, 2005; Frye, Zelazo, & Palfai, 1995; Hongwanishkul et al., 2005). In this task, children were first shown two boxes with target cards attached to the front of the box (i.e., blue card with a black star silhouette, red card with a black truck silhouette) and an open slot on the top of each box. Children were then handed sorting cards (i.e., blue cards with a black truck silhouette, red cards with a black star silhouette), and instructed to sort the cards according to shape (6 trials) and then according to color (6 trials). Before each shape and color trial, the experimenter stated the sorting rule and presented the child with a card. Once the child demonstrated color and shape knowledge, they advanced to the next level, where the target cards integrated the two sorting properties. Specifically, the target cards had either a red truck or blue star on white backgrounds, and sorting cards were red stars and blue trucks (12 trials). Finally, children were presented with target cards with and without a border. Children were instructed to sort by color if the card had a border and by shape if the card did not have a border (12 trials). Each trial was scored incorrect (0) or correct (1) and summed to create a total score ranging from 0–36, with higher scores representing better executive control.

Delay of Gratification was measured using the Gift Delay Task (Kochanska et al., 1996). In this task, the child was told that they will receive a present from the experimenter. They were instructed to sit on a chair facing in the opposite direction and specifically told not to peek while the experimenter wrapped their gift. The experimenter then noisily wrapped the present for 60 seconds while the child waited. The current study focused on latency to the child’s first peek (i.e., child’s head, shoulder, or body turning to peek at the gift) as an indicator of delay ability (range 0–60 seconds), with higher scores indicating better delay of gratification.

Regulation of Frustration was measured using two well-validated frustration tasks: Transparent Locked Box (LAB-TAB; Goldsmith & Rothbart, 1993) and Knotted Sack (Chaplin, Klein, Cole, & Turpyn, 2017). These tasks were alternated across the four assessments to mitigate practice effects associated with tasks including an element of deception (Monks, Smith, & Swettenham, 2005). In each task, the child was allowed to pick one of three toys. The experimenter then placed the chosen toy inside the locked box or knotted sack and told the child that once they opened the locked box/knotted sack, they could keep the toy. The experimenter explained that they “had some work to do”, turned their chair away from the child, and pretended to do work while the child attempted to open the box or sack. The child was left alone for 2 minutes to open the locked box with the incorrect set of keys or to open the tightly knotted sack. The tasks were coded from digital files using an adapted coding system (Dennis, 2004). The current study included a measure of on-task or problem-solving behavior scored in 4, 30-second epochs. On-task behavior could include any of the following: 1) child using strategies to solve the problem; 2) child working toward opening the box or sack; or 3) child seeking information from the experimenter about how to open the box or sack. Each epoch received a score of 0 (none or minimal on task behavior), 1 (on task behavior for about half of the epoch), or 2 (on task behavior for nearly all or all of the epoch) and 20% of participants were double-coded to assess reliability (average epoch ICC = .78). Scores for each epoch were then summed to create a total score ranging from 0–8, with higher scores indicating more on-task behavior.

Emotion Knowledge was measured using an adapted version of the Affective Perspective Taking Task (Denham, 1986; Pears & Moses, 2003; Zalewski, Musser, Binion, Lewis, & O’Brien, 2020). Prior to task initiation, children were shown 4 emotional faces (happy, sad, mad/angry, scared/afraid) and were instructed to identify all emotions both expressively, then receptively (by pointing). Next, an experimenter taught the child all four basic emotion faces to ensure the child’s understanding. The experimenter then presented 8 stereotypical situation vignettes using 2 puppets: a main character (“Nancy/Johnny”) who always matched the sex of the child, and a sibling who was always the opposite sex. The puppets acted out 8 scenes portraying various events that might elicit happiness, fear, sadness, or anger (e.g., getting an ice cream cone, being in a dark room). At the end of each vignette, the child was asked to choose a face that described how the puppet was feeling (happy, sad, mad/angry, scared/afraid). The child received 2 points if they matched the puppet’s expressed emotion to the correct face and 1 point if they chose the wrong face but correctly identified the emotion valence. Scores for each vignette were then summed to create a total score ranging from 0–18, with higher scores reflecting better affective perspective taking.

Child Internalizing and Externalizing Problems

To assess internalizing and externalizing problems, preschool teachers or daycare providers completed the Teacher Report Form (TRF; Achenbach, 1991) via the Qualtrics survey system (Qualtrics, Provo, UT). The TRF contains 102 items, assessing internalizing problems and externalizing problems. All items are rated on a 3-point Likert scale, ranging from 0 (not true), to 2 (very true, or often true), and are then summed to create an internalizing problems and externalizing problems score. Teacher data was available for 83% of the sample (n=64) at baseline and 65% (n=50) of the sample at the 12-month assessment. At baseline, missing data included 7 children who were not yet enrolled in preschool or daycare, 2 children whose parent did not give consent to contact their teacher, 1 child whose teacher declined participation, and 3 children for whom data was determined to be invalid (e.g., teachers completed the questionnaire in less than 5 minutes). At the 12-month assessment, missing data included 3 children who were still not enrolled in preschool or daycare, 19 children for whom data was missing due to preschool/school closures during the global pandemic, 2 children who stopped attending preschool or daycare (unrelated to the global pandemic), and 3 children who were lost to follow-up.

Covariates

Demographic Variables.

Child age, racial/ethnic minority status (0=white; 1=minority status), and sex (0=male; 1=female) were obtained via maternal report at the first assessment.

Receipt of Public Assistance.

Mothers reported on their annual income and whether or not they received any financial assistance (e.g., WIC, food stamps, welfare, etc.; 0=no public assistance; 1=received public assistance).

Estimated IQ.

Verbal IQ was estimated using age-based standard scores on the Peabody Picture Vocabulary Test Fourth Edition (PPVT-IV; Dunn & Dunn).

Data Analytic Plan

We conducted preliminary analyses to examine descriptive statistics by maternal group status and bivariate correlations between all study variables using IBM SPSS Statistics (Version 26.0). Additionally, we examined the extent to which our RCT produced changes in maternal ER difficulties using multigroup latent growth curve analysis (Jung & Wickrama, 2007) in MPlus version 8 (Muthén & Muthén, 2017). Full Information Maximum Likelihood (FIML) was utilized to handle missing data. In this model, negative slopes indicate improvements in ER difficulties while positive slopes indicate worsening ER difficulties. Slopes were compared using Satorra-Bentler scaled chi-square test statistic to compare fit for a model in which slope parameters were specified to vary to a model where slope parameters were constrained to be equal. This comparison was made for all possible group combinations (i.e., DBT vs. HC; WLC vs. HC; DBT vs. WLC).

All primary analyses were conducted in MPlus version 8 (Muthén & Muthén, 2017) using Full Information Maximum Likelihood (FIML) with robust standard errors (MLR) to handle missing data. Model fit was evaluated using standard criteria for chi-square, Comparative Fit Index (CFI; Bentler, 1990), and the Root Mean Square Error of Approximation (RMSEA, Browne & Cudeck, 1993). To examine within-individual change in child ER domains, we utilized a free curve slope intercept (FCSI) latent growth model (Wood, 2011). This model was chosen because it makes the fewest assumptions about the form or rate of growth (Wood, 2011; Wood & Jackson, 2013). Additionally, this model sets the intercept and the slope to be orthogonal (the covariance of between the intercept and slope variables are set to zero), allowing us to evaluate the relative amounts of between-person (intercept) and within-individual (slope) variability for each of the four ER domains. We report 1) intercept variance as the indicator of between-person differences in ER across 12 months in preschool; 2) squared slope loadings as an indicator of within-individual variability at each assessment; and 3) a comparison of intercept variance to squared slope loadings as an indicator of relative differences in between-person versus within-individual differences.

Next, we examined how change in maternal ER difficulties impacted differences in the starting point of preschooler ER growth trajectories (intercept: between-person differences) and their pattern of development over 12 months (slope: within-individual differences) above and beyond demographic covariates. To do so, we regressed the latent intercept and slope of each child domain on the slope of maternal ER difficulties while controlling for demographic covariates. Finally, to evaluate the extent to which each child ER domain was associated with internalizing and externalizing problems at 12-month follow-up, we regressed internalizing and externalizing problems (separately) on the latent intercept and slope of each child ER domains while controlling for internalizing or externalizing problems assessed at baseline, slope of maternal ER difficulties, and demographic covariates. Indirect effects were also assessed using the MODEL INDIRECT command in MPlus.

Results

Descriptive Statistics

Table 1 provides descriptive statistics for all study variables by maternal group status and Table 2 shows bivariate correlations from primary study variables across all assessments.

Table 1.

Descriptive statistics for all study variables by maternal group status.

	DBT (n=14)			TAU (n=25)			HC (n=38)
	N	M/%	SD	N	M/%	SD	N	M/%	SD
Maternal Characteristics
Age (years)	14	32.72	5.06	25	32.19	5.90	38	33.98	3.87
Minority Status	14	57%	--	25	48%	--	38	18%	--
Public Assistance	14	57%	--	25	56%	--	38	18%	--
IQ Estimate	14	94.29	16.38	24	98.92	13.57	38	104.37	13.84
Child Characteristics
Age (months)	14	42.21	4.26	25	41.88	4.01	38	42.97	3.47
Minority Status	14	0.64	0.50	25	0.64	0.49	38	0.29	0.46
Sex	14	0.57	0.51	25	0.48	0.51	38	0.61	0.50
IQ Estimate	14	105.36	16.04	24	108.04	20.06	38	113.87	15.93
Maternal ER
Baseline	14	112.07	22.39	25	108.00	24.00	38	58.05	13.92
4-months	12	99.58	22.74	15	100.40	30.81	36	55.92	12.75
8-months	12	85.00	27.91	12	96.25	19.19	26	56.96	11.97
12-months	12	85.25	17.52	20	95.05	23.34	35	56.11	11.76
Child ER
1. Executive Control (Dimensional Card Sort)
Baseline	13	15.23	5.13	25	15.56	4.96	38	18.45	4.83
4-months	10	18.30	8.03	12	19.67	7.89	24	24.46	8.49
8-months	8	23.00	9.34	10	23.30	9.37	19	27.63	6.25
12-months	8	27.50	6.26	10	23.60	8.13	16	28.25	6.84
2. Delay of Gratification (Gift Delay)
Baseline	12	27.75	25.69	23	21.78	22.66	36	41.31	24.10
4-months	12	36.08	25.13	15	30.93	24.28	36	40.06	21.56
8-months	11	38.64	25.30	12	32.33	25.25	27	39.30	23.81
12-months	8	34.00	23.44	10	34.10	27.69	16	43.69	24.44
3. Regulation of Frustration (Locked Box/Knotted Sack)
Baseline	14	5.14	2.62	24	5.12	2.62	35	4.97	2.56
4-months	11	5.18	2.40	12	3.66	2.53	34	5.20	2.47
8-months	9	5.00	1.80	10	4.70	2.31	24	4.87	2.40
12-months	6	3.33	2.33	8	6.50	2.00	13	4.92	2.72
4. Emotion Knowledge (Affective Perspective Taking)
Baseline	12	10.33	3.89	25	11.52	3.97	37	13.49	2.64
4-months	11	12.64	2.62	14	13.29	3.36	36	14.53	2.04
8-months	11	13.73	3.47	12	14.17	3.04	24	14.92	2.30
12-months	8	14.88	1.89	10	15.20	1.14	15	15.13	1.30
Teacher-Reported Problems
Internalizing Problems
Baseline	10	16.29	8.47	20	12.12	11.74	33	4.03	3.05
12-months	10	12.77	9.04	15	8.54	5.97	20	3.63	3.22
Externalizing Problems
Baseline	10	24.29	10.87	20	17.56	11.53	33	7.74	6.42
12-months	10	18.92	11.49	15	16.21	7.76	20	5.97	5.11

Table 2.

Correlations between main study variables across all timepoints

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23
Maternal ER
1. Baseline
2. 4-months	.86**
3. 8-months	.79**	.84**
4. 12-months	.86**	.79**	.71**
Child ER
Executive Control (Dimensional Card Sort)
5. Baseline	−.27*	−.32**	−.37**	−.26*
6. 4-months	−.34*	−.28	−.35	−.19	.66**
7. 8-months	−.28	−.31	−.33*	−.20	.62**	.68**
8. 12-months	−.10	−.21	−.20	−.10	.61**	.72**	.89**
Delay of Gratification (Gift Delay)
9. Baseline	−.39**	−.32*	−.37*	−.29*	.40**	.72**	.56**	.37*
10. 4-months	−.21	−.17	−.41**	−.23	.33**	.43**	.50**	.34	.29*
11. 8-months	−.19	−.27	−.21	−.10	.22	.30	.34*	.19	.26	.54**
12. 12-months	−.13	−.21	−.15	−.06	.37*	.09	.37	.17	−.02	.19	.28
Regulation of Frustration (Locked Box/Knotted Sack)
13. Baseline	−.14	−.19	.04	−.14	.04	.11	.12	.04	.15	.20	−.08	.00
14. 4-months	−.34**	−.34*	−.36*	−.36**	.22	.23	.33	.39*	.26	.23	.14	.28	−.01
15. 8-months	.00	−.18	−.06	.20	.22	.20	.12	.36	.21	.37*	.21	.09	.20	.27
16. 12-months	−.16	−.19	−.22	−.01	.03	.02	−.12	−.18	−.02	.14	.05	.16	.26	.06	.04
Emotion Knowledge (Affective Perspective Taking)
17. Baseline	−.33**	−.33*	−.31*	−.28*	.46**	.57**	.38*	.26	.40**	.18	.19	.22	.06	.17	.25	.32
18. 4-months	−.25	−.19	−.44**	−.06	.33*	.34*	.52**	.39*	.21	.19	.28	.10	−.01	.01	.31	.27	.55**
19. 8-months	−.20	−.18	−.27	−.08	.19	.10	.14	.17	.23	.30*	.35*	.17	−.01	.05	.25	.37	.43**	.69**
20. 12-months	.21	.14	.09	.30	.11	.22	.14	.23	−.07	.22	.39*	.24	−.27	−.24	.20	.15	.22	.33	.51**
Teacher-Reported Problems
Internalizing Problems
21. Baseline	.46**	.40**	.36*	.53**	−.24	−.14	−.09	−.12	−.24	−.15	−.04	−.16	−.01	−.32*	.01	−.22	−.32*	−.09	−.30	−.10
22. 12-months	.25	.29	.29	.21	−.06	−.12	−.36	−.18	−.17	−.09	.27	−.05	−.11	−.29	−.19	−.21	−.06	−.01	−.10	.19	.53**
Externalizing Problems
23. Baseline	.31*	.28*	.25	.39**	−.32*	−.21	−.10	−.01	−.31*	−.20	−.13	−.22	−.06	−.33*	−.08	−.12	−.34**	−.13	−.38*	−.16	.75**	.27
24. 12-months	.24	.37*	.38*	.18	−.14	−.24	−.35	−.30	−.27	−.39*	−.11	−.12	−.18	−.41*	−.14	.01	−.03	−.12	−.25	.04	.38*	.39*	.65**

^*p < .05;

^**p < .01

Treatment Effects on Maternal ER Difficulties

Figure S1 shows group trajectories from the multigroup LCGM assessing maternal ER difficulties across the study protocol (χ²(21) = 24.25, p = .28, CFI = .97, TLI = .97 RMSEA = .08). Mothers in the DBT (intercept: b = 110.80, p < .01) and WLC (intercept: b = 107.48, p < .01) groups demonstrated equivalent levels of ER difficulties at baseline (DBT vs. WLC: Δχ²=0.23, p>.05), and mothers in both groups demonstrated significantly more ER difficulties at baseline relative to mothers in the HC group (intercept: b = 56.77, p < .01; DBT vs. HC: Δχ²=30.86, p<.05; WLC vs. HC: Δχ²=47.40 p<.05). Additionally, mothers in the DBT group (slope: b = −9.22, p < .01) and WLC (slope: b = −4.58, p < .01) demonstrated greater improvements in ER difficulties across 12 months relative to mothers in the HC group, who showed no significant changes in ER difficulties (slope: b = - 0.12, p = .82; DBT vs. HC: Δχ²=10.28, p<.05; WLC vs. HC: Δχ²=8.69, p<.05). Mothers in the DBT group showed significantly steeper improvements in ER difficulties when compared to mothers in the WLC group (DBT vs. WLC: Δχ²=4.07, p<.05).

Between-Person Differences and Within-Individual Change in Child ER Domains

Table 3 presents model fit statistics for free curve slope intercept (FCSI) latent growth models for ER by domain, all of which demonstrated good model fit. Unstandardized parameter estimates for the intercept variance and squared slope loadings across assessments are also presented by ER domain.

Table 3.

Fit and model statistics for free curve slope intercept (FCSI) growth models for child emotion regulation by domain.

	χ2	df	p	CFI	RMSEA	Intercept	Baseline	4-month	8-month	12-month
						Variance	Loading2	Loading2	Loading2	Loading2
1. Executive Control (Dimensional Change Card Sort)	0.17	2	0.92	1.00	<0.001	0.32**	0.04	0.04	0.25	0.37
2. Delay of Gratification (Gift Delay)	0.54	2	0.76	1.00	<0.001	1.46**	3.31	0.04	0.02	0.90
3. Regulation of Frustration (Locked box/Knotted sack)	0.47	2	0.79	1.00	<0.001	1.16**	0.19	0.56	0.34	2.64
4. Emotion Knowledge (Affective Perspective Taking)	0.10	2	0.96	1.00	<0.001	2.10**	8.64	1.72	0.31	0.30

Note. Each row represents a separate model. Unstandardized parameter estimates for the intercept variance represent an estimate of relative between-person differences and squared slope loadings represent an estimate of relative within-individual differences at each timepoint across child ER domains.

^**p < .01 indicates significant between-person variance.

Executive Function.

The intercept variance indicates significant between-person differences in levels of executive control across the 12-month preschool period. An examination of squared slope loadings at each assessment shows increases in the relative amount of within-individual variability in executive control (i.e., 0.04 at baseline and 0.37 at 12 months). Comparing the intercept variance and the squared slope loadings at each assessment point reflects a shift from more between-person than within-individual differences to similar levels of between-person and within-individual differences in executive control over 12 months.

Delay of Gratification.

The intercept variance indicates significant between-person differences in levels of delay of gratification across the 12-month preschool period. An examination of slope loadings at each assessment shows the highest levels of within-individual variability at the first and last assessment (i.e., 3.31 at baseline and 0.90 at 12 months). A comparison of the intercept variance and the squared slope loadings at each assessment point shows more within-individual than between-person differences at the baseline assessment, with more between-person differences in delay of gratification at all subsequent assessments.

Regulation of Frustration.

The intercept variance indicates significant between-person differences in levels of frustration regulation across the 12-month preschool period. An examination of squared slope loadings at each assessment shows considerable increases in the relative amount of within-individual variability in the regulation of frustration (i.e., 0.19 at baseline to 2.64 at 12 months). Comparing the intercept variance and the squared slope loadings at each assessment point suggests that there is more between-person than within-individual differences in regulation of frustration across 12 months.

Emotion Knowledge.

The intercept variance indicates significant between-person differences in levels of emotion knowledge across the 12-month preschool period. An examination of squared slope loadings at each assessment suggests considerable decreases in the relative amount of within-individual variability in emotion knowledge (i.e., 8.64 at baseline and 0.30 at 12 months). Comparing the intercept variance and the squared slope loadings at each assessment point reflects a shift from more within-individual than between-person differences in emotion knowledge, to more between-person differences at all subsequent assessments.

Effects of Maternal ER Difficulties on Child ER Domains

Table 4 presents the associations between changes in maternal ER difficulties and changes in each child ER domain, controlling for child sex, child age-based estimated IQ, and family receipt of public assistance. All models demonstrated good fit. Standardized parameter estimates and standard errors for associations between change in maternal ER difficulties and between-person difference (intercept) and within-individual change (slope) in child ER domains are also presented.

Table 4.

Models testing associations between maternal changes in emotion regulation and between-person and within-individual differences in child emotion regulation by domain.

	χ2	df	p	CFI	RMSEA	Intercept (between-person)			Slope (within-individual)
						β	S.E.	p	β	S.E.	p
1. Executive Control (Dimensional Card Sort)	8.69	10	0.56	1.00	<.001	0.10	0.14	0.50	−0.35	0.16	0.03
2. Delay of Gratification (Gift Delay)	6.51	11	0.84	1.00	<.001	0.27	0.18	0.14	−0.23	0.22	0.29
3. Regulation of Frustration (Locked box/Knotted sack)	8.27	11	0.69	1.00	<.001	−0.38	0.62	0.54	0.49	0.42	0.25
4. Emotion Knowledge (Affective Perspective Taking)	3.17	10	0.97	1.00	<.001	0.14	0.09	0.10	−0.23	0.11	0.03

Note. Each row represents a separate model and significant effects are bolded. Coefficients are standardized parameter estimates (β) of the association between maternal changes in ER (slope) and between-person (intercept) and within-individual (slope) differences in child ER domains. All models control for the effects of child sex, child age-based estimated IQ, and family receipt of public assistance.

Executive Function.

Changes in maternal ER difficulties were unrelated to between-person differences in child executive function. There was a significant association between change in maternal ER difficulties and within-individual change in child executive function such that improvements in maternal ER difficulties predicted increases in child executive function.

Delay of Gratification.

There were no significant associations between changes in maternal ER difficulties and between-person differences or within-individual change in child delay of gratification.

Regulation of Frustration.

There were no significant associations between changes in maternal ER difficulties and between-person differences or within-individual change in child regulation of frustration.

Emotion Knowledge.

Changes in maternal ER difficulties were unrelated to between-person differences in child emotion knowledge. There was a significant association between changes in maternal ER difficulties and within-individual changs in child emotion knowledge whereby improvements in maternal ER difficulties predicted increases in child emotion knowledge.

Effects of Child ER Domains on Internalizing and Externalizing Problems

Table S1 presents the associations between between-person differences and within-individual changes in child ER domains and teacher-reported child internalizing and externalizing problems, controlling for teacher-reported problems at baseline, child sex, child age-based estimated IQ, and family receipt of public assistance. There was a significant effect of within-individual change in child executive function on teacher-reported externalizing problems. Additionally, there was a significant indirect effect whereby improvements in maternal ER difficulties predicted within-individual increases in child executive function and increases in child executive function predicted fewer teacher-reported externalizing problems at 12-month follow-up (Figure 1). There were no other significant effects on teacher-reported internalizing or externalizing symptoms.

Figure 1.

Model testing the indirect association between change in maternal emotion regulation and teacher-reported externalizing problems via within-individual change in child executive control. Note. Overall model fit was good (χ2 (21) = 17.27, p>.05; RMSEA = <.001; CFI = 1.00). Standardized (β) coefficients are shown for significant paths. Non-significant paths (i.e., with the intercept (between-person differences)) are not shown. Model controls for the effects of child sex, child age-based estimated IQ, and family receipt of public assistance. *p < .05; **p < .01.

Discussion

This study embarks on the ambitious task of applying a multidimensional RDoC framework to the development of ER. While widening the RDoC lens to include an emphasis on environmental and developmental influences, we also sought to sharpen the focus by illustrating how and when this can be accomplished using RDoC domains in this proof-of-concept design. The current study utilized repeated assessments across 12 months during preschool to demonstrate significant between-person differences and within-individual changes across several child ER domains. Additionally, it leveraged an RCT targeting maternal ER difficulties to test the causal impact of this known environmental influence on within-individual changes in preschoolers’ ER. Preliminary results suggest that treatment-driven improvements in maternal ER difficulties correspond with steeper growth in preschoolers’ executive control and emotion knowledge. Moreover, growth in child executive control predicted fewer teacher-reported externalizing problems at 12-month follow-up, even after accounting for baseline levels of externalizing problems. Here, we discuss these findings as a proof of concept to highlight how repeated assessments within an experimental intervention design afford the unique opportunity to test environmental mechanisms of change in developmental processes. In line with the goals of this special issue, we emphasize the importance of integrating environmental and developmental dimensions into the existing RDoC framework and offer specific recommendations about how to do so.

Charting Within-Individual Change in Developmental Processes

The use of longitudinal growth models, here the free curve slope intercept (FCSI) model, capitalized on our repeated assessment design and allowed us to parse meaningful between-person differences and within-individual change in four constructs associated with ER: executive function, delay of gratification, regulation of frustration, and emotion knowledge. Importantly, our results showed different patterns of within-individual variability whereby some constructs showed relatively higher levels of within-individual differences at baseline (i.e., 3 years of age; delay of gratification, emotion knowledge) while other constructs showed increases in within-individual differences over time (i.e., 4 years of age; executive function, regulation of frustration). These findings echo previous work demonstrating that functioning within these domains may develop along different timescales (Howse, Calkins, Anastopoulos, Keane, & Shelton, 2003; Lengua et al., 2015) and underscore the importance of charting multiple components of ER simultaneously. Additionally, results point to potential “turning points” or accelerated rates of growth that may vary by construct or domain of interest. This is consistent with the broader developmental psychopathology perspective that emphasizes the potential for non-linear growth, and discontinuity of change (Cicchetti & Rogosch, 1999; Rutter, 1996; Wood & Jackson, 2013). Taken together, this work highlights the utility of using latent growth modeling to capture meaningful within-individual change that is overlooked when using more traditional between-person models (Cole & Jacobs, 2018; Kim-Spoon & Grimm, 2016; Wood, 2011).

Our emphasis on modeling within-individual change is certainly not new (e.g., Baltes & Nesselroade, 1979), and neither is the call to harmonize developmental theory with methodological and analytic approaches (e.g., Cicchetti & Rogosch, 1999). However, the field has yet to routinely utilize within-individual approaches in practice, despite its consistency with theoretical conceptualizations of development and clear etiological implications. Expanding the RDoC to explicitly define development as a within-individual process would significantly advance research in this area by creating a unified framework and addressing perennial methodological issues that have limited rigor and reproducibility. Specifically, providing a structure that specifies the timing and frequency of assessments needed to capture within-individual change in developmental processes is needed (Fargas-Malet, McSherry, Larkin, & Robinson, 2010; Hopwood, Bleidorn, & Wright, 2021). The utility of this unified structure necessitates validating repeated assessments of developmentally and culturally sensitive domain-specific constructs to ensure our measurements capture the same underlying developmental phenomena over time. Reliable assessments of within-individual change would enhance our etiological understanding and allow for a more accurate characterization of causal effects, enriching our evaluation of known environmental influences on the development of ER.

Testing Environmental Influences as Causal Mechanisms of Within-Individual Change

By utilizing a prospective, experimental design, the current study was able to empirically test the influence of changes in maternal ER difficulties on within-individual change in child ER. To our knowledge, this proof-of-concept study was the first to utilize an RCT designed to link treatment-driven improvements in maternal ER difficulties with within-individual growth in child ER. Specifically, improvements in maternal ER difficulties predicted increasing trajectories of both child executive function and child emotion knowledge across 12 months during the preschool period. These findings are in line with core developmental psychopathology principles which conceptualize context as intricately intertwined with development (Sroufe & Rutter, 1984), and suggest that fully understanding (and quantifying) developmental change requires systematic manipulation of environmental influences (Garber & Bradshaw, 2020). Further, results demonstrate that maternal-driven growth in child executive function indirectly diminished externalizing problems across the same 12-month period. These findings shift our focus on maternal ER difficulties as a risk factor, to maternal ER difficulties as a malleable target for early intervention that promotes resilience in at-risk children, in line other emerging work in this area (Havighurst et al., 2013; Kehoe et al., 2020).

Notably, the influence of change in maternal ER difficulties on child ER was restricted to the RDoC defined cognitive (executive control) and social (emotion knowledge) systems. This is in line with research documenting the impact of parental influences on children’s attention deployment and set shifting (Fay-Stammbach, Hawes, & Meredith, 2014) as well as on children’s enhanced knowledge and awareness of emotions (Thompson et al., 2020). However, we failed to detect a link between change maternal ER difficulties and positive (delay of gratification) or negative (frustration tolerance) valence systems. This could be related to the small sample size and reduced power to detect effects. It is also possible that our measurement of these child ER constructs was affected by repeated administration, highlighting the need for future research focused on examining within-individual construct validity over time. Additionally, it is helpful to consider these findings within the context of the broader literature on the development of ER in children. Specifically, delay of gratification and regulation of frustration reflect more temperamental (impulsivity or negative reactivity) or phylogenetically older developmental processes (Howse et al., 2003; Lengua et al., 2015; Thomas, Letourneau, Campbell, Tomfohr-Madsen, & Giesbrecht, 2017). While these processes have shown malleability over time (Kiff, Lengua, & Zalewski, 2011; Slagt, Dubas, Deković, & van Aken, 2016), more significant or sustained environmental changes may be required to observe within-individual change. On the other hand, there may be cascading effects whereby maternal driven changes in child executive function may exert downstream effects on delay of gratification and regulation of frustration (Zelazo & Carlson, 2012), necessitating continued observations to capture this developmental sequencing. Finally, because our study focused on a global index of maternal ER difficulties and failed to consider mechanisms of change across parallel ER domains, our understanding of exactly how this intervention induced change in maternal ER difficulties is limited and, in turn, limits our understanding of the intergenerational transmission of ER.

Nonetheless, the current study highlights the ability of experimental intervention designs to more precisely quantify the causal impact of environmental influences on developmental processes across time. This work expands on traditional developmental studies that routinely describe and measure environmental factors but are rarely able to manipulate these influences. Importantly, this shift in design and focus directly aligns with the RDoC’s overarching goals, and enables us to translate mechanistic findings into specific prevention and intervention targets aimed at reducing risk for psychopathology (Cuthbert, 2014). However, the lack of prescriptive guidance within the RDoC framework prevents collective progress in this area and jeopardizes our ability to capture important explanatory variance in developmental processes of interest. Acknowledging the “environment” as an important dimension is necessary but not sufficient, and guidance for measuring core features of the environment is needed. This includes articulating domains of particular interest (e.g., family factors) and defining relevant constructs (e.g., parental psychopathology, parental influences) as well as including parallel levels of analysis with which these domains can be assessed (e.g., behavior, self-report), all of which would bolster the current RDoC framework. Further, to prevent a recapitalization of the “snap-shot” problem, these environmental influences should be assessed repeatedly, alongside repeated assessments of the developmental domain of interest. These extensions to the RDoC framework would result in greater precision and specificity about the impact of the environment on developmental processes, aiding in the identification of prevention and intervention targets.

Considering Developmental Influences

The current study focused on the preschool period, a developmental window known for rapid growth in ER development (Carlson & Wang, 2007; Denham et al., 2003). Perhaps unsurprisingly, our frequent, repeated assessments showed meaningful within-individual change across all child ER domains. The sensitivity of this developmental period was further underscored by heightened vulnerability to environmental influence (Cole et al., 2018), demonstrated here by the impact of maternal ER, and the link between changes in child ER and reduced risk for externalizing problems. Examining ER development within this key developmental period undoubtedly contributed to our ability to detect these associations and highlights the importance of explicit inclusion of a developmental dimension in the RDoC framework. While the ideal developmental window may vary – in both timing and tempo – depending on the developmental process of interest, incorporating a developmental dimension into the RDoC that explicitly defines when assessment and/or intervention are of utmost relevance would enhance our understanding of behavior and refine treatment targets.

Though we focused here on the preschool period, research clearly demonstrates that ER continues to develop across childhood and adolescence (Garber & Dodge, 1991; Zeman, Cassano, Perry-Parrish, & Stegall, 2006). Moreover, growing research supports the notion that environmental influences (e.g., parenting) continue to exert a critical influence on ER into adolescence (Klimes-Dougan et al., 2007; Laursen & Collins, 2009; Morris et al., 2017). For example, the transition to adolescence is another developmental period hallmarked by substantial neurobiological changes in systems underlying ER (Guyer, Silk, & Nelson, 2016; Steinberg & Morris, 2001), and this developmental period corresponds to the onset of various psychiatric disorders (Spear, 2009). Research suggests that the amplification of emotion during this period may necessitate prolonged emotional scaffolding (Guyer et al., 2016; Morris et al., 2017; Steinberg & Silk, 2002) and emerging work demonstrates that parental influences may exacerbate or mitigate ER difficulties (Byrd, Vine, Frigoletto, Vanwoerden, & Stepp, 2021) and emerging psychopathology (Vanwoerden et al., 2021). This further underscores the importance of incorporating what is known about developmental processes of interest into a well-defined RDoC dimension. Explicitly defining optimal windows of assessment and/or intervention would enable researchers to maximize their ability to detect and understand the impact of environmental influences on within-individual change.

Summary and Future Directions

In the past decade, the RDoC has increasingly taken hold as an alternative framework for the study of mental disorders (Cuthbert, 2014; Insel et al., 2010). Although environment and development are acknowledged as critical elements of the RDoC framework, the consideration of their influences are largely implicit. Scant guidance on how to incorporate these dimensions has created uncertainty for developmental psychopathologists who want to incorporate these two central tenets -- environment and development -- within the RDoC framework (Garber & Bradshaw, 2020). Here, in this proof-of-concept study, we demonstrate potential for the RDoC to significantly advance our understanding of environmental and developmental influences on within-individual change in developmental processes. Specifically, through experimental manipulation of the environment (maternal ER) we showed a causal effect on within-individual change in child ER during a key developmental window (preschool), enhancing our etiological understanding of this fundamental developmental process. Further, we found that these within-individual changes reduced externalizing problems during this same developmental window, pointing to maternal ER as an important prevention and intervention target. Thus, while the current study was notably limited by small sample size and the use of single constructs within domains of interest, our approach yielded results consistent with the overarching goals of the RDoC framework: advancing our understanding of behavior and identifying and refining targets for intervention. Here, we illustrate the ways in which environmental and developmental dimensions can be explicitly incorporated into the RDoC framework, paralleling conceptualizations of current dimensions. Future work incorporating these dimensions will hasten our progress toward better characterizing within-individual change in behavior and the extent to which that change is influenced by environmental factors.