Parental history of childhood maltreatment and child average RSA shape parent–child RSA synchrony

Abstract

We examined whether dynamic parent–child RSA synchrony varied by individual differences in child average RSA and parental history of childhood maltreatment (CM), which has been linked to parental behavioral and physiological dysregulation. We also examined whether RSA synchrony was curvilinear, reflecting homeostatic regulation. Synchrony was defined as the dynamic association between parent and child RSA reactivity (change relative to their own mean) within epoch across a challenging task. Eighty-three mother–preschooler and 61 father–preschooler dyads participated. State-trait modeling showed that RSA synchrony was curvilinear such that significant relations were only found at lower and higher child reactivity. Children’s higher task average RSA predicted maternal RSA augmentation and lower task average RSA predicted maternal RSA withdrawal, regardless of whether child reactivity in the moment was low or high, suggesting individual differences in child regulatory capacity were associated with dynamic maternal reactivity. When maternal CM history and child average RSA were both higher, mothers showed RSA augmentation. Father–child synchrony was not moderated by child average RSA but greater paternal CM history predicted fathers’ greater RSA withdrawal regardless of whether child RSA reactivity was low or high. Findings offer novel insights into the nature and meaning of RSA synchrony with parents at risk.

1 |. INTRODUCTION

Early childhood is a critical period for the development of regulatory skills, which are shaped in the context of parent–child interactions (Kim & Kochanska, 2012; Scholtes et al., 2020). During episodes of child distress, primary caregivers support the child’s efforts to regain a calm, balanced state through moment-to-moment responses on physiological, emotional, and behavioral levels, eliciting responses in the child and creating a dyad-specific “dance” of mutuality (Beeghly et al., 2016; Feldman, 2012). Over time, these repeated biobehavioral transactions support children’s positive developmental and self-regulatory outcomes (Beeghly et al., 2016). Specifically, synchrony of respiratory sinus arrhythmia (RSA) in parents and preschoolers seems to be a meaningful process related to child regulatory skills and developmental psychopathology (Lunkenheimer et al., 2015; Lunkenheimer, Tiberio, et al., 2018). We define synchrony as one subdimension of parent–child coregulation, one that includes positive synchrony (related change in the same direction), negative synchrony (related change in opposite directions), or asynchrony (no related change) of parent–child processes in close temporal proximity.

Parent–child RSA synchrony appears to be shaped by physiological regulatory capacities in the parent (Skoranski et al., 2017). Especially during challenging tasks, when parents must respond to children’s needs and distress, a critical question is how RSA synchrony is affected when parent self-regulation is compromised. A parental history of childhood maltreatment (CM) has been associated with difficulties in parenting, emotion dysregulation, and altered stress physiology (Lomanowska et al., 2015), all of which may affect physiological synchrony (Davis et al., 2018). When parents experience greater stress in the parenting role due to traumatic experiences around their own upbringing, parenting can be difficult and dysregulating (Lomanowska et al., 2015) and their children are more likely to have regulatory difficulties (Martinez-Torteya et al., 2014; Spieker et al., 2018). History of CM alters parents’ adrenocortical synchrony with infants (Fuchs et al., 2016) and preschoolers (Fuchs et al., 2017), but it is yet unclear whether and how it affects parent–child RSA synchrony.

Additionally, we lack knowledge about how individual physiological regulatory capacities in the child, both isolated and in tandem with parental history of CM, shape parent–child RSA synchrony. Higher resting RSA is commonly linked with better adjustment and regulatory capacities and is theorized to support social engagement (Holzman & Bridgett, 2017; Porges, 2007). For children with higher resting RSA and their parents, coregulation and thus RSA synchrony may be easier to accomplish and may even buffer potentially adverse effects of a parental history of CM (Rousseau et al., 2020).

RSA synchrony is usually examined assuming linear associations between parent and child RSA (Fuchs et al., 2021; Palumbo et al., 2017). However, linear models may fall short modeling coregulatory processes, as they disregard that parent–child dyads could physiologically coordinate with each other to maintain an optimal dyadic-level baseline, reflecting homeostatic regulatory processes (Feldman, 2012; Saxbe et al., 2020). Direction and strength of RSA synchrony could be subject to change, as stronger responses in one partner may provoke stronger responses in the other, or both partners may be in a homeostatic state where regulation is not necessary (Girme, 2020). However, as of yet, nonlinear associations have not been examined in parent–child RSA synchrony.

To address these gaps, we examined parent–child RSA synchrony with a special focus on how parent risk factors (i.e., their history of CM) and child physiological regulatory capacities shaped parental RSA reactivity in the context of a challenging interaction. In other words, we examined how child RSA reactivity (“state RSA”) dynamically predicted parent state RSA in each time segment over the course of a challenging task and whether this synchrony was moderated by factors expected to influence parent reactivity, namely parental history of CM and child average RSA during the task, that is, individual differences in child physiological regulation. By accounting for both child state and average RSA, we were able to examine their interrelations and unique contributions to RSA synchrony. We also explored whether relations between child and parent state RSA were curvilinear in nature. Further, we examined these processes in both mother–child and father–child dyads given the lack of research on RSA synchrony with fathers.

2 |. RSA AS A MARKER OF REGULATORY CAPACITIES

RSA is an index of the parasympathetic nervous system (PNS) that reflects variability in heart rate associated with the respiration rate (Porges, 2007). Normative and adaptive RSA functioning is commonly described in terms of higher resting RSA, which is thought to reflect greater capacity to respond to challenge, regulate stress and emotional arousal (Holzman & Bridgett, 2017; Thayer & Lane, 2000), and engage with the social environment (Porges, 2007). In contrast, lower resting RSA may reflect general PNS hypoactivation, related to psychopathology and early adversity exposure in both adults (Sigrist et al., 2020) and children (Beauchaine et al., 2019; Holochwost et al., 2020; Patriquin et al., 2015).

When situational stressors are present and mobilization of resources is required, for example when parents are prompted to support their children in challenging laboratory tasks, mild-to-moderate decreases in RSA (withdrawal) are commonly observed (Kahle & Hastings, 2015; Ostlund et al., 2017; Shih et al., 2019). Atypical RSA responses to challenge such as laboratory stressors include RSA increases (augmentation), which suggest disengagement, avoidance, and/or hyporeactivity (Beauchaine, 2001), as well as excessive RSA decreases, which suggest heightened regulatory effort or hyperreactivity (Porges, 2007). Whereas prior research relatively consistently suggests that lower resting RSA represents a risk factor and indexes lower regulatory capacities (Beauchaine et al., 2019; Holochwost et al., 2020; Patriquin et al., 2015), the literature focusing on the adaptiveness of RSA reactivity is less clear, for example, reporting different results depending on sample type, for example, clinical versus healthy or community samples, and type of task, for example, whether tasks challenge social, cognitive, sensory, and emotional domains (Bush et al., 2011; Graziano & Derefinko, 2013; Obradović et al., 2011). Considering the specific context is of particular importance since task differences may be associated with differences in normative RSA reactivity. For example, social and sensory challenges may be associated with RSA withdrawal, whereas cognitive and emotional challenges may be associated with RSA augmentation relative to resting RSA (Bush et al., 2011).

3 |. PARENT–CHILD RSA SYNCHRONY

In affective and behavioral synchrony research, higher parent–child synchrony of positive behaviors has been linked with positive child outcomes (Lunkenheimer et al., 2020; Woltering et al., 2015). However, the adaptiveness of parent–child RSA synchrony is still not fully understood. Some studies demonstrate positive RSA synchrony is normative and adaptive in community samples, associated with supportive parenting and better child self-regulation (Lunkenheimer et al., 2015; Skoranski et al., 2017), and that it becomes negative or asynchronous at higher risk, for example with maltreating mothers (Creaven et al., 2014; Lunkenheimer, Busuito et al., 2018; Lunkenheimer et al., 2019) or dyads with higher psychopathology symptoms (Lunkenheimer et al., 2015; Lunkenheimer, Tiberio, et al., 2018; Suveg et al., 2019). However, there is also evidence suggesting negative synchrony may be adaptive for children at higher risk, for example when parents have clinical diagnoses (Merwin et al., 2018; Suveg et al., 2019) or poorer emotional regulation skills (Creavey et al., 2020). In such cases, children may be protected by not attuning to parents who have compromised self-regulation (Lunkenheimer, Busuito, et al., 2018). Thus, whether synchrony is adaptive or not may depend on the nature, degree, and source of risk. In the present study, it was assumed that parent history of CM confers greater risk, and that RSA synchrony may be negative or asynchronous with higher levels of parental history of CM, analogous to findings in maltreating families (Creaven et al., 2014; Lunkenheimer, Busuito et al., 2018; Lunkenheimer et al., 2019).

Thus far, most studies on RSA synchrony focus on mothers. Yet preliminary evidence shows that father–child dyads demonstrate second-by-second synchrony in positive affect, that father–child affective synchrony positively predicts child outcomes, and that father–child and mother–child affective synchrony patterns may be distinct (Feldman, 2003; Lunkenheimer et al., 2020). However, few studies have investigated father–child physiological synchrony, and the majority that do tend to address endocrinological processes during infancy or adolescence (e.g., Saxbe et al., 2017). As for cardiovascular processes, a recent study showed that children’s externalizing symptoms were related to both mother– and father–preschooler RSA synchrony, and also that father–child RSA patterns were particularly associated with positive affect expressed during the interaction (Lunkenheimer et al., 2021). This finding may point to an important role of positive affect in father–child RSA synchrony. Further, some preliminary results with older children suggest father–child synchrony is present but may vary by child age, assessment, and individual and dyadic characteristics (Waters et al., 2020). For example, in one study examining pre-ejection period synchrony in 7- to 11-year-olds and their parents, synchrony was found in both mother– and father–child dyads, but directionality differed such that children were synchronized to their mothers, whereas fathers were synchronized to their children (Waters et al., 2020). Other studies with adolescents suggest no father–child RSA synchrony is present (Li et al., 2020; Roman-Juan et al., 2020). More data is needed to understand RSA synchrony in father–child dyads and to examine the parent and child factors that explain variations in these processes.

Studies examining child resting RSA or RSA reactivity during parent–child interaction increasingly point to the importance of testing nonlinear models of RSA regulation over time (Miller et al., 2013). For example, both atypically high and low resting RSA seem to have a negative relation with children’s social and regulatory outcomes (Miller et al., 2017; Zhang & Wang, 2019). Although most prior RSA synchrony research has been based on linear modeling (Fuchs et al., 2021; Palumbo et al., 2017), there is an increasing effort to promote the use of nonlinear and dynamic modeling techniques (Helm et al., 2018). According to theory, we expect parent–child dyads to coordinate their physiology to maintain an optimal dyadic-level baseline (Feldman, 2012; Saxbe et al., 2020). However, linear models may fall short in modeling coregulatory processes as they imply consistency in direction and strength of synchrony (Girme, 2020). Nonlinear modeling allows to test whether synchrony may be stronger when one dyadic partner shows more marked physiological responses, which may also lead to stronger regulatory responses in the other partner in order to restore some homeostatic state (Girme, 2020; Saxbe et al., 2020). It further allows one to examine whether synchrony may be weaker when a dyad has reached said homeostatic dyadic state (Girme, 2020). As nonlinear modeling of RSA synchrony may better match theoretical conceptualizations of coregulatory processes, and an alignment of conceptualizations and methods is vital going forward, nonlinear models of RSA synchrony may also resolve some inconsistencies in prior literature (Creaven et al., 2014; Palumbo et al., 2017). Thus, by examining curvilinear effects, we examine the conceptualization of homeostatic processes in RSA synchrony that allow for variation in the presence, direction, and strength of synchrony depending on the intensity and direction of physiological responses in parent and child (Girme, 2020; Saxbe et al., 2020).

Lastly, prior work on RSA synchrony has highlighted the importance of moving away from static measures of synchrony that may mask the dynamic nature of RSA synchrony over time (DePasquale, 2020). We implemented multilevel intradyad dynamics modeling, also called state-trait modeling (Bolger & Laurenceau, 2013; Shanahan et al., 2014), which allows us to examine how dynamic changes in one individual’s state RSA (i.e., increases or decreases with respect to that individual’s average RSA), may be related to dynamic changes in their partner’s state RSA (i.e., increases or decreases with respect to the partner’s average RSA). It further allows us to parse effects of trait-like between-person differences in child regulatory functioning over the course of a task (average RSA) from the effects of RSA reactivity in the moment on parent–child RSA synchrony.

4 |. AVERAGE RSA AS A MODERATOR OF RSA SYNCHRONY

A prominent advantage of state-trait modeling is the independent examination of dynamic, in-the-moment state RSA associations and the association between each individual’s average RSA across the time series of a given interaction (Bolger & Laurenceau, 2013; DePasquale, 2020). Parsing the effects of dynamic and average associations between parent and child RSA allows us to account for whether child average RSA in the context of parent–child interactions, including challenging interactions, may be relevant in shaping RSA synchrony. Recent findings have suggested that child average RSA is highly correlated with resting RSA and can thus be interpreted as trait-like RSA functioning and an index of children’s regulatory capacity in the parent–child interaction context (Fuchs et al., 2021; Lunkenheimer et al., 2021).

This contextualized average RSA is important because lower average RSA is thought to represent a risk factor in children linked with regulatory difficulties and psychosocial maladjustment such as mental health problems (Beauchaine et al., 2019; Holzman & Bridgett, 2017). Thus, higher average RSA may be a buffering factor: Due to their higher regulatory capacities, children with higher average RSA may be better equipped to interact with the social environment and to regulate stress and emotion (Porges, 2007). Additionally, parental responses to changes in child state RSA may differ depending on children’s general regulatory capacities. For example, parents of children with higher average RSA may have less difficulties establishing synchronous coregulatory patterns with their children if higher average RSA reflects better regulation (Rousseau et al., 2020). Alternatively, parents of children with lower average RSA may be even more attuned to their children due to their higher needs for external regulation. Second, recent findings suggest that average RSA constitutes a biological susceptibility to environmental influences, influencing how children react and respond to negative or positive qualities of the caregiving environment (Eisenberg et al., 2012; Richardson et al., 2019). For example, children with higher average RSA may be more sensitive to external influences such as parent physiology and behavior than children with lower average RSA (Eisenberg et al., 2012; Richardson et al., 2019). Although continued work is needed to better understand its role in dyadic RSA, recent studies showed that child average RSA interacted with parental psychological distress and child externalizing symptoms to shape parent–preschooler RSA synchrony (Fuchs et al., 2021; Lunkenheimer et al., 2021), underscoring its relevance for RSA synchrony.

5 |. HISTORY OF CHILDHOOD MALTREATMENT, RSA, AND RSA SYNCHRONY

A history of childhood maltreatment (CM) refers to one’s experiences of abuse and/or neglect in childhood and adolescence. Adverse consequences of CM have been shown to last into adulthood and seem to be observable in parenting situations (Alink et al., 2019). A parental history of CM may shape parent–child RSA synchrony via several potential pathways, including parent regulatory functioning in behavior and physiology (Gruhn & Compass, 2020; Sigrist et al., 2020) and parent–child interaction quality (Vaillancourt et al., 2017). Firstly, CM disrupts healthy behavioral and physiological self-regulation, for example, through promoting the use of avoidance or suppression of emotions as regulatory strategies (Gruhn & Compass, 2020). Impaired emotion regulation in children could thus, in adulthood, affect the ability to coregulate with one’s children (Suveg et al., 2019). Further, a recent meta-analysis showed that particularly in older samples, severity of CM was significantly associated with lower resting RSA, suggesting dysregulated RSA functioning in parents with a history of CM (Sigrist et al., 2020). There is also evidence that a history of CM may be associated with blunted RSA (hyporeactivity) in response to caregiving-related stressors in parents (Oosterman et al., 2019) and that a maternal history of neglect specifically may be related to RSA hyperreactivity (Buisman et al., 2019). Recent work has started to extend this research to children of parents with CM history, suggesting there may be intergenerational transmission of altered physiological stress regulation (Alink et al., 2019; Thomas et al., 2018). In line with this, studies show that parental history of CM is associated with RSA dysregulation in the next generation, specifically, children of parents with more severe CM seem to generally show lower average RSA (Glackin et al., 2020; Gray et al., 2017). Thus, RSA synchrony may be a pathway linking parent and child physiological and emotional dysregulation in context of parental history of CM.

Second, CM history has been shown to impair parent–child interaction quality (Savage et al., 2019; Vaillancourt et al., 2017), and seems to undermine parents’ ability to tend to and regulate children’s arousal and emotions (Lieberman et al., 2011). A parent’s CM history has further been identified as the strongest antecedent of their perpetration of child maltreatment, which is not only associated with significant alterations to parent resting heart rate (Van Ijzendoorn et al., 2020) and RSA functioning in parenting situations (Skowron et al., 2013; Wells et al., 2020) but also parent–child RSA synchrony (Creaven et al., 2014; Lunkenheimer, Busuito et al., 2018; Lunkenheimer et al., 2019). RSA synchrony seems to be negative (i.e., discordant) or absent in maltreating families (Creaven et al., 2014; Lunkenheimer et al., 2018, 2019). However, the question remains whether RSA synchrony differs in parents with a history of CM and their children.

6 |. PRESENT STUDY

Our aim was to examine whether child state RSA (i.e., reactivity) dynamically predicted parent state RSA (i.e., reactivity) in each time unit over the course of a challenging task, and whether these dynamic synchrony patterns were moderated by parent and child factors expected to influence parent physiological reactivity, namely child average RSA and severity of parent history of CM. We implemented multilevel intradyad dynamics modeling to test within-dyad (i.e., state RSA) and between-dyad (i.e., average RSA, history of CM) effects, which allowed us to model both the dynamic nature of RSA synchrony and individual contributions to RSA synchrony. We also explored curvilinear synchrony patterns and thus included quadratic state RSA parameters in analytic models. Further, as our main focus was to understand the association of child state and average RSA and parental CM with parent RSA reactivity, which has been shown to be linked with parenting (Connell et al., 2017; Wells et al., 2020), we chose parent reactivity as outcome variable. Certain covariates were included given their potential relevance for the present research questions. We included observed dyadic positive affect, indexing the affective climate and the degree of emotional arousal experienced in the task, as positive affect has been linked with RSA functioning (Oveis et al., 2009; Wang et al., 2013) and dyadic positive affect has been shown to play an important role in father–child RSA synchrony (Lunkenheimer et al., 2021). Parent resting RSA was included because individuals with higher baseline levels may have greater potential for reactivity (law of initial value) (Graziano & Derefinko, 2013). Lastly, we included time to account for the effects of elapsed task time on RSA reactivity (Bolger & Laurenceau, 2013). Our research questions were examined with respect to preschoolers (age 3) given the importance of RSA synchrony for children’s developing regulatory capacities in early childhood (Lunkenheimer et al., 2015; Lunkenheimer, Tiberio, et al., 2018).

Questions were tested in two steps. First, we tested how child state RSA was related to parent state RSA in the moment and how this relation was moderated by child average RSA. We hypothesized there would be a significant positive relation between child and parent state RSA, and that this relation would depend on child average RSA. However, given the novelty of this question we did not specify the direction of effects. Second, we added CM history severity as an additional moderator and hypothesized it would moderate RSA parent–child synchrony. We tentatively expected positive synchrony in lower-CM history parents and their children and negative synchrony in higher-CM history parents and their children as negative synchrony is usually found in at-risk samples and specifically in families with substantiated child maltreatment (Lunkenheimer, Busuito, et al., 2018; Suveg et al., 2019). To our knowledge no prior work has tested the curvilinear form of RSA, thus this part of the study was exploratory.

7 |. METHOD

7.1 |. Participants

Data was drawn from a three-wave longitudinal study on parent–child coregulation and familial risk. A total of 150 families participated, oversampled for low-income, child maltreatment risk based on Child Protective Services (CPS) involvement, or higher life stress on the Social Readjustment Rating Scale (Holmes & Rahe, 1967). Families were recruited via flyers at preschools, CPS, and family service agencies. Exclusion criteria included the inability to read, write, and speak English, cardiac conditions, and diagnosed developmental disability. Children (53% girls) were 2.5 years (SD = 0.14) on average at Time 1 and 3.03 years (SD = 0.11) on average at Time 2. Child race/ethnicity was reported as 65% White, 22% Latinx, 2% Black, 2% Native American, 8% Multiracial, and 1% Other or Unknown. Parent relationship status was 66% married, 13% living together, 9% separated or divorced, and 12% single. Mothers’ education ranged from junior high school to graduate level, with median educational level being an Associate’s degree. Average household income was $30,000 to $39,000 and the majority of families used government assistance (80%).

For the current study, data from Time 1 (questionnaires) and Time 2 (questionnaires, physiology, and affect) was analyzed. Eighty-three mother–preschooler and 61 father–preschooler dyads had complete RSA data and were thus included in analyses. Out of those, 10 mothers and 10 fathers had missing CTQ questionnaires, resulting in an overall analytic subsample of 71 mother–preschooler and 47 father–preschooler dyads with complete RSA, affect, and questionnaire data. Little’s missing completely at random (MCAR) test was not significant (χ² = 458.67, df = 460, p = .509) indicating that the missingness pattern was MCAR. Further, parent–child dyads with complete data did not differ from those with incomplete data with respect to any primary or control variables, nor sociodemographic variables except for child age (t(115) = −2.01, p < .05), with families with complete data having children slightly older than those with incomplete data. Families were excluded if parents could not read, write, and speak in English, if participants had a cardiac condition that would alter RSA data, or if the child had a diagnosed developmental disability.

7.2 |. Procedure

After approval by the Institutional Review Board, families were invited to participate and provided informed consent. Parents participated with one target child. At Time 1, questionnaire data regarding parental CM was collected and at Time 2 mother– and father–child dyads completed dyadic tasks during lab visits that were scheduled on different days and lasted 2 hours for mothers and 1 hour for fathers. Sessions began with electrode and respiratory belt application and a resting task in which parents and children watched a 3-minute nature video while resting RSA was collected. Stimuli for dyadic tasks were counterbalanced across parents. Mothers were reimbursed $135 and fathers were reimbursed $75 for participating.

7.3 |. Measures

7.3.1 |. Parent–child challenge task (PCCT)

RSA was collected during the PCCT (Lunkenheimer et al., 2017), a 10-minute collaborative and challenging task designed to capture change across baseline, challenge, and recovery conditions. For baseline (4 minutes), parents and children were asked to complete a puzzle and told that if they completed all three designs, they would win a prize. Parents were asked to use their words only and refrain from physically assisting their children. For the challenge condition (3 minutes), the experimenter interrupted the participants and informed them they only had two minutes left to finish. Lastly, in the recovery condition (3 minutes), children were praised, received the prize (art materials) regardless of completion status, and parent–child dyads were asked to play together with the art materials. PCCT has been shown to reliably elicit affective, behavioral, and RSA responses in parent–child dyads (Lunkenheimer et al., 2017).

7.3.2 |. RSA

RSA data was collected using the Mindware wireless electrocardiograph (ECG) MW3000A with an ECG sampling rate of 500 Hz. A crystal respiratory effort belt was placed below the diaphragm to monitor respiration. RSA was operationalized as high-frequency heart rate variability, the natural log of the variance of heart period within the frequency band related to respiration (0.12–0.40 for parents and 0.24–1.04 for children; Shader et al., 2018). Data was transmitted via wireless devices worn in backpacks and was processed and cleaned by trained research assistants using Mindware Heart Rate Variability 3.0 software. RSA was calculated for each 30-second segment of the 10-minute PCCT task (20 data points) and the 3-minute resting task (6 data points) for each individual. Segments with 10% or greater noise or artifact were excluded from analysis. As respiratory fluctuations have been shown to influence RSA in the context of a history of CM (Cyranowski et al., 2011), respiration rate was examined as a potential covariate; it was not correlated with maternal (r = .05, p = .600) nor paternal RSA (r = .06, p = .589) and was therefore not considered further.

7.3.3 |. Dyadic positive affect

Positive affect was coded by trained research assistants using the Dyadic Interaction Coding System (Lunkenheimer, 2009) for early parent–child interactions. Positive affect valence and intensity was rated second-by-second using Noldus Observer XT 10.0 software ranging from medium/high positive to neutral. Medium/high positive affect was coded, for example, when positive fluctuations in vocal tones such as higher pitch or sing-song rhythm were observed, or smiles with teeth showing, laughter or giggling was present. Low positive affect was coded, for example, when positive fluctuations in vocal tone or closed mouth smiles were present. Neutral affect was coded when there was a relative absence of facial expression and vocal tone was neither positive nor negative. Parents and children were coded separately. Affect was coded on a continuous second-by-second time scale, requiring coders to capture the same affect at the same window of time using a standard 3-s criterion in Noldus Observer to determine agreement. Dyadic positive affect was calculated as the total time in which one individual displayed low or medium/high positive affect, and at the same time the partner displayed either neutral or positive affect as well. Twenty percent of videos were double-coded and coders completed periodic tests for drift reliability. Average interrater agreement was 78%.

7.3.4 |. History of childhood maltreatment

Parental history of CM was assessed with the Childhood Trauma Questionnaire short form (CTQ-SF; Bernstein et al., 2003), a retrospective 28-item self-report inventory about childhood abuse and neglect with five subscales: emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect. Items are rated on a five-point Likert scale ranging from 5 (no history) to 25 (severe history of abuse/neglect). A total sum CM score was calculated for all subscales indexing CM severity (min 25, max 125). The CTQ has demonstrated strong psychometric properties in community samples (Scher et al., 2001). Cronbach’s α for mothers (α = .95) and fathers (α = .94) was excellent.

7.4 |. Analytic plan

Multilevel modeling (MLM) accounts for the nested nature of dyadic data and data interdependency (Gonzalez & Griffin, 2012), which can be measured by the intraclass correlation (ICC; Du & Wang, 2016). Unconditional means models and ICCs were computed to test whether MLM was fitting for our data structure. A higher ICC indicates higher variance explained by the grouping variable, the dyad (Garson, 2019). ICCs were 64.7% for mothers and 60.9% for fathers and unconditional means models showed significant random effects, confirming that MLM was appropriate (Garson, 2019).

Following Bolger and Laurenceau (2013), child RSA was split into a time-varying variable (child state RSA, within-dyad; WD) and a time-invariant variable (child average RSA over the course of the task, between-dyad; BD). Estimation of state RSA effects allowed for the examination of dynamic relations between parent and child RSA (Creaven et al., 2014; Suveg et al., 2019), whereas estimation of average RSA allowed for the examination of individual differences in regulatory functioning. Child average RSA was the grand mean-centered child RSA during the PCCT. Thus, higher or lower average RSA indicated that children had higher or lower average RSA compared to other children during the task. Child state RSA represented the deviation of child RSA from their average in each respective time segment of the task (Bolger & Laurenceau, 2013). Hence, a state RSA value of zero equaled child average RSA, the child’s mean RSA during the task. A negative RSA value indicated a decrease from that average in this particular moment, and a positive RSA value indicated an increase from that average in this particular moment.

A positive prediction of parent state RSA by child state RSA was interpreted as positive synchrony, meaning both children and parents increased or both decreased RSA. A negative prediction was interpreted as negative synchrony, reflecting that children increased and parents decreased RSA, or vice versa. Besides the ability to portray directionality, WD effects also account for magnitude of change: synchrony parameter values are lower for dyads wherein parents exhibit larger RSA changes and children exhibit smaller changes at a given moment compared to dyads with similar magnitude of change (Suveg et al., 2019).

To account for the effects of elapsed interaction time, time was included as a covariate (Bolger & Laurenceau, 2013). Model intercepts represented the model-implied mean at the first segment. Also, we included the covariates of average parent resting RSA to account for the law of initial value (Graziano & Derefinko, 2013) and dyadic positive affect to account for the affective climate of the interaction (Oveis et al., 2009; Wang et al., 2013). Covariates and history of CM were grand mean-centered so that values of 0 represented the respective sample averages. In order to examine potential nonlinear patterns of RSA synchrony we included both linear and quadratic child state RSA terms.

To test whether (1) child average RSA moderated RSA synchrony and (2) child average RSA and parental CM history jointly moderated RSA synchrony, four analytical models were performed: mother– and father–child Average RSA models and mother– and father–child Average RSA-CM models. It should be noted that most children completed the interaction tasks with mother and father, so that data for mother and father models was based on overlapping sets of children. Data were analyzed using the lme4 package in R (Bates et al., 2014). Random intercept models were fitted. Eighty-three mother–child and 61 father–child dyads with complete RSA data generated 1522 and 1160 observations for analyses, respectively. For CTQ analyses, we were able to analyze 73 mother–child (1334 observations) and 51 father–child dyads (963 observations). Simulation studies indicate that even with a sample size of 50, given a high ICC, reliable and valid estimates can be obtained with the present analytic approach (Du & Wang, 2016).

The first research question of child average RSA moderating the association between child and parent state RSA was tested by specifying separate models for fathers and mothers:

Within-dyad RSA: Level 1 child average RSA and child average RSA-CM history models

$$\begin{gathered} Y_{P}ij=\beta 0j_P+\beta 1.1j_{CP}\times \left({\text{RSAstate}}_{CP}\right)i+\beta 1.2j_{CP}\times \left({\text{RSAstate}}_{CP}^2\right)i \\ +\beta 2j_p\times (\text{time})i+{\epsilon }_{p}ij. \end{gathered}$$ (1)

Between-dyad RSA: Level 2 child average RSA model

$$\begin{gathered} \beta 0j_p=\gamma 00+\gamma 01\times ({\text{restingRSA}}_P\text{)}j+\gamma 02\times (\text{dyadicpositiveaffect})j \\ +\gamma 03\times \left({\text{averageRSA}}_{CP}\right)j+\mu 0j_p, \\ \beta 1.1j_{CP}=\gamma 04.1+\gamma 04.11\times \left({\text{averageRSA}}_{CP}\right)j, \\ \beta 1.2j_{CP}=\gamma 04.2+\gamma 04.21\times \left({\text{averageRSA}}_{CP}\right)j, \\ \beta 2j_P=\gamma 20. \end{gathered}$$ (2.1)

Equation 1 specifying parental RSA (Y_Pij) during PCCT at time i in dyad j includes an intercept for parent RSA specific to dyad j (β0j_P), a slope specific to dyad j representing within-person variation in child state RSA (β1.1j_CP), a slope for quadratic child state RSA (β1.2j_CP), a slope representing time (β2j_P), and a residual specific to time i for dyad j (ε_Pij). Equation 1 specifies between-dyad variations in the coefficients of Level 1 equations, including an intercept, child average RSA, parent average resting RSA, dyadic positive affect, and a residual component specific to each dyad. Between-dyad variations in slopes β1.1j_CP and β1.2j_CP are a function of intercepts and effects of child average RSA, and β2j_P is represented by a constant on Level 2.

For the second research question testing whether child average RSA and severity of parental CM history jointly moderated RSA synchrony, we again specified separate models for fathers and mothers with the same Level 1 Equation 1. Level 2 Equation 3 included the history of CM severity parameter:

Between-dyad RSA: Level 2, child average RSA-CM history model

$$\begin{gathered} \beta 0j_P=\gamma 00+\gamma 01\times {\text{(restingRSA}}_P\text{)}j+\gamma 02\times (\text{dyadicpositiveaffect})j \\ +\gamma 03\times \left({\text{averageRSA}}_{CP}\right)j+\gamma 04\times \left({\text{ctq}}_P\right)j \\ +\gamma 05\times \left({\text{averageRSA}}_{CP}\times ct{q}_P\right)j+\mu 0j_P \\ \beta 1.1j_{CP}=\gamma 06.1+\gamma 06.11\times \left({\text{averageRSA}}_{CP}\right)j+\gamma 06.12\times \left({\text{ctq}}_P\right)j \\ +\gamma 06.13\times \left({\text{averageRSA}}_{CP}\times {\text{ctq}}_P\right)j \\ \beta 1.2j_{CP}=\gamma 06.2+\gamma 06.21\times \left({\text{averageRSAA}}_{CP}\right)j+\gamma 06.22\times \left({\text{ctq}}_P\right)j \\ +\gamma 06.23\times \left({\text{averageRSA}}_{CP}\times {\text{ctq}}_P\right)j \\ \beta 2j_P=\gamma 20. \end{gathered}$$ (2.2)

Equation 3 specifies the between-dyad variations in the coefficients of the Level 1 equations, though variations in intercepts are now also a function of CM history severity and an interaction term between CM history severity and child average RSA. Thus, between-dyad variations in slopes β1.1j_CP and β1.2j_CP are a function of intercepts and the effects of child average RSA, CM history, and the CM history-average RSA interaction. β2j_P is again represented by a constant value on level 2.

8 |. RESULTS

Means, standard deviations, and correlations of study variables are shown in Table 1. There were no significant correlations between primary study variables and sociodemographic factors. Child average RSA was positively correlated with child resting RSA (during mother–child task: r = .85; p < .01 during father–child task: r = .79***, p < .01), maternal average RSA was positively correlated with maternal resting RSA (r = .86, p < .01), and paternal average RSA was positively correlated with paternal average RSA (r = .83, p < .01), supporting the use of average RSA use as a representation of individual differences in regulatory functioning.

TABLE 1.

Correlations, means, and standard deviations of study variables

Variable	M (SD)	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
(1) History of CM m	41.01(17.81)
(2) History of CM f	38.88(14.07)	0.28*
(3) Dyadic positive affect m	75.38 (53.73)	0.12	−0.07
(4) Dyadic positive affect f	70.37 (56.77)	0.03	0.09	0.16
(5) Resting RSA m	6.10(1.32)	−0.10	0.08	0.19	0.00
(6) Resting RSAf	6.13(1.32)	−0.05	−0.05	−0.11	0.19	0.22
(7) Average RSA cm	4.92(1.12)	0.05	−0.29*	−0.02	−0.03	0.18	0.03
(8) Average RSA cf	4.81 (1.01)	0.05	−0.05	−0.07	0.10	0.22	0.00	0.65***
(9) State RSA cm	0 (0.72)	0	0	−0.01	0.01	0.02	−0.01	0.02	0
(10) State RSAcf	0 (0.69)	−0.01	0	0	−0.02	−0.02	−0.01	0.01	0.02	0.07*

m = mother, f = father, cm = child with mother, cf = child with father; CM = (severity of history of) childhood maltreatment; resting RSA = mean RSA during the resting task; average RSA = mean RSA during the dyadic parent–child task; positive affect m/f = duration of positive dyadic affective states (in seconds) for mother–child and father–child interactions.

Note.

^*p < .05

^***p < .001. State RSA represents the deviation of the child's RSA from their average RSA. The mean state RSA is zero, reflecting that there was no change in child RSA from their average.

8.1 |. Child average RSA moderation models

Table 2 presents parameter estimates for maternal and paternal child average RSA models. There was considerable intercept variability between dyads as revealed by significant random intercept coefficients. For both mothers and fathers there was a positive relation between average resting RSA and parent state RSA during challenging parent–child interactions. Fathers but not mothers showed a positive association between state RSA and task time, suggesting they were more likely to show RSA augmentation as the task progressed.

TABLE 2.

Child average RSA Models

Child average RSA models Parameter	Child RSA predicting maternal RSA			Child RSA predicting paternal RSA
	Estimate (SE)	95% CI	p Value	Estimate (SE)	95% CI	p Value
Fixed effects
Timeγ20	−0.005 (0.003)	[−0.011,0.002]	.175	0.013 (0.004)	[0.005,0.020]	<01**
Resting RSAγ01	0.790 (0.042)	[0.708,0.871]	<.001***	0.737 (0.053)	[0.634, 0.840]	<.001***
Dyadic positive affectγ02	0.001 (0.001)	[0.708,0.871]	.470	−0.001 (0.001)	[−0.003,0.003]	.766
Child average RSAγ03	−0.019 (0.046)	[−0.109,0.071]	.685	0.036 (0.068)	[−0.094,0.167]	.600
Child state RSAγ041	0.013 (0.028)	[−0.042,0.067]	.651	−0.061 (0.034)	[−0.128,0.005]	.070†
Child state RSA2γ04.2	0.036 (0.026)	[−0.009, 0.080]	.116	0.002 (0.033)	[−0.062, 0.066]	.957
Child average RSA × state RSAγ0411	−0.002 (0.023)	[−0.047,0.043]	.937	0.036 (0.034)	[−0.031,0.104]	.290
Child average RSA × state RSA2γ04.21	0.052 (0.017)	[0.019,0.086]	<.01**	−0.054 (0.032)	[−0.118,0.009]	.093†
Random effects
Interceptμ0j	0.176 (0.419)	[0.341,0.490]	<.001***	0.234 (0.483)	[0.380, 0.576]	<.001***

Estimate = unstandardized coefficients, CI = confidence interval.

Note. A total of 1522 observations nested within 83 dyads for mothers, 1160 observations nested within 61 dyads for fathers.

^***p < .001

^**p < .01

^†p < .10. Parameter nomenclature is in accordance with Equations 1 and 2.1.

For mothers, the interaction between child average RSA and quadratic child state RSA was significant, suggesting the real-time association between child and maternal state RSA was curvilinear and dependent on child average RSA (Figure 1). Probing of this effect via Johnson–Neyman region of significance analysis revealed that child state RSA was associated with maternal state RSA when child average RSA was either very low (< −2.53) or moderate-to-high (> 0.17; child state RSA ranged from −3.07 to 3.88). Specifically, when children had higher average RSA, mothers displayed RSA augmentation whether children showed either low (withdrawal; negative synchrony) or high state RSA (augmentation; positive synchrony) in that time segment. In contrast, when children had lower average RSA, mothers displayed RSA withdrawal, whether their children showed either low (withdrawal; positive synchrony) or high state RSA (augmentation; negative synchrony) in that time segment. Thus, child average RSA was associated with maternal RSA reactivity in a particular direction and thereby with a specific synchrony pattern. There was no significant synchrony when child state RSA values were close to the child’s average RSA value (Figure 1). Main effects of child average and state RSA were not significant.

FIGURE 1. Child average RSA moderating mother–child RSA synchrony.

Note. This plot illustrates marginal effects. Child state RSA was associated with maternal state RSA when child average RSA was lower (←2.53) or moderate-to-higher (>0.17); SD = Standard deviation (+1 SD = 1.12, −1 SD = −1.12).

For fathers, neither child average RSA nor quadratic child state RSA were related to parent state RSA. However, linear child state RSA and the interaction between child average RSA and quadratic state RSA were marginally significant in negatively predicting father state RSA.

8.2 |. Child average RSA-CM history moderation models

Table 3 presents parameter estimates for maternal and paternal models of RSA synchrony moderated by child average RSA and parental history of CM severity. In line with results from child average RSA models, random intercept parameters were significant. Again, average resting RSA was related to parental state RSA values during the PCCT and fathers’ RSA was associated with task time.

TABLE 3.

Child Average RSA-CMModels

Child average RSA-CM models Parameter	Child RSA predicting maternal RSA			Child RSA predicting paternal RSA
	Estimate (SE)	95% CI	p Value	Estimate (SE)	95% CI	p Value
Fixed effects
Timeγ20	−0.007 (0.004)	[−0.014,0.000]	.054†	0.014 (0.004)	[0.005,0.0223]	<01**
Resting RSAγ01	0.806 (0.046)	[0.718,0.776]	<.001***	0.736 (0.061)	[0.622,0.850]	<.001***
Dyadic positive affectγ02	0.001 (0.001)	[−0.001,0.003]	.488	−0.001 (0.002)	[−0.004,0.002]	.720
Child average RSAγ03	−0.021 (0.051)	[−0.119,0.076]	.682	0.070 (0.081)	[−0.083,0.222]	.397
History of CMγ04	−0.002 (0.004)	[−0.009,0.005]	.619	0.008 (0.008)	[−0.008,0.023]	.367
Child average RSA × CMγ05	0.000 (0.004)	[−0.008,0.007]	0.940	0.004 (0.009)	[−0.013,0.021]	.687
Child state RSAγ061	0.041 (0.030)	[−0.017,0.099]	.165	−0.033 (0.038)	[−0.107,0.041]	.383
Child average RSA × state RSAγ0611	0.026 (0.026)	[−0.024,0.076]	.311	0.020 (0.041)	[−0.060,0.101]	.629
Child state RSA × CMγ06.12	−0.001 (0.002)	[−0.004,0.002]	.576	−0.001 (0.004)	[−0.009,0.006]	.750
Child state RSA × CM × average RSAγ06.13	−0.004 (0.002)	[−0.008, −0.001]	<.01**	−0.003 (0.004)	[−0.012,0.005]	.474
Child state RSA2γ06.2	0.003 (0.029)	[−0.053,0.059]	.909	−0.015 (0.036)	[−0.085,0.056]	.677
Child average RSA × state RSA2γ06.21	0.026 (0.023)	[−0.019,0.070]	.265	−0.060 (0.039)	[−0.137,0.017]	.129
Child state RSA2 × CMγ06.22	0.000 (0.002)	[−0.004,0.003]	.795	−0.007 (0.003)	[−0.014, −0.001]	<.05*
Child state RSA2 × CM × average RSAγ06.23	0.004 (0.001)	[0.001,0.006]	<.05*	−0.008 (0.004)	[−0.016,0.000]	.059†
Random effects
Intercept	0.185 (0.430)	[0.339,0.498]	<.001***	0.266 (0.516)	[0.386,0.606]	<.001***

CM = (severity of history of) childhood maltreatment; Estimate = unstandardized coefficients; CI = confidence interval.

Note. A total of 1334 observations nested within 73 dyads for mothers, 963 observations nested within 51 dyads for fathers.

^***p < .001

^**p < .01

^*p < .05

^†p < .10. Parameter nomenclature is in accordance with Equations 1 and 2.2.

For the maternal model, results showed significant three-way interactions for both linear state RSA and quadratic state RSA, suggesting curvilinear effects of child state RSA that differed by maternal history of CM severity. Because the inclusion of a quadratic term changes the interpretation of the linear term, we only probed the quadratic interaction. Simple slopes testing the effect of average RSA and severity of CM history showed that RSA synchrony was observed only in mothers with more severe history of CM (≥1 SD; 17.58) and children with higher average RSA (≥1 SD; 1.12) (Figure 2). When both factors were more severe/higher, mothers displayed RSA augmentation when children showed either RSA withdrawal or augmentation within time segment. Accordingly, more severe history of CM and average child RSA were related to positive synchrony (joint RSA augmentation) when children had higher state RSA, and negative synchrony (mother RSA augmentation, child RSA withdrawal) when children showed lower state RSA. In contrast, there was no RSA synchrony when mothers had mean or lower severity history of CM, nor when children showed mean or lower average RSA.

FIGURE 2. Maternal child maltreatment history and child average RSA moderating mother–child RSA synchrony.

Note. This plot illustrates marginal effects. Child state RSA was associated with maternal state RSA when child average RSA was higher (≥1 SD; 1.12) and maternal childhood maltreatment history was more severe (≥1 SD; 17.58); SD = Standard deviation.

For fathers, there was a significant two-way interaction between quadratic child state RSA and history of CM severity (Figure 3). Johnson–Neyman analyses illustrated synchrony was not significant for CM history scores up to 18.18 (range = −13.88 to 25.12). Thus, for fathers with less severe CM history and when children showed state RSA close to average levels, there was no significant RSA synchrony. However, fathers with more severe CM history showed lower state RSA (withdrawal) when children showed either lower state RSA (withdrawal) or higher state RSA (augmentation). Thus, higher paternal CM history was associated with positive synchrony (joint withdrawal) when children had lower state RSA, and negative synchrony (father withdrawal, child augmentation) when children had higher state RSA.

FIGURE 3. Paternal child maltreatment history moderating father–child RSA synchrony.

Note. This plot illustrates marginal effects. Child state RSA was associated with paternal state RSA when paternal childhood maltreatment history was more severe (≥18.18); SD = Standard deviation (+1 SD = 13.65, −1 SD = −13.65).

9 |. DISCUSSION

Parenting young children is challenging and requires self-regulation that may be compromised by risk factors such as history of CM, which is why it is important to focus on predictors of parental physiological reactivity (Lomanowska et al., 2015). The present work offered novel insights into maternal and paternal RSA reactivity and how it is shaped by child and parent factors, namely child average RSA and parental history of CM; in turn, these relations offered new information about the nature and meaning of parent–child RSA synchrony. Further, we expanded on prior research by testing both RSA reactivity in the moment and individual differences in child physiological regulatory functioning to better understand their contributions to dyadic RSA synchrony; we also addressed whether relations between parent and child reactivity were curvilinear in nature.

There are a few major takeaways from this study. First, dynamic relations between mother and child RSA reactivity were curvilinear and varied by high or low (but not mean) child average RSA; this suggests individual RSA components are informative for understanding dyadic RSA synchrony and that synchrony seems more evident when stronger RSA reactivity responses are observed. Second, individual differences in child task average RSA (both higher and lower) play an important role in maternal RSA reactivity, suggesting children’s trait-like self-regulatory capacities may guide maternal physiological responding as much as the child’s reactivity in the moment. We interpreted that maternal RSA withdrawal, associated with children’s lower average RSA, reflected mothers’ engagement and challenge, whereas maternal RSA augmentation, associated with children’s higher average RSA, reflected disengagement or the absence of challenge. Third, the combination of more severe maternal CM history and higher child average RSA was associated with maternal RSA augmentation regardless of the nature of child RSA reactivity, suggesting maltreated mothers were less likely to engage in tasks that required maternal support when children had higher regulatory capacity. Fourth, CM history had the opposite effect on fathers: Fathers with greater CM history showed RSA withdrawal regardless of child reactivity, suggesting their past maltreatment was associated with greater challenge and/or engagement with preschoolers. Our results suggest that RSA synchrony is not inherently adaptive or maladaptive and must be interpreted based on individual RSA functioning, risk factors, and task context.

9.1 |. RSA synchrony shows a curvilinear pattern

Both mother–child and father–child RSA synchrony show a curvilinear pattern such that significant positive or negative relations between child and parent RSA changes in the moment were only found at more pronounced degrees of reactivity, regardless of the direction (RSA augmentation or withdrawal) of that reactivity. There was no significant synchrony when children’s state RSA changes were close to zero, that is, close to their average RSA. Dyadic regulatory processes are particularly critical in the context of changes to the system, such as one partner changing from one state to another (Granic et al., 2016). Without change, there may be no need for active regulation. Dysregulation can be signaled by a deviation from an average or homeostatic state (Timmons et al., 2015). Prior research suggests healthy parent–child coregulation is characterized by both brief episodes of dysregulation or conflict and successful repairs of that conflict (Skowron et al., 2010; Tronick, 1989). Thus, our findings suggest RSA synchrony may be more meaningful when changes occur and less relevant when homeostasis is achieved, supporting the use of curvilinear models to test dynamic parent and child RSA relations. Emphasis on linear relations may be one reason for prior mixed findings regarding the direction of effects in parent–child RSA synchrony and the adaptiveness of those effects (Creavey et al., 2020; Girme, 2020). However, another possible explanation for the lack of significant synchrony when children were more constant in their physiological responding is a relative lack of variance with which to model associations. In future research, the inclusion of concurrent observed behavioral data could reveal more about the meaning of these moments and whether the absence of synchrony during episodes of “homeostasis” is due to a lack of need for regulation or related to methodological artifact.

9.2 |. Children’s individual differences moderate mother–child RSA synchrony

We found that child average RSA moderated dynamic relations between child and parent reactivity for mothers but not fathers. Mother–child RSA synchrony was significant only when children had lower or higher average RSA than the sample average. In this study, we treated average RSA during the task as a proxy for one’s physiological regulatory capacities, akin to the use of resting RSA in prior research, and our analyses indicated that average RSA was highly positively correlated with resting RSA. Generally, higher resting RSA is thought to be adaptive, and lower resting RSA maladaptive in relation to child outcomes (Beauchaine, 2001; Porges, 2007). Thus, lower average task RSA likely reflected children’s poorer baseline regulatory capacities whereas higher average task RSA likely reflected higher regulatory capacities to respond to challenge.

Interestingly, if children had higher average RSA overall and showed RSA augmentation or withdrawal in the moment, mothers showed RSA augmentation. If children had lower average RSA overall and showed RSA augmentation or withdrawal in the moment, mothers showed RSA withdrawal. Thus, mothers seemed to respond in sync with children’s average RSA (augmentation with higher RSA and withdrawal with lower RSA). This may imply that mothers generally respond to their experience of the child’s overall regulatory capacity and respond more when children tend to need support and less when they do not. Given that we examined an at risk-sample, two further interpretations are possible: Our results may suggest that mothers experienced the task of supporting their children as stressful, and thus the two patterns could reflect either disengagement or avoidance in mothers of children with higher regulatory capacities, and effortful withdrawal in mothers of children with lower regulatory capacities. Alternatively, it could imply that the protective factor of higher regulatory capacities in children may have prompted mothers, despite their risk status, to perceive children as resourceful enough to be able to handle the present task.

Further research is needed on how individual differences in child physiological functioning contribute to dyadic RSA synchrony, how they shape maternal RSA reactivity, and whether they are adaptive or maladaptive for child development. In cases where mothers showed RSA augmentation, we might also consider that children with higher average RSA did not perceive the task as challenging and therefore mothers may have not needed to to support them. However, we used a validated task shown to prompt RSA change in children at risk (PCCT; Lunkenheimer et al., 2017), and children’s average task RSA was significantly lower than their resting RSA, suggesting children were indeed challenged.

For father–child synchrony, there was only a marginal effect of child average RSA on child and parent state RSA relations. Given trend-level significance, we hesitate to interpret this finding; however, it was opposite from the effect for mothers, and the significant CM history finding for fathers (discussed below) was also in the opposite direction as that for mothers. Although we could not directly compare mother and father results with these particular analytic models and varying sample sizes, findings suggest that in future research with larger samples it may be worthwhile to explore whether RSA synchrony is fundamentally distinct for mothers and fathers.

9.3 |. Parent CM history significantly moderates parent–child RSA synchrony

In our higher-risk community sample, RSA synchrony was observed in dyads with greater parental CM history. For mother–child dyads, higher child average RSA and more severe history of CM seemed to constrain RSA synchrony such that it was only associated with maternal RSA augmentation regardless of the direction of child reactivity in the moment. RSA augmentation during challenge could reflect disengagement from or avoidance of a stressful stimulus (Beauchaine, 2001; Porges, 2007). A history of CM is related to avoidant and suppression emotion regulation strategies (Gruhn & Compas, 2020). Also, it is associated with hyporeactivity in response to caregiving-related stressors (Oosterman et al., 2019), and hyporeactivity in mothers seems to be associated with an avoidant attachment style in infants (Groh et al., 2019; Hill-Soderlund et al., 2008). Thus, it is possible that more severe maternal CM history may exacerbate the tendency for mothers to disengage from supporting their children when they have higher regulatory capacity, regardless of children’s needs in the moment. As CM history is a risk factor for hyporeactivity, future research could examine relations to child outcomes to verify if mother–child RSA synchrony characterized by maternal hyporeactivity is indeed maladaptive for child development. Interestingly, when CM history was taken into account, RSA synchrony was only present in mothers and children with higher regulatory capacities. Children’s higher regulatory capacities may have buffered adverse effects of maternal CM history such that children were generally better equipped to regulate stress and emotion adaptively (Porges, 2007). In consequence, those mother–child dyads may have had less difficulties establishing a coregulatory pattern (Rousseau et al., 2020). Alternatively, children with higher average RSA may have been more susceptible to maternal influences in affect, behavior, and physiology (Eisenberg et al., 2012; Richardson et al., 2019). The question remains whether a potential “buffering effect” or higher susceptibility on part of the child would be adaptive in this context (Suveg et al., 2019).

Father–child dyads also showed RSA synchrony when fathers reported more severe CM history. Fathers with more severe history of CM showed RSA withdrawal regardless of child reactivity, suggesting their higher CM may have been associated with greater engagement and experience of challenge with preschoolers. Thus, positive synchrony was observed when children showed RSA withdrawal and negative synchrony was observed when children showed RSA augmentation. Fathers with a more severe CM history may have perceived the task of regulating the child as more effortful, thus showing a pattern of RSA withdrawal in response to any changes in child RSA reactivity. Fathers also could have been less accustomed to engaging in challenging problem-solving tasks with their preschoolers in comparison to mothers.

The present findings also suggest other potential differences in mother–child and father–child RSA synchrony to be explored in future work. Mother–child synchrony was associated with variations in both child average and state RSA whereas father–child synchrony was associated with child state RSA only. Mothers may be sensitive to both trait-like characteristics and children’s needs in the moment, whereas fathers may be more heavily influenced by the present moment and context. Mothers are more likely to be primary caregivers and spend more time with the child and thus may be more sensitive to or adept at reading children’s needs (Petts & Knoester, 2018; Planalp & Braungart-Rieker, 2016). A recent meta-analysis supports the notion that mothers are more influenced by or responsive to child traits than fathers, who seem to be more heavily influenced by contextual facets of the interaction or their own personality traits (Klahr& Burt, 2014). Further, fathers’ RSA was related to concurrent positive affect whereas mothers’ RSA was not, underscoring that fathers’ reactivity may have been more dependent on the immediate context.

9.4 |. Limitations

Due to the characteristics of our higher-risk community sample, our findings may not generalize to lower-risk community or clinical samples. We chose a broad measure of CM history but did not disentangle effects of maltreatment subtypes and polyvictimization, which could be studied in future research. CM history was measured using subjective parent reports; despite validation of such reports (Hardt & Rutter, 2004), recent evidence suggests subjective and objective measures of CM history do not necessarily identify the same individuals (Danese, 2020). Although sample sizes were adequate for dynamic analysis of time series data, larger sample sizes would still have increased power to detect effects, especially with respect to father–child dyads. Also, interpretation of results could have benefitted from measures assessing subjective stress in parents during the lab assessment, but these measures were not available. Lastly, in order to highlight whether a lack of synchrony is truly associated with the dyads being in a homeostatic state and not with a lack of variance in RSA values, future study should add moment-to-moment observations of changes and dysregulation in the dyad and link these observations to moment-to-moment RSA assessments.

10 |. CONCLUSIONS

Our results highlight that parent–child RSA synchrony in early childhood is shaped by individual differences in child and parent factors, which may aid in understanding the operations and function of synchrony patterns. Synchrony with mothers appears to be linked with both trait-like regulatory capacities of the child and dynamic child RSA reactivity, whereas only the latter seemed to be linked with synchrony with fathers. They call for further examination of nonlinear models of synchrony and suggest that synchrony may be more evident when greater dynamic RSA reactivity levels are observed. When informed by prior theory and empirical evidence, inclusion of risk factors such as parental CM history may help to interpret the adaptive versus maladaptive nature of parent–child synchrony. CM history severity was associated with dynamic RSA augmentation in mothers and RSA withdrawal in fathers, which in turn was associated with synchrony patterns. These findings suggest that further investigation is warranted into the adaptiveness of synchrony patterns in higher-risk families. Lastly, while the present work informs research methodology and calls for further examination of parent–child RSA synchrony, it also supports the recent movement to consider biologically-informed prevention and intervention efforts to reduce parenting stress and restore or support adaptive physiological regulation in parents, especially in at-risk populations (Ha & Granger, 2016).

DATA AVAILABILITY STATEMENT

The data used herein is not publicly available.