Interactive Effects of Maternal Physiological Arousal and Regulation on Maternal Sensitivity: Replication and Extension in an Independent Sample

Abstract

This study examined the extent to which mothers’ physiological arousal (i.e., skin conductance level [SCL] augmentation) and regulation (i.e., respiratory sinus arrythmia [RSA] withdrawal) interacted to predict subsequent maternal sensitivity. Mothers’ (N = 176) SCL and RSA were measured prenatally during a resting baseline and while watching videos of crying infants. Maternal sensitivity was observed during a free play task and the still-face paradigm when their infants were 2 months old. The results demonstrated that higher SCL augmentation but not RSA withdrawal predicted more sensitive maternal behaviors as a main effect. Additionally, SCL augmentation and RSA withdrawal interacted, such that well-regulated maternal arousal was associated with greater maternal sensitivity at 2 months. Further, the interaction between SCL and RSA was only significant for the negative dimensions of maternal behavior used to derive the measure of maternal sensitivity (i.e., detachment and negative regard) suggesting that well-regulated arousal is particularly important for inhibiting the tendency to engage in negative maternal behaviors. The results replicate findings from mothers in previous studies and demonstrate that the interactive effects of SCL and RSA in relation to parenting outcomes are not sample specific. Considering joint effects of physiological responding across multiple biological systems may enhance understanding of the antecedents of sensitive maternal behavior.

Introduction

Leerkes and colleagues demonstrated that maternal physiological arousal, indexed by Skin Conductance Level (SCL) augmentation, and regulation, indexed by Respiratory Sinus Arrhythmia (RSA) withdrawal, in response to infant crying: (a) predicted adaptive maternal social cognitions about crying which in turn predicted maternal sensitivity (Leerkes et al 2015; 2016), (b) had direct effects on higher maternal sensitivity in a fear-eliciting context (Augustine & Leerkes, 2018), and (c) predicted better infant attachment and mental health outcomes (Leerkes et al., 2017). Specifically, maternal arousal in response to infant crying was associated with more adaptive parenting and infant outcomes when accompanied by high maternal regulation and with maladaptive outcomes when accompanied by low regulation. Although the pattern was robust with both prenatal and postnatal measures of maternal physiology and with multiple outcomes, it has not been replicated in other samples. Thus, the goals of this study are to determine: (1) if the adaptive nature of well-regulated physiological arousal in response to infant crying for parenting replicates in a different sample; and (2) which specific dimensions of maternal behavior are undermined by dysregulated arousal.

SCL, a widely used measure of emotional arousal, is the amount of sweat that rises from sweat ducts to the skin and reflects sympathetic nervous system (SNS) activation (Stern et al., 2001). RSA, or vagal withdrawal, is a parasympathetic nervous system (PNS) response that reflects vagal regulation of the heart when confronted with challenge and is viewed as an indicator of physiological regulation (Porges, 2007). Specifically, a reduction in vagal control over the heart allows for a greater sympathetic response in the moment, and this is believed to support active coping. Generally, the SNS (fight or flight) and PNS (rest and digest) are viewed as acting in an antagonistic fashion, such that if one is high the other is low. However, distinct patterns of activation across the two are apparent (e.g., co-activation, co-inhibition; dominance of one system over the other), and both health and behavioral outcomes vary as a function of patterns across these systems (Berntson et al., 2008; Miller et al., 2015). In relation to parenting, Leerkes and colleagues (2015, 2016) asserted that the combination of higher SCL augmentation and higher RSA withdrawal in response to infant crying is adaptive in relation to parenting. This combination promotes more adaptive social cognition about crying and reflects a state in which mothers are aware of and moved by their infants’ distress and are simultaneously able to regulate their own arousal, which facilitates the ability to focus on infants’ needs rather than their own needs in the moment. Consistent with this view, mothers who demonstrated higher SCL augmentation coupled with higher RSA withdrawal were more likely to engage in infant-oriented cry processing, characterized by accurate distress detection, and empathic emotional and cognitive responses to crying, which in turn predicted higher sensitivity to distress (Leerkes et al., 2016). In contrast, mothers who demonstrated higher SCL augmentation coupled with lower RSA withdrawal were more likely to engage in mother-oriented cry processing, characterized by negative emotional and cognitive responses to crying, which in turn predicted lower maternal sensitivity to distress (Leerkes et al., 2015).

To our knowledge, only one subsequent study has tested the interactive effect between RSA and SCL in relation to parenting. Specifically, in a sample of 150 parents (primarily mothers) of 6- to 12-year-olds, Zhang et al. (2021) assessed change in parental SCL and RSA from baseline to a task in which parents passively watched their children’s performance during a frustrating task being criticized. SCL augmentation was associated with lower observed parental emotional availability in a joint problem-solving task and less supportive self-reported emotion socialization, but there were no statistically significant main or moderating effects of RSA withdrawal. Notably, the nature of the task in which physiology was assessed may have elicited disappointment/shame in the child’s performance rather than concern for the child and the subsequent dyadic task was challenging, but not particularly distressing, all of which may have contributed to discrepant findings. There is some evidence that RSA withdrawal may be particularly adaptive for parenting in emotion-eliciting situations because it supports a fight or flight response that may prompt mothers to ensure child safety, whereas RSA augmentation may be adaptive in less distressing problem-solving tasks because it prompts social engagement which may contribute to joint problem-solving (Augustine & Leerkes, 2019; Miller et al., 2015; Ravindran et al., 2022). Consistent with this view, RSA withdrawal has operated as a moderator of the association between other risk factors and emotion-related parenting outcomes in studies that assessed RSA change from baseline to a stressful parent-child interaction. For instance, RSA withdrawal buffered the association between: (a) maternal post-traumatic stress symptoms and non-supportive emotion socialization in a sample of mothers of 6- to 12-year-olds exposed to domestic violence (Gurtovenko & Katz, 2020); and (b) maternal anxiety and greater overprotection in fear-eliciting contexts among mothers of toddlers (Jones et al., 2021).

Given our primary goal is replication, we attempted to closely mirror the methods used by Leerkes and colleagues (2015, 2016). Specifically, we employed the same infant cry stimuli and covariates, except adult attachment which was not measuredin the current study. However, the current study varies in that we included multiparous mothers in addition to primiparous, observed maternal behavior when infants were younger (2 versus 6-months old), and employed a different coding scheme to rate maternal sensitivity (NICHD ECCRN, 1999; Mills-Koonce et al., 2011). We view the latter as a strength because it is a frequently used coding scheme and allows for examination of specific maternal behaviors in addition to sensitivity.

We test the following hypotheses: high SCL augmentation is associated with (1) higher mother-oriented cry processing, (2) lower infant-oriented cry processing, and (3) lower maternal sensitivity only when RSA withdrawal is low. Then, we extend prior work by examining the interaction between SCL and RSA in relation to sensitivity dimensions (i.e., sensitivity to distress and non-distress cues, intrusiveness, detachment, positive regard, negative regard) to determine if the joint effects of arousal and regulation are particularly robust in relation to certain parenting dimensions. The latter is warranted given evidence that specific features of insensitive maternal behavior have distinct predictors (Bailes & Leerkes, 2021; Ensink et al., 2017; Michl-Petzing et al., 2019).

Material & Methods

Participants

Participants were 176 mothers and their typically developing infants (50% female) participating in a study on the origins of childhood obesity (Leerkes et al., 2020). All were recruited prior to the onset of COVID 19. Mothers were diverse with respect to race (approximately 46% non-Hispanic White, 32% non-Hispanic Black, 9% multiracial, 4% from other racial groups, 3% did not disclose their race, and 8% were Hispanic), education (median = 2-year college degree, range = GED or less–post-graduate degree), age (M = 29 years, range = 18–47), and income-to-needs ratio (i.e., family income divided by the federal poverty level for a family of that size; M = 3.10, range = 0–17). The majority (77%) of mothers were married to or cohabitating with the infant’s father, and 45% of mothers were primiparous. Most (130) dyads participated in a laboratory visit when infants were 2 months old. Reasons for missing the visit include: infant mortality or health complication (2), withdrawal from study (5), inability to re-contact or schedule within the window (16), and COVID-related suspension of in-person data collection (23). Relative to the sample from the prior study we seek to replicate, women in the current sample were significantly older (M = 29 vs. 25 p < .001, η² = .11), modestly more educated (Med = 4, 2-year college degree vs. 3, some college, p < .05, η² = .01), and more likely to be multiparous (55% vs 0%, p < .001). The samples do not vary significantly on race, income-to-needs ratio, marital status, or infant gender.

Procedures

Expectant mothers completed online questionnaires in their third trimester prior to visiting the campus laboratory. During the lab visit, expectant mothers engaged in a 5-minute resting baseline and then viewed four 1-minute video clips of crying infants used by Leerkes et al. (2015). Mothers completed questionnaires about their perceptions of the infants and their own reactions to the videos after each clip. Information on infants (e.g., date-of-birth, sex) was obtained via phone approximately one-week post-birth. Then, mothers and 2-month-old infants (M = 2.24 months, SD = .56) participated in a 7-minute free play task and the still face paradigm (Tronick et al., 1978).

Measures

Physiological measures.

Maternal SCL and heart rate were recorded using a Biolog (UFI, Morro Bay, CA) during the baseline and each cry video clip, and files were downloaded for processing. SCL was recorded in microsiemens at a sampling rate of 100Hz. Mothers’ electrocardiogram was recorded at a sampling rate of 1 kHz. Artifact editing and RSA estimation were done via the CardioEdit software (Brain Body Center, University of Illinois at Chicago) using Porges’ (1985) methods. If more than 15% of RSA data within a task required editing, it was treated as missing. RSA, in msec², was calculated for 15-second epochs and then averaged across epochs within a task. RSA and SCL were missing for one mother who did not complete the cry video task. Of the remaining mothers, all had complete SCL data; 109 had complete RSA data, 55 had some missing RSA data, and 11 had no usable RSA data due to poor signal quality/exceeding the editing threshold. Repeated measures ANOVAs with follow-up contrasts demonstrated that the cry videos effectively elicited higher SCL and lower RSA relative to the baseline; p < .01for all contrasts from baseline. SCL and RSA change scores from baseline were calculated for each clip and then averaged across available clip scores. SCL augmentation reflects an increase in SCL from baseline to the cry clips (α = .94), and RSA withdrawal reflects a decline in RSA from baseline to the cry clips (α = .90).

Cry processing.

After viewing each videoclip, mothers rated how distressed the infant was on a scale from 1 (high positive) to 7 (high negative) with 4 labeled as neutral, noted all displayed emotions, and noted the dominant or strongest displayed emotion. Following Leerkes et al. (2015), this yielded 3 initial scores: (1) the number of clips (0 to 4) in which the mother correctly noted the infant was distressed (i.e., not positive or neutral); (2) the number of clips (0 to 4) in which the correct dominant emotion was noted; and (3) a weighted score reflecting the egregiousness of errors made if the incorrect dominant emotion was selected (i.e., positive emotion most egregious, 3; non-emotion word like bored/ tired, 2; wrong negative emotion, 1; no error, 0) summed across clips (0 to 12). These individual variables correlated significantly (rs.28 to −.79, all p < .01), and were used to create an accurate emotion identification composite by averaging standardized variables (egregiousness reversed).

Mothers also reported the extent to which they felt 20 specific emotions during each video on a 4-point scale ranging from not at all to very strongly on the My Emotions Questionnaire (Leerkes & Qu, 2020). This yielded scores for empathy, sympathy, anger, and anxiety for each clip that were subsequently averaged across clips (α’s ≥ .83). Participants also rated the extent to which they endorsed specific causal attributions about why each infant was crying using a 4-point scale ranging from strongly agree to strongly disagree (Leerkes & Siepak, 2006). This yielded scores for situational-emotional attributions (e.g., baby was upset by the situation), emotion minimizing attributions (e.g., baby is tired), and negative attributions (e.g., baby is spoiled) for each clip which were then averaged across clips (α’s ≥ .88). Then, after viewing all clips, expectant mothers rated their beliefs about crying on the Infant Crying Questionnaire (Haltigan et al., 2012) yielding two scores: infant-oriented cry beliefs (e.g., babies cry to communicate, α = .77) and mother-oriented cry beliefs (e.g., babies cry to manipulate others, α = .83). All cry processing scores were submitted to an exploratory factor analysis that yielded the two anticipated factors with eigenvalues greater than 2 that accounted for 54% of the variability. Each score was standardized and averaged as follows: infant-oriented cry processing (i.e., accuracy identifying infant distress/emotions, empathy and sympathy, situational/emotional causal attributions, infant-oriented cry beliefs; α = .72) and mother-oriented cry processing (i.e., anger, anxiety, minimizing and negative causal attributions, mother-oriented cry beliefs; α = .78). The indicators and factor structure mirror that used by Leerkes and colleagues (2015; 2016).

Maternal emotional risk.

Prior to the prenatal visit, mothers reported their depressive symptoms using the Center for Epidemiologic Studies–Depression Scale (Radloff, 1977; α = .90), trait anxiety using the State Trait Anxiety Inventory (Spielberger et al., 1970; α = .91), emotion regulation difficulties using the Difficulties in Emotion Regulation Scale (Gratz & Roemer, 2004; α = .94), and neuroticism and agreeableness using the NEO–Five Factor Inventory (McCrae & Costa, 2004; α’s = .84 and .76, respectively). These scores were standardized and averaged (agreeableness reversed), yielding a single score (α = .85).

Observed mother and infant behavior.

Maternal sensitivity to distress and non-distress, intrusiveness, detachment and infant negative affect were rated using 7-point scales adapted for each of three separate tasks: free play and the engagement and re-engagement episodes of the still-face task (ECCRN, 1999; Mills-Koonce et al., 2011). Ratings on parallel dimensions of behavior were correlated across tasks/episodes (mean r = .55) and were averaged across tasks (α = .62, .77, .73, and .69 for sensitivity to distress, sensitivity to non-distress, intrusiveness, and detachment, respectively). In addition, maternal positive regard and negative regard were rated on the same scale a single time based on maternal behavior across all tasks. Total maternal sensitivity was created by averaging sensitivity to distress and non-distress cues, positive regard, and reversed scores for intrusiveness, detachment, and negative regard (α = .88). Infant negative affect was averaged across tasks (α = .97). Inter-rater reliability was established based on 33 cases via intra-class correlation coefficients and was good for total sensitivity (.78) and infant negative affect (.97) and adequate for the specific maternal ratings (.67 −.76).

Results

Outliers (3 RSA withdrawal, 4 SCL augmentation, 2 intrusiveness, and 3 detachment) were winsorized prior to conducting analyses. Descriptive statistics and intercorrelations are presented in Table 1. Primary analyses were conducted in Mplus Version 8.6 (Muthén & Muthén, 2017) using full information maximum likelihood to handle missing data. Overall, 6.02% of data was missing, and the data were missing at random according to Little’s test, χ² (315) = 313.33, p =.52. Race, education, maternal emotional risk, infant-oriented and mother-oriented cry processing, and concurrent infant distress each correlated significantly with at least one rating of maternal behavior or the sensitivity composite and were retained as covariates.

Table 1:

Correlations and Descriptive Statistics

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. Race	-
2. Education	.42**	-
3. Emotional Risk	−.13	−.29**	-
4. Infant Distress	.06	.01	−.09	-
5. IO Cry Process	.15*	.19*	−.03	.01	-
6. MO Cry Process	−.20**	−.23**	.28**	.07	−.11	-
7. SCL Augmentation	.22**	.01	−.05	.20*	−.05	.02	-
8. RSA Withdrawal	.23**	.16*	−.19*	−.11	.06	−.15	.12	-
9. Sen Distress	.09	.22*	−.16	−.40**	−.01	−.26**	.10	.04	-
10. Sen Non-Distress	.13	.20*	−.22**	−.24**	.15	−.26**	.15	.03	.61**	-
11. Intrusiveness	−.10	−.03	.13	.26**	−.12	.31**	−.17*	−.05	−.50**	−.77**	-
12. Detachment	−.11	−.31**	.17*	.30**	−.16*	.28**	.09	−.10	−.48**	−.52**	.38**	-
13. Positive Regard	.22**	.31**	−.22**	−.40**	.19*	−.28**	.02	.04	.58**	.63**	−.42**	−.63**	-
14. Negative Regard	.12	.10	.10	.30**	−.02	.18*	−.00	.00	−.40**	−.30**	.49**	.39**	−.28**	-
15. Total Sensitivity	.12	.19*	−.23**	−.46**	.16	−.35**	.05	.06	.72**	.83**	−.76**	−.71**	.82**	−.61**	-
Mean	.46	4.24	−.001	2.88	.00	.00	2.02	.32	3.99	4.98	2.03	1.41	5.34	1.42	5.79
SD	.50	1.88	.79	1.59	.70	.75	1.84	.79	1.0	1.0	.80	.59	1.15	.73	.73
N	176	172	172	128	175	175	175	162	84	127	128	128	128	128	128

Note:

^*p < .05;

^**p < .01.

First, two hierarchical regressions were calculated predicting infant-oriented and cry mother-oriented cry processing, respectively. In each, the covariates and main effects for RSA and SCL were entered in the first block. The SCL by RSA interaction (i.e., multiplicative product of the two grand mean-centered terms) was entered in Block 2. Given the focus of the paper, we focus exclusively on the results related to SCL augmentation, RSA withdrawal, and their interaction in the narrative summary. There were no statistically significant main or moderating effects of physiology on mother-oriented or infant-oriented cry processing (complete results available in supplemental Table 1).

Next, hierarchical multiple regressions were calculated predicting the total sensitivity composite and each specific dimension of maternal behavior, with infant-oriented and mother-oriented cry processing included as additional predictors in Block 1. As noted in Table 2, SCL augmentation was associated with higher total sensitivity (qualified by an interaction), sensitivity to both distress and non-distress cues, and with lower intrusiveness. There were no statistically significant main effects of RSA withdrawal. The interaction between RSA and SCL was statistically significant for total sensitivity, detachment, and negative regard. None of the confidence intervals (CI) for the interaction terms included zero: 95% CIs were [.002, .14], [−.14, −.02] and [−.17, −.003], respectively. Interactions were probed via simple slope analyses at ± 1SD from the mean for RSA withdrawal and regions of significance were identified via plots generated by Mplus. As illustrated in Figure 1, Panel A higher SCL augmentation was positively associated with total sensitivity only when RSA withdrawal was high. According to the regions of significance plots in Mplus, the association between SCL augmentation and sensitivity was significant at mean levels of RSA withdrawal and higher. As illustrated in Figure 1 Panel B, SCL augmentation was only positively associated with detachment when RSA withdrawal was low; this association was significant when RSA withdrawal was at or below the mean according to the regions of significance plots. As illustrated in Figure 1 Panel C, SCL augmentation was only negatively associated with negative regard when RSA withdrawal was high; this association was significant when RSA was at or above the mean (from regions of significance plot).

Table 2.

Multiple Regression Results Predicting Total Sensitivity and Behavioral Dimensions

	Sen Comp	Det	Neg Reg	Sen-ND	Sen-D	Pos Reg	Intr
	β	β	β	β	β	β	β
Main Effects R 2 Δ	.41 **	.24 **	.17 **	.25 **	.23 **	.36 **	.23 **
Maternal Race	.02	.02	.12	.01	.05	.11	−.03
Maternal Education	.06	−.24**	.13	.11	.18	.18*	.08
Emotional Risk	−.19*	.08	.14	−.18*	−.12	−.17*	.10
Infant Distress	−.51**	.29**	.32**	−.33**	−.33**	−.45**	.31**
Infant Oriented Cry Processing	.13	−.10*	−.05	.15	−.06	.14	−.12
Mother-Oriented Cry Processing	−.26**	.18*	.18*	−.17*	−.18*	−.15*	.28**
SCL Augmentation	.17*	.02	−.10	.23*	.20*	.11	−.25**
RSA Withdrawal	−.11	.00	.06	−.13	−.06	−.14	.08
2 Way Interaction Terms R 2 Δ	.01 *	.03 *	.03 *	.00	.02	.01	.00
SCL X RSA	.14*	−.19*	−.17*	.07	.13	.11	−.02
Total R 2 Δ	.42 **	.27 **	.20 **	.25 **	.25 **	.37 **	.23 **

Note:

^*p < .05;

^**p < .01.

β = beta at entry block. Sen comp = sensitivity composite score, Det = detachment, Neg Reg = negative regard, Sen-ND = sensitivity to non- distress cues, Sen-D = sensitivity to distress cues, Pos Reg = positive regard, Intr = intrusiveness.

Figure 1.

Interaction effect between physiological arousal and regulation in relation to sensitivity (Panel A), detachment (Panel B) and negative regard (Panel C). RSA = respiratory sinus arrhythmia; SCL = skin conductance level

Given the difference in parity status between this sample and that from Leerkes and colleagues (2015, 2016), post hoc analyses were conducted to determine if parity was a significant moderator. Parity, parity X RSA withdrawal, parity X SCL augmentation, and parity X SCL X RSA were added to the regressions described above for cry processing (Supplemental Table 2) and maternal behaviors (Supplemental Table 3). Two of 27 (7.4%) possible two-way and 3-way interactions involving parity were statistically significant. First, the 3-way interaction was significant in relation to negative regard, β = −.28, p = .03, CI[−.38,−.02]. Probing indicated that the 2-way interaction between SCL and RSA was significant only for multiparous mothers, β = −.26, p = .008, CI[−.26,−.04]. Among multiparous mothers, SCL augmentation was negatively associated with negative regard when RSA withdrawal was high, β = −.41, p = .002, but not when RSA withdrawal was low, β = .14, p = .41. Second, SCL augmentation interacted with parity in relation to sensitivity to distress cues, β = .37, p = .005, CI[.08, .48]. SCL augmentation was associated with higher sensitivity to distress among primiparous mothers, β = .43, p = .001, but not among multiparous mothers, β = −.08, p = .54.

Discussion

The goals of this study were to (a) determine if the previously reported adaptive interactive effect between SCL augmentation and RSA withdrawal in relation to cry processing and maternal sensitivity would replicate in an independent sample and (b) extend prior results by determining which dimensions of sensitivity were supported or hindered by specific patterns of maternal physiological arousal and regulation. The results demonstrated SCL augmentation in response to infant crying was generally adaptive in relation to sensitive maternal behavior. Further, the previously reported interactive effects of SCL and RSA in relation to parenting outcomes are not sample specific.

Although not predicted, the positive main effect of SCL augmentation in response to infant crying on sensitive maternal behavior underscores the view that negative emotions or arousal may be adaptive and motivate effective parenting under certain circumstances (Dix, 1991). In this case, the cry videos included infants who were in obviously challenging situations and very distressed. As such, arousal may reflect a normative and adaptive awareness of infants’ emotional needs. In contrast, high arousal in response to other child stimuli (e.g., poor child performance on a task) may be non-normative and prompt less sensitive behavior (Zhang et al., 2021).

The proposed interaction between physiological arousal and regulation was supported in relation to total sensitivity, detachment, and negative regard, although the pattern of simple slopes varied for detachment versus sensitivity and negative regard. On one hand, higher maternal physiological arousal in response to infant crying appears to promote more adaptive parenting outcomes when well-regulated, as evidenced by the positive association with sensitivity and the negative association with negative regard when RSA withdrawal in response to crying was high. On the other hand, higher arousal was a risk factor for greater detachment when RSA withdrawal was low. It may be that high arousal accompanied by poor regulation prompts a focus on self, explaining the tendency to be detached, but high arousal accompanied by adaptive regulation promotes active responding to infant signals explaining the tendency to be more sensitive and less negative. In our prior research, we also observed promotive effects of arousal when well-regulated and deleterious effects when poorly regulated (Leerkes et al., 2015, 2016). Notably, inspection of the left side of each panel in Figure 1 illustrates that the combination of low arousal and high regulation, which may reflect over-regulation, is associated with the most negative parenting behavior. In our previous research, this pattern of physiology was related to higher infant attachment disorganization and behavior problems (Leerkes et al., 2017). The pattern of results across studies suggests that over- and under-regulation both have negative implications for parenting and child well-being. In contrast, a functional balance between arousal and regulation (e.g., no need to regulate when not aroused) is adaptive as evidenced by mothers with low SCL arousal and low RSA augmentation displaying the least detachment (left side of Panel B). This is consistent with Dix’s (1991) contention that emotions that are too weak or too strong may undermine parenting.

Notably, the only specific behavioral dimensions linked with the interaction between SCL and RSA were detachment and negative regard which suggests that well-regulated arousal is particularly important for inhibiting the tendency to engage in negative maternal behaviors. Consistent with this view, RSA withdrawal operated as a meaningful buffer in reducing the likelihood of mothers with high post-traumatic stress engaging in non-supportive emotion socialization but did not enhance the likelihood of them engaging in supportive emotion socialization (Gurtovenko & Katz, 2020).

In contrast to prior results from Leerkes et al. (2015), the effect of physiology on parenting was direct rather than indirect via cry processing. Although the data collection methods in both studies are comparable in many ways, four methodological differences may contribute to this difference. First, the current sample included both multiparous and primiparous mothers, whereas the prior research focused exclusively on primiparous mothers. However, post-hoc analyses demonstrated that the results did not vary consistently as a function of parity (i.e., only 2 of 27 interactions significant; simple slopes significant for multiparous mothers in one case and for primiparous mothers in the other); as such it is unlikely this in and of itself contributed to this difference in results. Second, there was a shorter lag between the prenatal and postnatal data collection in the current study compared to the prior study (approximately 3 versus 7 months), which may also have increased the odds of observing direct effects from prenatal physiology to subsequent parenting behavior. That is, there was less time for experience-dependent change in physiological reactions to infant crying in the current study. Third, our measures of maternal sensitivity and specific behaviors were aggregated across distress-eliciting and non-distress eliciting contexts because we only had one very brief distress eliciting task in this study (2 minute still-face re-engagement), whereas the prior research focused exclusively on sensitivity in a series of distressing contexts (Leerkes et al., 2015, 2016). Finally, in the prior work, we included adult attachment coherence as a predictor of sensitivity which may predict overlapping variance in sensitivity undermining our ability to detect direct effects of physiology in the prior study relative to this one. Despite differences in the pathways to more sensitive behavior across studies, the overall pattern of findings demonstrates that physiological arousal in response to crying as indexed by SCL augmentation is only maladaptive in relation to sensitivity when poorly regulated (i.e., both over and under) as indexed by concurrent RSA withdrawal.

Strengths of this study include efforts to replicate Leerkes and colleagues (2015, 2016) prior methods as closely as possible, employing similar covariates, observing and coding mother and infant behavior using well-established paradigms and coding schemes, and the diverse, moderately sized sample. Limitations include the data are from a community sample with relatively low levels of negative maternal behaviors, and that we are unable to focus exclusively on maternal behavior in distress eliciting contexts. Future research should expand upon these findings to examine if similar results emerge in higher risk samples and with other negative parenting behaviors (e.g., hostility, abuse/neglect), and if the pattern of prediction is similar or different in distress and non-distress tasks. Ideally, future replication will occur in independent laboratories.

Conclusion

In conclusion, the finding that prenatal physiological arousal in response to infant crying is only maladaptive in relation to parenting outcomes when poorly regulated is not sample specific. Furthermore, the joint effects of arousal and regulation appear to be most important in relation to inhibiting negative parenting behavior. Given parenting is a complex human behavior, likely driven by the interplay of multiple physiological systems, parenting researchers should consider complex patterns across systems. Viable alternatives to interaction effects for this purpose are to use person-oriented analyses (e.g., Zeytinoglu et al., 2021) or to employ the method proposed by Bernston et al (2008) when cardiac-based measures of both SNS and PNS are employed (see Miller et al., 2015).

Data Availability Statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.