Maternal dynamic respiratory sinus arrhythmia during toddlers’ interactions with novelty

Abstract

Maternal psychophysiological responses to toddlers’ distress to novelty may have important implications for parenting during early childhood that are relevant to children’s eventual development of social withdrawal and anxiety. Likely, these responses depend on intrapersonal, interpersonal, and contextual factors. The current study investigated the time course of respiratory sinus arrhythmia (RSA) across two laboratory novelty episodes, one low threat and one moderate threat, in 120 mothers of 2-year-old toddlers. Growth models tested context differences in and correlates of dynamic patterns of RSA. Dynamic patterns differed between tasks and according to mothers’ perceptions of and distress about toddler shyness. Thus, changes in mothers’ RSA across toddlers’ interactions with novelty seem to depend on the context as well as how mothers perceive and respond to their toddlers’ shyness.

1 |. INTRODUCTION

Parents must be able to regulate their emotional and physiological responses in order to optimally manage a child’s shyness, fear, inhibition, and distress to novelty (Root et al., 2016; Tiwari et al., 2008). Shyness, fear, inhibition, and distress to novelty are related, temperamentally based constructs indicative of children’s risk for anxiety development (Rubin et al., 2009). Individual differences in parents’ physiological responses to young children’s interactions with novelty are understudied but important given that the physiological processes that underlie parents’ emotions also fundamentally support the deployment of parenting behaviors (Dix, 1991). Thus, fully understanding antecedents of parenting behaviors associated with children’s anxiety risk requires an understanding of parents’ physiological responses to children’s interactions with novelty. Toward this end, the current study examined mothers’ respiratory sinus arrhythmia (RSA) during two contexts in which their toddlers interacted with novelty. We also investigated mothers’ perceptions of their toddlers’ shyness, as well as their own distress about toddler shyness, as correlates of mothers’ physiological regulation.

1.1 |. Maternal physiological regulation and parenting

A putative psychophysiological marker of individual differences in self-regulatory ability is RSA, a parasympathetic measure of the vagus-mediated variation in time between heart cycles (Porges, 1997, 2001). RSA is typically studied either at a trait level when measured during a resting baseline period, or at a state level when responding to tasks that may elicit regulatory efforts (Bush et al., 2011; Burt & Obradović, 2013; Obradović & Boyce, 2012). At a trait level, higher RSA tends to relate to greater regulatory capacity, while lower RSA has been associated with regulatory difficulties (Beauchaine, 2017).

Changes in RSA in response to a specific task are theorized to depend on the context and individuals’ perceived capacities for coping with that context. Porges’s polyvagal theory posits that RSA suppression (a decrease in RSA from baseline to task) indexes withdrawal of parasympathetic control over autonomic activity, which allows for an increase in heart rate and more efficient responses to the environment (e.g., Calkins, 1997; Porges, 2001, 2007). RSA suppression is thought to signify a readiness for an appropriate behavioral response in challenging situations (Bornstein & Suess, 2000; Calkins, 1997) and has been linked to higher levels of regulatory abilities (i.e., better emotion and physiological regulation). A failure to suppress RSA in challenging or distressing situations has been associated with deficits in regulation (e.g., Bornstein & Suess, 2000; Calkins, 1997).

On the other hand, RSA augmentation (an increase in RSA from baseline to task) accompanies increased parasympathetic control over physiological arousal in response to environmental stressors (Porges, 1997, 2001, 2007). Polyvagal theory (Porges, 2001, 2007) posits that continued or increased activation of the vagus nerve (manifested as maintenance or augmentation of RSA) in a safe environment prevents activation of the fight/flight system and supports constructive interaction and engagement. Thus, RSA augmentation would be expected to be adaptive in contexts that necessitate calm behavioral engagement rather than autonomic arousal (Beauchaine et al., 2008; Hastings, Nuselovici, et al., 2008; Hastings, Sullivan, et al., 2008; Mathewson et al., 2013; Stifter et al., 2011). Taken together, theory and empirical research suggest that suppression is most adaptive in situations that are distressing or threatening, whereas augmentation may be ideal in nonthreatening situations that call for attention and calm engagement (Bonanno & Burton, 2013; Brooker & Buss, 2010).

Challenge and threat theory (Blascovich & Mendes, 2000; Blascovich & Tomaka, 1996) similarly identifies context as an important determinant of whether RSA suppression or augmentation would be considered the expected response. According to this theory, the ratio of the context’s demands to the perceived individual’s resources determines whether the context is deemed a challenge (where resources outweigh demands) or a threat (where demands outweigh resources). Demands may be either conscious or unconscious; therefore, both manipulated aspects of laboratory contexts as well as individuals’ perceptions determine demands. Situations in which toddlers interact with novelty place inherent demands on parents, depending on the circumstances that determine their toddlers’ responses. Demands that vary could include both the overt features of the context as well as preexisting perceptions and expectations brought in by the parent.

Overt features that may increase demands on toddlers, and subsequently parents, include the stimulus being more unpredictable/less controllable by the toddler and the stimulus coming closer in proximity to the toddler. These overt features influence toddlers’ displays of negative emotion and bids for support (Buss, 2011), which place demands on parents (Kiel & Buss, 2012). In the current study, we chose two contexts (viewing a puppet show and an interaction with a friendly clown) that resemble interactions a toddler might encounter outside of the laboratory, such as at a public event or birthday party, and elicit individual differences in inhibition to novelty (Nachmias et al., 1996), but nevertheless differ in their predictability and proximity.

Factors that may directly influence perceived demands on parents include parents’ existing perceptions of their toddlers’ dispositional shyness, and how distressed parents feel by their toddlers’ shyness. Shyness present in toddlerhood relates to neural activity consistent with heightened fear responses that also characterize behavioral inhibition and heightened risk for anxiety (Poole & Schmidt, 2020; Rubin et al., 2009). Transactional models of parent–child relationships theorize that children’s dispositions shape parenting beliefs and behaviors (Sameroff, 2010). Following this, parents’ expectations for interactions in a given novel situation would be shaped by their preexisting evaluations of their toddlers’ shyness. Parents are theorized to be more emotionally reactive to situations involving novelty when they view their toddlers’ shyness as more dispositional, rather than situational (Rubin et al., 2009). Therefore, we would also expect physiological responding to relate to parent perceptions of their toddlers’ dispositional shyness. The extent to which parents’ perceptions of toddler shyness relate to their physiological responses to toddlers’ interactions with novelty would likely further depend on how distressing they find their toddlers’ shyness to be. Distress surrounding toddler shyness may result in parents being more vigilant toward its display and, as a result, change the extent to which perceptions of shyness relate to a physiological response. Previous work has shown that mothers’ physiological responses to infant crying relate to mothers’ own distress (Kim et al., 2016; Laurent & Ablow, 2012) as well as difficulties engaging in sensitive parenting (Joosen et al., 2013). In line with challenge and threat theory, parents who perceive their children as shy and find shyness distressing may perceive a situation as more threatening and less calm/engaging than a parent who does not view their toddler as shy or feel distressed about shyness.

Therefore, both polyvagal theory and challenge and threat theory would suggest that whether RSA suppression or augmentation is expected to occur depends on the context and the individual’s interpretation of that context. In line with these theories, we examined maternal RSA in two contexts (varying in demands) and tested mothers’ perceptions of their toddlers’ shyness and distress in response to toddler shyness as correlates of RSA.

1.2 |. Dynamic RSA

Traditionally, studies have investigated RSA using an average of values from smaller epochs across a task (i.e., mean task RSA) in comparison with a similarly aggregated baseline measure of RSA (Alkon et al., 2003; Brooker & Buss, 2010; Buss et al., 2005; Obradovic & Boyce, 2012; Obradović et al., 2010). Measuring RSA in aggregate, however, does not capture dynamic fluctuations in regulatory patterns across a situation (Brooker & Buss, 2010; Calkins et al., 2007; Larsen et al., 2008). Because the true time scale for changes in RSA is much shorter than the typical challenge task or baseline episode, a mean-level aggregate measure may be insufficient to accurately capture more nuanced patterns of fluctuation in RSA as individuals respond to environmental cues and demands (e.g., Larsen et al., 2008).

Advances in statistical approaches have allowed researchers to model RSA fluctuations on smaller time scales, readily capturing dynamic changes in RSA within episodes characterized by emotional challenges (Brooker & Buss, 2010; Cui et al., 2015; Miller et al., 2013, 2016). This approach models RSA across the time period of a task in a growth model, wherein the intercept estimates an initial RSA level, and the slopes associated with linear and quadratic time variables test whether and how RSA changes across the episode. According to polyvagal (Porges, 2007) and challenge and threat (Blascovich & Mendes, 2000) theories, to the extent that a task represents a challenge or threat, we would expect an increase or decrease in values across the episode, respectively. If the task has an immediate effect, this would be reflected in a higher intercept (reflecting augmentation) or lower intercept (reflecting suppression) as compared to the baseline value. We may see further increases or decreases as the episode progresses. To the extent that individuals experience recovery, or a transition back toward baseline values, we would expect a significant quadratic effect indicating a change in direction: from initial deviation from baseline occurring at the onset of the episode, to back toward baseline RSA toward the end of the episode. If individuals do not experience recovery during the task itself, we may expect a flat slope (non-significant linear and quadratic trends indicating maintenance of a particular value) or linear and quadratic components that lead to increases or decreases that do not change direction back toward the baseline value.

To date, dynamic approaches have primarily been deployed in child samples. Thus, although the current study focused on temporal patterns of RSA among parents, we drew on concepts first presented in the child literature. Studies with older children and adolescents (Cui et al., 2015; Miller et al., 2013), whose RSA patterns more closely resemble those of adults (Wallis et al., 2005), suggest that a higher intercept, negative linear trend, and positive quadratic trend relate to adaptive regulatory behaviors. In other words, RSA values started high at the beginning of the task, decreased initially following the onset of a stressor, and then increased back toward starting values. This pattern would suggest the task is stressful enough to evoke suppression and then recovery. Against this backdrop of the typical pattern, adolescents with regulation difficulties started with lower RSA values and had a flatter slope across time. Both the linear and quadratic components, which would be obscured though traditional averaging, were necessary to discern these group differences. A study of adults found that dynamic RSA in marital dyad members related to the quality of interpersonal interaction, while static RSA yielded no relation (Gates et al., 2015). Although this study used a different approach than we use, it does lend support toward examining dynamic RSA in adults.

The pattern expected among adults in a parenting context is unclear, as this population has not been studied in regards to dynamic RSA. If mothers followed a similar pattern as above, it may indicate that viewing one’s toddler during interactions with novelty are “threatening” enough to induce suppression that is then recovered. Deviations from this overall pattern may indicate regulatory difficulties. Alternatively, we might expect that interactions with toddlers during low-threat laboratory tasks may be more of a “challenge” than a “threat,” in which case we may expect initial augmentation followed by “recovery,” indicated as a downturn toward baseline. To the extent that the context or mothers’ expectations resulted in intensified versus diminished demands of the situation, we would expect change in one or more of these parameters. Challenge and threat theory (Blascovich & Mendes, 2000) suggests using a convergent validity approach wherein context and individual differences in evaluations are used to understand the nature of psychophysiological responding.

The literature requires a foundation for determining whether a dynamic approach might further clarify the expected patterns of mothers’ physiological regulation and influences of both context and mothers’ experiences on these patterns. To this end, we examined how maternal perceptions of toddler shyness and maternal distress responses to shyness related to dynamic RSA patterns. Further, we did so in consideration of the suggestion from polyvagal theory (Porges, 2001, 2007) and challenge and threat theory (Blascovich & Tomaka, 1996; Blascovich & Mendes, 2000) that patterns of RSA also depend on the context in which they are observed.

1.3 |. The current study

This investigation aimed to provide insight into mothers’ psychophysiological response to their toddlers’ distress to novelty. We focused on two contexts of comparable length that each involved a social component. Based on previous literature (Buss, 2011), we expected an interaction with a friendly clown to be somewhat higher in demands on mothers than watching an interactive puppet show. The clown requests direct interaction, is more unpredictable in her movements, and could presumably move close to the toddler and mother. The puppet show, on the other hand, allows the toddler “off” time while puppets converse/interact with each other, is predictably stationary based on the solid stage, and does not change in proximity to the toddler. Although neither post true “threat” to the toddler’s physical safety, they may differ in how the toddler or mother perceives “threat” to the toddler’s psychological comfort. Previous studies provide evidence that as situations increase in ambiguity, uncertainty, and direct requests for interaction, so too do toddlers’ behavioral and neural indicators of subjective threat (Brooker et al., 2016; Buss, 2011). Given that the two contexts under investigation resembled typical activities characterized by social interaction with one’s child, we hypothesized that the overall pattern of maternal RSA would be augmentation from baseline (the expected response to mild challenge), followed by recovery toward initial levels. We hypothesized that this pattern would be weaker (i.e., blunted) in the context presenting mothers with greater parenting demands (interaction with a friendly clown) than the more passive context (viewing a puppet show). We also hypothesized that this pattern would become increasingly blunted as mothers perceived their toddlers as more shy and reported greater distress responses to toddler shyness, putatively placing higher psychological demands on mothers.

2 |. METHOD

2.1 |. Participants

The current study comes from a larger investigation of 165 mother–toddler dyads. The present study was conducted according to guidelines laid down in the Declaration of Helsinki, with written informed consent obtained from a parent or guardian for each child before any assessment or data collection. All procedures involving human subjects in this study were approved by the Institutional Review Board at Miami University. Electrocardiogram (ECG) monitoring appropriate to the current study’s analyses was added after a number of participants (n = 24) had completed the study, so participants missing ECG were excluded. Other participants were excluded because they were siblings of other participants (n = 9), they did not complete the laboratory portion of the study (n = 8), or because of technical difficulties with the ECG recording (n = 4). Thus, analyses focus on the remaining 120 participants, who did not differ from excluded participants on demographic characteristics (toddler biological sex, household income, toddler or maternal age, race, or ethnicity; all ps > .10). Toddlers (50 female) were approximately 2 years of age (M = 26.85 months, SD = 2.06 months, Range = 23.49–33.05 months). Mothers ranged in age from 19.82 to 45.11 (M = 32.41, SD = 5.23) years of age at the time of study participation. As reported by mothers, 89.2% of mothers and 81.7% of toddlers were European American; 3.3% of mothers and 1.7% of children were Asian, Asian American, or Pacific Islander; 1% of children were Black or African American; 8.3% of children were multiracial; and 2% of mothers and 2% of children were reported as a different racial background; 1% of mothers and 1.7% of children were reported to identify as Latinx. On average, mothers were college-educated (M = 15.45 years, SD = 2.69, Median = 16 years of education), with 18% of the sample reporting a high school education or less. The mean and median household gross annual income of participants was in the $41–50K range, with 24% of participants reporting less than $20K/year.

2.2 |. Procedure

All procedures were approved by the campus Institutional Review Board. Mothers were recruited from birth announcements in local newspapers, through flyers at local establishments (e.g., pediatrician’s offices, daycare centers, farmers’ markets), and in person at a Women, Infants, and Children office. Some mothers had participated in a previous phase of the larger study (not discussed further) and were contacted directly about continued participation. Upon the parent expressing interest in participating, a research assistant scheduled a visit to the laboratory and sent the mother a consent form and packet of questionnaires to complete and bring to the visit. At the laboratory, a primary research assistant (E1) explained all of the procedures and provided time for the child to become comfortable with the laboratory setting. E1 then explained the wireless heart rate monitor and gave the mother privacy to apply electrodes, using the traditional 3-lead formation. The 3-lead (Lead II) formation is a standard alternative for the clinical standard 12-lead application. The 12-lead approach is unnecessary for studying heart rate variability and offers no novel information beyond the 3-lead setup (Pipberger et al., 1961). As such, the Lead II formation is suggested as the common standard for ECG/HRV data collection in research settings (BIOPAC Systems, 2016; Mindware Technologies, LTD, 2014). We used a modified formation, with electrodes placed on the chest rather than arms/legs. Mothers also wore a respiration belt. After checking for adequate reception of the ECG and respiration signals by a BIOPAC MP150 system and AcqKnowledge software, the mother was asked to sit calmly and breathe normally for a 5-min baseline recording. The toddler remained in the room so that the presence of the child did not differ between baseline and tasks. The toddler and mother then engaged in a variety of laboratory activities over the course of 2.5–3 h. Mothers continued wearing electrodes and belt for ECG and respiration recording throughout the laboratory visit. Tasks were video-recorded for later scoring.

The relevant tasks for the current study include Clown and Puppet Show episodes modeled after tasks from the Laboratory Temperament Assessment Battery (Lab-TAB; Buss & Goldsmith, 2000), and other studies in the literature (Buss, 2011; Nachmias et al., 1996). These tasks were specifically chosen because they have been shown to identify toddlers at risk for anxiety-spectrum problems (Buss, 2011) and yield toddler behavior more highly related to maternal perceptions of toddler temperament than tasks that elicited higher levels of distress (Kiel & Hummel, 2017). For all tasks, mothers were instructed to behave naturally (“However you typically would, or however feels right to you”). In Clown (approximately 5 min in length), E1 escorted the mother and toddler into an empty room. Shortly thereafter, a second female experimenter (E2) entered, dressed in a clown outfit, wig, and red nose. With a friendly and enthusiastic demeanor, the clown greeted and invited the toddler to play three 1-min games (blowing bubbles, catch with beach balls, musical instruments). After the games, the clown invited the child to help clean up her toys, after which she said good-bye and left the room. In Puppet Show (approximately 6 min in length), the mother and child entered the room, and in the opposite corner was a small wooden stage with a curtain, behind which E2 controlled two plush puppets that engaged in a friendly manner and invited the toddler to interact with them for two, 1-minute games (catch, fishing). At the end of these games, the puppets invited the toddler to take a sticker from them. Tasks were presented in the same order for all participants, with an episode not designed to elicit negative emotions and a resting period in between (mean time between Clown and Puppet Show = 28.77 min, SD = 5.94 min). Mothers were paid $50 for their time, and toddlers received a small gift for their participation.

2.3 |. Measures

2.3.1. | Maternal RSA

Mothers’ ECG data were recorded at a sample rate of 1000 Hz using AcqKnowledge software. Text files of raw ECG data were imported into MindWare HRV Analysis 3.1.5 software and visually inspected and cleaned for artifacts and missing/ectopic beats offline. Log-transformed RSA values were then calculated for each 1-min segment of the baseline and 30-s segment of each episode from the respective interbeat interval (IBI) time-series filtered at the standard frequency of adult respiration (0.12–0.40 Hz). For baseline recording, we followed Task Force (1996) recommendations to analyze 1-min epochs of data. Shorter 30-s intervals were used for Puppet Show and Clown tasks to allow for more nuance in changing values, as is typical in studies of dynamic RSA (e.g., Brooker & Buss, 2010; Fisher & Woodward, 2014; Gates et al., 2015; Miller et al., 2013). Intervals missing more than 10% of beats were not analyzed. Values were averaged across the 1-min baseline epochs for the final variable of baseline RSA. Individual 30-s epochs were averaged separately within Clown and within Puppet Show for descriptive purposes, but individual epoch values were retained for growth modeling.

2.3.2 |. Maternal perceptions of toddler anxious behavior

Prior to the laboratory visit, mothers completed the Early Childhood Behavior Questionnaire (Putnam et al., 2006), a well-established survey measure of temperament for 18- to 36-month-old toddlers. The ECBQ has been found to show adequate internal and inter-rater reliability (Putnam et al., 2006). The current study focuses on the Shyness subscale (5 items, α = 0.77; e.g., “When approached by an unfamiliar person in a public place (for example, the grocery store), how often did your child cling to a parent?”). Items were averaged to yield the final measure of mother-perceived toddler shyness.

2.3.3 |. Maternal distress about toddler anxious behavior

Mothers also completed the Coping with Toddlers Negative Emotions Scale (CTNES; Spinrad et al., 2004). This measure presents parents with 12 hypothetical vignettes in which they are instructed to imagine their toddlers expressing specific expressions or behaviors indicative of particular emotions. To maximize relevance of this measure to laboratory procedures, we focused on the four vignettes describing the toddler in fear-eliciting contexts (i.e., going to the doctor or getting shots, getting nervous with a new babysitter, being left alone to go to sleep, getting stuck on playground equipment). For each of the four vignettes, mothers rated the likelihood that they have various responses, from which we focus on the Distress items (e.g., “Feel upset or uncomfortable because of my child’s reactions”). The four Distress items (α = 0.80) were averaged to yield the final variable of maternal distress.

2.3.4 |. Observed toddler shyness

Toddler shyness was rated by trained coders for the Puppet Show and Clown tasks. Using an existing coding scheme for these episodes as well as episodes presenting higher threat and uncertainty (Buss, 2011), coders scored toddlers’ overall level of shyness on a 5-point scale (1 = none; 2 = slight withdrawal, fidgety, reduced activity; 3 = tense in posture, hiding face, moderately withdrawn for first part of episode; 4 = very tense, inactive, very fidgety, frozen, avoidant for most of episode; 5 = extremely shy, frozen, totally avoidant or resistant of stimulus throughout entire episode). Different coders scored the two episodes. Coders were required to meet reliability (intraclass correlation coefficient [ICC] = 0.80) with a master coder before coding independently. The master coder double-scored approximately 20% of cases throughout coding to prevent coder drift. Final reliability (prior to resolving discrepancies) was found to be good for both Puppet Show (ICC = 0.96) and Clown (ICC = 0.85).

3 |. RESULTS

3.1 |. Preliminary analyses

3.1.1 |. Missing data

All 120 participants provided baseline RSA. No epochs of RSA could be calculated for nine Clown and 14 Puppet Show episodes because the signal was either too noisy or absent. These 23 cases included 18 mothers contributing one episode and five mothers contributing neither, the latter of whom could not be included in primary analyses. Individual epochs with missing RSA values did not exclude participants from analysis, as multilevel modeling can accommodate these missing values in estimating growth by using available observations within a participant (Heck et al., 2014; Snijders & Bosker, 2004). Eight mothers were missing either perceptions of toddler shyness or distress about toddler shyness; seven mothers were missing both. Observation of toddler shyness was missing for five Puppet Show episodes and six Clown episodes. Overall, this amounted to 6.67% of observations missing. Little’s Missing Completely at Random (MCAR) test, including primary variables and demographic information (maternal age, maternal education, household income, child gender, participation in earlier phase of study) suggested that the pattern of missing values did not significantly deviate from the MCAR pattern (χ²[249] =276.54, p =.111). Missing data were handled by running analyses using both available cases and multiple imputation (20 imputations, algorithm included primary and demographic variables) for analyses of mean-level differences in RSA. The pattern of results was identical across approaches, so analyses with available cases are reported. Missing data were handled at the lower level of multilevel models.

3.1.2 |. Descriptive statistics

We provide descriptive statistics for and correlations among primary variables in Table 1. We examined bivariate associations between demographic variables and average RSA values to determine potential covariates. RSA did not differ according to child gender, child age, maternal education, or household income, so these variables were not considered further. Maternal age at the time of study participation did relate to average RSA in the Puppet Show (r[102] = −.24, p = .015) and Clown (r[108] = −.21, p = .028) episodes. Therefore, maternal age was included as a covariate in growth models. On average, both Puppet Show and Clown were characterized by moderate shyness (Table 1). As a validation of our conceptualization as the Clown context as higher in perceived threat than Puppet Show, we performed a paired-samples t test in observed shyness scores between the episodes. As expected, shyness in Clown was significantly higher than in Puppet Show (t[111] =4.73, p < .001; Cohen’s d = 0.50).

TABLE 1.

Descriptive statistics and bivariate correlations for primary study variables

Variable	Mean (SD)	Range	n	2	3	4	5	6	7
1. Maternal RSA: Baseline	5.47 (0.98)	2.50–8.00	120	.76***	78***	−.06	−.18†	.01	−.11
2. Maternal RSA: Puppet Show	6.12 (1.00)	4.03–8.93	106	–	.76***	−.06	−.07	−.03	−.10
3. Maternal RSA: Clown	6.09 (1.00)	2.59–8.40	111		–	−.05	−.13	−.09	−.20*
4. Mother-perceived toddler shyness	3.80 (1.19)	1.00–6.60	109			–	.22*	−.07	.29**
5. Maternal distress about shyness	3.15 (1.48)	1.00–6.75	109				–	.20*	.13
6. Observed shyness: Puppet Show	2.03 (1.02)	1.00–5.00	115					–	.37***
7. Observed shyness: Clown	2.56 (1.00)	1.00–5.00	114						–

Note: RSA values represent averages across all epochs of their respective episodes. Maternal-perceived toddler shyness represents the average of items from the Shyness scale of the Early Childhood Behavior Questionnaire (Putnam et al., 2006) and can have a possible range of 1.00 to 7.00. Maternal distress about shyness represents the average of four items about maternal distress reactions to toddler shy or fearful behavior from the Coping with Toddlers’ Negative Emotions Scale (Spinrad et al., 2004) and can range from 1.00 to 7.00.

Abbreviation: RSA, respiratory sinus arrhythmia.

^†p <.10,

^*p <.05,

^***p <.001.

3.2 |. Mean-level difference in RSA

We used paired-sample t tests to determine whether average RSA values differed between each episode and baseline (Figure 1). Average RSA across the Puppet Show task (Mean = 6.12, SD = 1.00) was significantly higher than baseline RSA (Mean = 5.48, SD = 0.97; t[105] =9.70, p < .001; Cohen’s d = 0.89). Similarly, average RSA across the Clown task (Mean = 6.09, SD = 1.00) was significantly higher than baseline RSA (Mean = 5.48, SD = 0.97; t[110] =9.79, p < .001; Cohen’s d = 0.65). Examining the spread in individual differences in these changes revealed that 84% of mothers showed augmentation (positive values of the difference between task and Baseline) in the Puppet Show episode, and 83.8% of mothers showed augmentation in the Clown episode. Average RSA did not differ between Puppet Show and Clown episodes (t[101] =0.60, p = .550; Cohen’s d = 0.04). Thus, in accordance with interpretation of these situations as mildly challenging, RSA augmentation occurred for both tasks in comparison with baseline. The lack of difference between the tasks themselves cannot speak to nuances in dynamic patterns across time.

FIGURE 1.

Maternal respiratory sinus arrhythmia (RSA) across baseline and tasks. Note. Mean comparisons were completed using paired-sample t tests. Bars represent standard errors. ***p <.001

3.3 |. Multilevel models

3.3.1 |. Analysis plan

Multilevel modeling was used to account for the nesting of RSA values (Level 1) within participants (Level 2). Although “episode” could be considered an intermediary level in between epochs of RSA and participant, context effects were of primary interest, rather than nuisance variance (and episode could not be considered a variable if it was also used to specify the nesting structure). Models were estimated using full maximum likelihood.

Prior to testing hypothesized effects, a model with no predictors and RSA as the dependent variable was run to calculate the intraclass correlation coefficient (ICC) for RSA, which provides the proportion of between-subjects variability. Next, potential covariates were entered into the model to determine whether they should be included in primary models. Patterns of growth in RSA across episodes were tested by creating a time variable counting each 30-sec epoch, starting at 0 so the intercept would provide the expected value of RSA for the first epoch and adding an integer of 1 for each subsequent epoch. A quadratic time variable was created by squaring the linear values. As Level 1 variables, we tested whether either the linear or quadratic time variables should include random effects in addition to fixed effects using deviance change tests between a fixed-effect only model and another with the random component included. As participant characteristics, maternal perceptions of toddler shyness and distress reactions to fear resided at Level 2 and so could not have random components.

Context was dummy-coded such that the Clown episode was assigned a score of 1 and Puppet Show acted as the reference group with a score of 0. Maternal perceptions of toddler shyness and maternal distress in response to toddler fear were centered at their means. Interaction terms were created by calculating the cross-products between relevant variables. Observed shyness variables for Puppet Show and Clown were each included as covariates in the model including maternal perceptions of toddler shyness.

We tested the hypothesized interactions in a top-down manner, such that the main effects of linear time, quadratic time, relevant moderators, and interaction terms were included simultaneously. Significant interactions were probed by recoding the dichotomous moderator to change the reference group or recentering continuous moderators at low (−1 SD) and high (+1 SD) values and interpreting the time variables within these various conditions. Non-significant interactions were dropped, and the model was re-run to examine lower-order effects. Interactions involving the time variables would suggest that the moderator(s) impacted the slope of RSA across the episode. Main effects of time variables affected the intercept, or the initial status of RSA.

3.3.2 |. Preliminary models

An empty model (with no predictors) revealed an ICC of 0.56, suggesting that 56% of the variance in RSA was between participants, warranting the multilevel approach. Given an ICC of 0.56, 1475 total observations of RSA across 115 mothers, and an average of 13 observations per mother, we calculated an effective sample size (the equivalent single-level sample size to our nested sample of observations; Snijders & Bosker, 2004) of 191. A priori power analyses suggested that power of 0.80 for medium effects, alpha = 0.05, and 15 predictors (the maximum number of predictors per model) could be achieved with a sample size of 139. Our effective sample size of 191 exceeds this value, suggesting we were adequately powered for subsequent models. The linear and quadratic time variables were added one at a time, and tests revealed that adding random components with an unstructured covariance matrix did not improve the model for either the linear time term (χ²[2] =0.79, p = .672) or the quadratic effect (χ²[2] =0.13, p = .935); thus, we retained only the fixed effects for both time variables.

3.3.3 |. Context differences in dynamic RSA

The context interactions for both the linear and quadratic growth terms were significant (Table 2). The puppet show episode (interpreted within the original model) was characterized by positive linear (γ = 0.09, SE = 0.04, 95% CI = [0.02, 0.16], t[1350.30] =2.41, p = .016) and negative quadratic (γ = −0.01, SE = 0.00, 95% CI = [−0.02, −0.01], t[1353.04] = −3.41, p = .001) growth, yielding a pattern of increasing and then decreasing RSA values over the course of the episode. After recoding the context dummy code and recreating interaction terms, the clown episode was found to be characterized by non-significant linear (γ = −0.01, SE = 0.04, 95% CI = [−0.08, 0.06] t[1361.10] = −0.38, p =.702) and quadratic (γ = −0.00, SE = 0.00, 95% CI = [−0.01, 0.01], t[1376.14] = −0.63, p =.527) growth. Visualization of these patterns (Figure 2) suggests that although both episodes resulted in augmentation from baseline, RSA values in puppet show, the lower demand context, trended back toward baseline values, whereas the decrease was not significant in the clown episode, the higher-demand context. We suggest caution in interpreting the precise timing of return to baseline, as fewer mothers provided RSA values for the last epochs.

TABLE 2.

Multilevel model predicting context difference in growth of respiratory sinus arrhythmia (RSA)

Fixed effect	γ (SE)	95% CI	t Test	p
Intercept	6.11 (0.11)	5.90, 6.32	57.22	<.001
Maternal age	−0.04 (0.17)	−0.08, −0.01	−2.67	.009
Linear growth	0.09 (0.04)	0.02, 0.17	2.41	.016
Quadratic growth	−0.01 (0.00)	−0.02, −0.01	−3.41	.001
Context	0.11 (0.10)	−0.08, 0.30	1.10	.270
Linear × Context	−0.10 (0.05)	−0.21, −0.002	−2.00	.046
Quad × Context	0.012 (0.006)	0.0002, 0.02	1.99	.047
Random effect	Estimate (SE)	95% CI	Wald Z	p
L1 residual variance	0.64 (0.02)	0.59, 0.69	25.90	<.001
L2 variance	0.72 (0.10)	0.55, 0.96	6.94	<.001

Note: Maternal age was mean-centered. Growth variables were coded such that the first time point equaled zero. Context was dummy-coded with puppet show = 0 and clown = 1.

FIGURE 2.

Interaction between growth and context in relation to estimated log respiratory sinus arrhythmia (RSA). Note. Both linear and quadratic growth parameters interacted with context to predict RSA. Positive linear and negative quadratic trends were significant for the puppet show episode; neither growth parameter was significant for clown. Baseline RSA is included as a reference line although was not modeled

3.3.4 |. Maternal perceptions of toddler shyness and distress

Final models added maternal perceptions of toddler shyness and self-reported distress in response to toddler shyness. Three-way interactions among growth (linear or quadratic), maternal variable (perceptions of shyness or distress about shyness), and context were investigated to determine whether effects of maternal variables on dynamic RSA differed across contexts.

The first model included maternal perceptions of toddler shyness. Three-way interactions among growth, context, and perceptions of shyness were not significant for either linear (γ = 0.02, SE = 0.05, 95% CI [−0.07, 0.12], t[1193.67] =0.50, p = .615) or quadratic (γ = −0.00, SE = 0.01, 95% CI [−0.01, 0.01], t[1201.17] = −0.07, p = .941) parameters, so they were dropped and the model was re-run. A two-way interaction emerged between quadratic growth and perceptions of shyness, above and beyond interactions between growth parameters and context (Table 3). Probing of this interaction revealed that quadratic growth (most visible in our models as the downward turn toward baseline) weakened as perceptions of toddler shyness increased from low (−1 SD; γ = −0.023, SE = 0.01, 95% CI [−0.03, −0.01], t[1194.683] = −3.99, p < .001), to mean (γ = −0.016, SE = 0.00, 95% CI [−0.02, −0.01], t[1190.39] = −3.44, p =.001), to high (+1 SD; γ = −0.008, SE = 0.01, 95% CI [−0.02, 0.003], t[1193.828] = −1.47, p = .143) levels. Visualization of this pattern suggests that increased perceptions of toddler shyness related to slower (less steep) return toward baseline (Figure 3).

TABLE 3.

Multilevel models predicting maternal respiratory sinus arrhythmia (RSA) in two contexts

Fixed effect	Moderator: Maternal perceptions of toddler shyness				Moderator: Maternal distress about toddler shyness
	γ (SE)	95% CI	t Test	p	γ (SE)	95% CI	t Test	p
Intercept	6.10 (0.12)	5.86, 6.32	52.70	<.001	6.08 (0.11)	5.86, 6.30	54.51	<.001
Maternal age	−0.04 (0.02)	−0.07, −0.00	−2.20	.030	−0.04 (0.02)	−0.08, −0.01	−2.30	.024
Observed shyness: Puppet Show	−0.01 (0.10)	−0.21, 0.18	−0.12	.903	–	–	–	–
Observed shyness: Clown	−0.08 (0.11)	−0.28, 0.13	−0.74	.462	–	–	–	–
Linear growth	0.10 (0.04)	0.02, 0.18	2.53	.012	0.10 (0.04)	0.03, 0.18	2.65	.008
Quadratic growth	−0.02 (0.01)	−0.02, −0.01	−3.44	.001	−0.02 (0.00)	−0.02, −0.01	−3.51	<.001
Context	0.14 (0.10)	−0.06, 0.34	1.35	.178	0.16 (0.10)	−0.04, 0.35	1.59	.112
Moderator	—0.01 (0.09)	−0.18, 0.16	−0.14	.893	0.04 (0.08)	−0.11, 0.19	0.51	.610
Linear × Context	—0.11 (0.06)	−0.22, −0.00	−2.04	.042	−0.12 (0.05)	−0.23, 0.01	−2.16	.031
Quadratic × Context	0.01 (0.01)	−0.00, 0.03	1.93	.054	0.01 (0.006)	0.001, 0.03	2.03	.043
Linear × Moderator	−0.03 (0.02)	−0.08, 0.02	−1.32	.189	−0.06 (0.03)	−0.11, −0.002	−2.03	.042
Quadratic × Moderator	0.01 (0.00)	0.00, 0.01	2.05	.040	0.01 (0.00)	−0.001, 0.01	1.74	.083
Context Moderator	0.02 (0.04)	−0.05, 0.10	0.56	.576	−0.20 (0.07)	−0.33, −0.07	−2.94	.003
Linear × Moderator × Context	–	–	–	–	0.09 (0.04)	0.02, 0.16	2.44	.015
Quadratic × Moderator × Context	–	–	–	–	−0.01 (0.00)	−0.02, 0.00	−1.86	.063
Random effect	Estimate (SE)	95% CI	Wald Z	p	Estimate (SE)	95% CI	Wald Z	p
L1 residual variance	0.63 (0.03)	0.58, 0.69	24.31	<.001	0.63 (0.03)	0.58, 0.68	24.87	<.001
L2 variance	0.76 (0.11)	0.57, 1.03	6.51	<.001	0.74 (0.11)	0.55, 0.99	6.66	<.001

Note: Maternal age, observed shyness, and moderators were centered at their means. Growth variables coded the first time interval as zero. Three-way interactions were not significant for maternal perceptions of toddler shyness, so they were dropped for the model reported here.

FIGURE 3.

Interaction between growth parameters and maternal perceptions of toddler shyness in relation to estimated log respiratory sinus arrhythmia (RSA). Note. Both linear and quadratic growth parameters weakened as maternal perceptions of shyness increased. Effects were above and beyond context effects on growth. Observed shyness in Puppet Show and Clown were each included as covariates. Baseline RSA is included as a reference line although was not modeled

In the model with maternal distress responses to toddler shyness, the three-way interaction among linear growth, context, and distress was significant, and the three-way interaction among quadratic growth, context, and distress was marginally significant (Table 3). We probed these interactions by recoding context to examine two-way interactions between growth and distress for each context within the larger model (Figure 4). In the puppet show episode, two-way interactions were significant between linear growth and distress (γ = −0.06, SE = 0.03, 95% CI = [−0.11, −0.002], t[1245.80] = −2.03, p = .042), and marginally significant between quadratic growth and distress (γ = 0.006, SE = 0.003, 95% CI = [−0.001, 0.01], t[1247.06] =1.74, p = .083). We recentered distress to examine changes in growth patterns at various levels. Linear and quadratic growth parameters weakened as mothers reported increasing distress from low (linear: γ = 0.18, SE = 0.06, 95% CI = [0.07, 0.30], t[1243.67] =3.22, p = .001; quadratic: γ = −0.02, SE = 0.007, 95% CI [−0.04, −0.01], t[1246.41] = −3.58, p < .001) to mean (linear: γ = 0.10, SE = 0.04, 95% CI = [0.03, 0.18], t[1243.61] =2.65, p = .008; quadratic: γ = −0.02, SE = 0.004, 95% CI = [−0.02, −0.01], t[1246.68] = −3.51, p <.001) to high (linear: γ = 0.02, SE = 0.05, 95% CI [−0.09, 0.13], t[1245.93] =0.38, p = .701; quadratic: γ = −0.01, SE = 0.01, 95% CI [−0.02, 0.01], t[1247.40] = −1.17, p = .241) levels. In the clown episode, distress did not interact with either linear (γ = 0.03, SE = 0.02, 95% CI [−0.01, 0.08], t[1254.27] =1.38, p =.168) or quadratic (γ = −0.02, SE = 0.003, 95% CI = [−0.01, 0.003], t[1264.89] = −0.83, p =.405) growth terms.

FIGURE 4.

Interactions among context, maternal distress about toddler shyness, and growth parameters in relation to maternal respiratory sinus arrhythmia (RSA). Note. In the lower-threat puppet show context (Panel A), the interaction between linear growth and maternal distress was significant, and the interaction between quadratic growth and distress was marginally significant. Interactions were not significant for the moderate-threat clown context (Panel B). Baseline RSA is included as a reference line although was not modeled

4 |. DISCUSSION

Given rapid fluctuations in emotions and concomitant psychophysiology within and across situations, the use of dynamic approaches and analysis of contextual effects have been evidenced to be vital when investigating patterns of regulation (Brooker & Buss, 2010; Cui et al., 2015; Miller et al., 2013; Miller et al., 2016; Obradovic et al., 2010). The current study investigated maternal dynamic RSA, as predicted by the specific novelty context (lower versus higher demand on the toddler and parent), maternal perceptions of toddler shyness, and maternal distress about toddler shyness.

Prior to examining dynamic patterns, we examined mean-level differences in RSA between baseline and each of the novel contexts. Across the sample, each context exhibited RSA augmentation in comparison with baseline. According to Porges’s (2001, 2007) polyvagal theory, augmentation may be the expected response when individuals are faced with situations requiring social engagement. The increase in parasympathetic control puts a stronger break on sympathetic activation to allow for attentional and behavioral engagement. According to challenge and threat theory (Blascovich & Mendes, 2000), augmentation may be the typical response to situations that individuals find challenging, but not overtly threatening. In line with these theories, differences in average RSA in novelty contexts as compared to baseline suggest that mothers’ interactions with their toddlers in situations involving mild to moderate uncertainty are socially engaging and somewhat challenging, relying on parasympathetic control. In these analyses, average RSA values in the low-threat and moderate-threat contexts did not differ from one another, but stopping here would have prevented the detection of nuanced differences between contexts that emerged only when examining dynamic changes in RSA across time.

We examined linear and quadratic trends in RSA across the two contexts, finding differences between the episodes in growth. The context characterized as low threat for toddlers, a friendly puppet show, was expected to place the lowest demands on mothers. In this low-demand episode, RSA started higher than the baseline value, reflecting the mean-level RSA augmentation found above, increased slightly, and then decreased back toward baseline values. This pattern is consistent with the idea that initial attention and social engagement, facilitated by increased parasympathetic control, gradually returns toward a resting state once mothers acclimate to the situation and their toddlers’ responses to it. When contextual demands were higher in the episode characterized by moderate threat and uncertainty for toddlers, these trends disappeared, reflecting RSA augmentation that remained relatively steady across the episode. This may suggest that mothers maintain attention and engagement, as the situation maintains moderate demands on them to discern their toddlers’ responses. The clown stimulus was more mobile and less predictable than the puppet show, which presents more uncertainty to toddlers and therefore requires more of mothers. Challenge and threat theory might suggest that, in both situations, mothers generally have resources that outweigh the demands of the situation and can meet the parenting challenges posed by them; however, summoning these resources is only necessary at the beginning of the lowest threat episode and is maintained during the moderate-threat episode. The current study did not examine a “high threat” episode to determine whether the pattern would change to RSA suppression, as predicted by both polyvagal and challenge versus threat theories. In the case of the procedures used presently, a task of similar length to the included tasks was not available for an appropriate comparison. Comparing our results to those found with additional contexts will be an important future direction.

In addition to hypothesizing context differences, we also investigated maternal characteristics that may reveal individual differences in dynamic RSA patterns. Mothers’ perceptions of toddler shyness were chosen as one construct that might change the perceived demands of a novel context. Shy and anxiety-prone children elicit more parental involvement during novel and uncertain situations (Kiel & Buss, 2012), so we posited that preconceived perceptions of toddler shyness would increase the demands of the situation. Similarly to what happened when the context putatively increased demands, greater perceptions of toddler shyness related to maintained augmentation, suggesting that mothers’ parasympathetic control continued to engage their attention and engagement throughout their toddlers’ interactions with novelty. Maternal perceptions appear to be particularly important, as the interaction occurred above and beyond observed toddler shyness in the laboratory tasks. The interactive effect is consistent with the notion that mothers’ resources outweigh the demands placed on them and that they can muster the resources needed to manage the situation. It could be, however, that longstanding demands of this sort may weigh on mothers as they parent shy children throughout development. Whether maintained augmentation during a moderately demanding situation predicts adaptive or maladaptive outcomes for parents will need to be determined in future work.

The interaction between maternal perceptions of toddler shyness and growth in relation to RSA did not depend further on context. In other words, maternal perceptions of toddler shyness acted as a moderator of growth in a similar manner across the low-threat and medium-threat tasks. When mothers report on their perceptions of their toddlers’ shyness, they likely draw on longstanding interactions across a variety of contexts. Perhaps the broad nature of perceptions has a broad effect on psychophysiology, across acute situations. Given the inclusion of toddlers’ observed shyness, the role of maternal perceptions in dynamic RSA was above and beyond toddlers’ context-specific displays of shyness, supporting this interpretation. It is also possible that the demands of the different situations did not differ enough to change how perceptions would relate to parasympathetic regulation.

Another factor that maintained augmentation was maternal distress in relation to toddler shy behavior. In other words, the more distress mothers reported in relation to their toddlers becoming fearful or shy, the weaker the pattern of recovery from initial augmentation to resting values of RSA. This pattern was specific to the lower-threat context. At high maternal distress, dynamic RSA across the lower-threat context mirrored the general pattern of maintained augmentation across the context of moderate threat. Thus, in addition to toddler characteristics, mothers’ own emotional responses related to sustained RSA augmentation across a situation in which most mothers showed recovery toward baseline. Again, this pattern seems adaptive in the situation at hand, with maintained parasympathetic control allowing for the attentional resources and social engagement required to manage the situation. This pattern may also indicate that mothers are actively regulating their own arousal throughout this lower-threat situation. As mothers engage in this pattern across their children’s development, however, it remains to be seen whether this predicts adaptive or maladaptive outcomes for mothers.

Unlike for maternal perceptions of shyness, the context was important for the role of maternal distress about toddler shy behavior in dynamic RSA. One explanation for the specific role of context may be that the nature of the situation may provide an indication to mothers about how atypical their child’s behavior might be, with shyness being more confusing in a situation with low demands for interaction and greater control (e.g., the stimulus cannot move from its location). Perhaps when the stakes are particularly low (as in our puppet show episode), mothers’ general levels of distress matter for the psychophysiological regulation. Another explanation may be that the low-threat context is necessary to display the relevance of the maternal trait of distress proneness. Just like how toddlers who are prone to anxiety are most easily identified in their reactions to in low-threat situations (Buss, 2011), perhaps mothers who are prone to heightened distress about toddler shyness are most easily identified in their reactions to a low-threat situation, but in their case, by their putative difficulty returning to baseline parasympathetic activity in a low-threat situation. Studies specifically aimed at parsing apart mechanisms of maternal psychophysiology across varying contexts are needed to provide further support for either of these interpretations.

Thus, increases in contextual and personal demands related to maintained RSA augmentation, rather than initial augmentation followed by recovery toward baseline. These results conform to hypotheses derived from Porges’s (2001, 2007) polyvagal theory and Blascovitch and Mendes’s (2000) challenge versus threat theory. The current study adds to the growing literature supporting the investigation of parental psychophysiology during interactions with their children. Just as individual differences in parent physiology have emerged in relation to the attachment relationship and children’s externalizing behaviors (Hill-Soderlund et al., 2008; Lorber & O’Leary, 2005; Mills-Koonce et al., 2009), there appear to be important individual differences in RSA patterns when mothers engage in situations that may elicit distress to novelty in their toddlers. Replication of these results in independent samples will be important. A next step for this line of inquiry will be determining how features of dynamic RSA predict specific parenting behaviors that have implications for toddlers’ development from distress to novelty to social withdrawal and anxiety. Mothers’ baseline RSA has already been shown to determine when maternal anxiety relates to overprotective behavior (Root et al., 2016). Understanding the role that temporal aspects of psychophysiology play in anxiogenic parenting behaviors could strengthen support for specifically targeting parents’ physiological reactivity and regulation during family-focused prevention and intervention programs for child anxiety. Equally important will be future studies understanding how these patterns predict mothers’ own well-being, subjective parenting experience, and individual adaptation.

It is important to consider these results in light of existing limitations of the study. Given that all mothers experienced the laboratory episodes in the same order (with the moderate-threat episode occurring before the lower-threat episode), we cannot rule out the possibility of carry-over effects. It is possible that mothers of children who were highly distressed in the moderate-threat task were still dealing with that experience when the low-threat episode began, and that parasympathetic activity reached a point of exhaustion, which looked like recovery. Given the finding of individual differences consistent with theory, we believe concerns about context effects should be minimal. However, an experimental manipulation of episode order would be needed to rule out carry-over effects.

We did not measure the sympathetic nervous system, which would be necessary to fully understand overall levels of arousal. The current study used a moderately sized sample that was primarily European American and middle class, although it was relatively representative of low-income house-holds (21% of participants reported less than $20K/year). Future research would benefit from studying dynamic RSA in larger, more diverse samples to determine whether a parent’s own regulation and/or toddler distress to novelty affects psychophysiology in the same manner. Additionally, the study focused on concurrent measures, and the literature would benefit from studies employing longitudinal designs in order to more fully understand the relations among maternal physiological measures of regulation, subjectively experienced or behaviorally observed regulation strategies, and child behavior in the parenting context. Our measure of maternal distress about toddler shy behavior asked mothers to report on toddler fear in situations that did not precisely match the contexts observed in the laboratory. Finally, future research may benefit from investigating toddler outcomes to understand how maternal physiological and emotional regulation may predict emotion and emotion regulation development in toddlers.

In summary, the current study found that engaging in novel situations with toddlers related to mothers’ RSA augmentation and that contextual and individual difference factors related to growth patterns in maternal RSA. These results warrant continued investigation into the psychophysiological antecedents of parents’ experiences during toddlers’ interactions with novelty and subsequent outcomes. Future work may compare dynamic patterns of RSA to averaged values to determine whether the dynamic approach offers added predictive utility for understanding parenting responses to their toddlers in novel situations.