Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jan 1.
Published in final edited form as: J Hum Behav Soc Environ. 2021 Jun 2;32(5):574–590. doi: 10.1080/10911359.2021.1931618

Social Network and Behavioral Synchrony Influences On Maternal and Infant Cortisol Response

Laura A Thompson a, Bryan White b
PMCID: PMC9216205  NIHMSID: NIHMS1714216  PMID: 35757570

Abstract

This study examined the contributions of mothers’ social network stability and mother-infant behavioral synchrony on cortisol response in infants and their mothers during separation. The quality and stability of mothers’ social network system and mother-infant bond have both been shown to affect infant neuroendocrine response. Yet, no studies have directly addressed how these two forms of social relationships might differentially affect infants’ and mothers’ neuroendocrine responses during separation. First-time mothers (N = 133) and their 3-month-old infants participated in the study. Maternal social network stability, mother-infant behavioral synchrony, and mother and infant cortisol response during an infant challenge task were assessed. Behavioral synchrony accounted for significant variance in infants’ cortisol response, and after adjusting for synchrony, mothers’ network stability measures did not explain variance in infant cortisol. Social network stability, but not synchrony, accounted for significant variance in mothers’ cortisol response. These results demonstrate that, when mothers and infants experience brief separation, the quality of their bond is associated with a lower stress response for infants; but for mothers, it is the longevity of her social relationships outside of the mother-infant relationship context that is associated with her lower stress response.

Keywords: cortisol, social networks, infants, behavioral synchrony, hypothalamic-pituitary-adrenal axis

All of us must “get our act together” in the morning, sometimes rushing through our morning routine to drive to a new place, meet new people, and possibly do something we’ve never done before. When a mother and her young infant arrive at our laboratory to participate in an experiment for the first time, both she and her infant are sharing in a mildly challenging experience, engaging their biobehavioral response systems in a way that is by now commonplace for most infants. And yet, due to genetic programming, maturity, and experience, while they are together in the same event, mothers’ and infants’ physiological experiences may be vastly different. Mothers enter the new experimental context with stress response systems evolved to protect and care for their offspring (Feldman et al., 2007; Taylor et al., 2000); infants’ biobehavioral stress response systems are designed to be affected by this protection and nurturance. In addition, mothers and infants differ in terms of the knowledge they hold about their connections to other people who are from multiple layers of an ecological system of social support (Bronfrenbrenner, 1979; 1986; Nath et al., 1991), people who, even though not physically present, may affect their biobehavioral response to challenge.

Lin et al. (1979) define social support as “support accessible to an individual through social ties to other individuals, groups, and the larger community” (p. 109). For young infants, the majority of their experiences in social connection have involved their mothers, and likely a handful of other individuals. Social support available to the mother affects not only maternal attitudes and life satisfaction, but mother-infant interactive behaviors (Crnic et al., 1983), and the quality of infants’ attachments to their mothers (Crockenberg, 1981). For mothers, their social connections have widened and deepened over the years to include a large social sphere of human relationships that could affect their physiological response to challenge. Levitt et al. (1986) found that mothers with infants reported an average social network size of 13 people, and that the majority (52%) said that they would like more people in their networks that they could depend on. The primary question addressed by the present study is whether the quality of two types of social support influences, namely, the mother-infant relationship and the mother’s social network, have differing effects on their cortisol responses when mothers and infants are sharing the same environment in a novel event.

Drawing upon a vast literature (Taylor, 2002; Taylor et al., 2000), Taylor (2006) proposed a model to help understand how human affiliative efforts affect, and are influenced by, stress physiology. As a first step, the hormone oxytocin promotes humans to seek contact with others. These affiliative efforts can result in both positive and negative contacts. When efforts are positive, oxytocin, in concert with opioid and dopaminergic systems, dampen psychological and biological stress responses, reducing both sympathetic nervous system (SNS) and hypothalamic-pituitary-adrenal (HPA) axis responses (see also Heinrichs et al., 2003). However, when negative contacts are made, stress responses increase. Investigators have found that physiological stress responses are associated with activity in dorsal anterior cingulate cortex and Brodmann’s area 8, neural regions known both for cognitive processing and for modulating affective responses to distress at social separation (Eisenberger et al., 2007). Thus, the quality of social relationships formed through affiliative relationships can have both stress buffering, and stress exacerbating, physiological effects, which in turn affect cognitive and emotional processing.

Research with humans and other highly social animals shows that the quality of adult relationships is a major predictor of the biobehavioral stress response (Hennessy et al., 2009; Kikusui et al., 2006; Ozbay et al.; Southwick, 2008). For example, Stetler and Miller (2008) experimentally manipulated the amount of daily social contact women received and found significant differences in the steepness of diurnal cortisol functions when they ate lunch alone, compared to when they had lunch partners with whom they were in daily contact. Also in line with expected effects on the HPA axis due to positive affiliative contacts, women with high levels of perceived support experienced lower cortisol levels across pregnancy, and those with high cortisol levels experienced low levels of perceived support (Giesbrecht et al., 2013). Moreover, within Mexican-American families, lower cortisol output was present in low-income mothers of infants with high, compared to low, family support (Jewell et al., 2015). However, while maternal bonding behaviors were higher for infants’ mothers who exhibited higher levels of oxytocin, investigators in one study reported a lack of a correlation between oxytocin and maternal cortisol (Feldman et al., 2007).

Less is known about how social contacts may impact infants’ biobehavioral responses to stress, and related findings are inconclusive. Prior to birth, greater levels of family support predicted more gradual increases across pregnancy in placental corticotropin-releasing hormone, which is a hormone involved in the HPA stress response (Hahn-Holbrook et al., 2013). While some research has failed to reveal significant correlations between maternal sensitivity and infant cortisol reactivity measures in both cross-sectional (Thompson & Trevathan, 2008; 2009) and longitudinal (Thompson et al., 2015) studies, other research has found that infants who are securely attached to their mothers show less elevated cortisol during challenging situations (e.g., Gunnar et al., 1996). The difference in findings across studies may be primarily due to the different forms of measurement (attachment security versus mother-infant interactive behaviors as a measure of maternal sensitivity). Alternatively, maternal sensitivity influences may be revealed when they are considered in combination with wider influences of social support that may act upon infants’ biobehavioral systems. Moreover, there exists some evidence that maternal insensitivity is associated with an increase in infants’ autonomic stress responses, and is marginally associated with increases in infant cortisol (Enlow et al., 2014). The latter finding is consistent with research on both humans (Kirschbaum et al., 1995) and other animals (Kikusui et al., 2006) showing that the presence of another individual, instead of having a social buffering effect, can actually serve as a stressor.

To date, there have been no studies that directly address potential effects of both the quality of the mother-infant relationship and mothers’ social support on infant cortisol response to challenge; likewise, this is an unexplored research area in terms of assessing how these two forms of social relationships affect maternal cortisol response. Although network size was found in a recent meta-analysis to be the primary means of measuring the quality of social networks in studies of psychological disorders (Degnan et al., 2018), there are many possible ways of describing and measuring social network quality that take into account more nuanced aspects of the structural characteristics of the network, as well as the quality of relationships between individuals. We decided to employ such a measure, one that is used both in social science research and by family clinicians, called the Social Network Map (Tracy & Whittaker, 1990). Moreover, we focused on one aspect of network quality, namely, the longevity or stability of a new mother’s relationships with others outside of the mother-infant relationship, for two reasons: a) a recent study in rhesus macaques showed an association between greater social instability and cortisol elevations (Vandeleest et al., 2019); and, b) sample size limitations meant that not all measures from the Social Network Map could be tested simultaneously to achieve sufficient power.

In measuring the quality of the mother-infant relationship, we employed a system that is derived from a dyadic view of how best to conceive of maternal sensitivity (Censullo et al., 1987; Feldman, 2003, 2007; Isabella & Belsky, 1991; Jaffe et al., 2001; Tronick & Gianino, 1986). Maternal sensitivity was assessed at each phase by observing the behaviors of mother-infant dyads during a 10-min laboratory play session using a method nearly identical to that introduced by Isabella et al. (1989). Their method outlines mothers’ sensitive and appropriate caregiving responses to infants’ behaviors, and a percentage of time spent in behavioral synchrony is the outcome measure for both mother and infant. For the purposes of the present study, a higher quality mother-infant social connection is indicated by a greater percentage of time spent in behavioral synchrony in the laboratory play session.

The present study focuses on the question of how the stability of the mother’s social network and maternal-infant synchrony might be associated with a reduction, or an exacerbation, in infants’ and their mothers’ HPA response to a mildly challenging event for the infant (maternal separation, combined with cognitive challenge). The experimental paradigm did not specifically challenge the mothers. However, given that: a) mothers had brought their infants to the laboratory and watched them participate, b) cortisol levels can resonate between observers and those experiencing a stressful event (Buchanan et al., 2012; Waters et al., 2014), and c) there are variations in mothers’ cortisol response to this laboratory paradigm (Thompson et al., 2015), we felt it was reasonable to expect that social network stability may be associated with variations in mothers’ cortisol response while their infants were engaged in a mildly challenging situation. Furthermore, due to mothers’ and infants’ different places in the lifespan, we predicted that mother-infant behavioral synchrony would exert a significant buffering influence on infant cortisol response, but would have a negligible effect on mothers’ cortisol responses.

Method

Participants

The current study used a subset of data from a 5-panel longitudinal study of infant cortisol response during learning events (Thompson et al., 2015). The study was conducted in the laboratory of a university located in a community in southern New Mexico. To control for variation in cortisol response and mother-infant interaction due to maternal experience, the participant sample of 133 mother-infant pairs included only first-time mothers, who were recruited through newspaper advertisement and by posting fliers throughout the city. In the current study, data were obtained from two laboratory visits, when the infants were: for Session 1, 1- to 2.5-months-old; and for Session 2, 3-months-old. There were 131 mothers that completed the Social Network Map in Session 1, but 10 mothers did not return for Session 2, so the final sample included 123 mother-infant pairs. None of the mothers’ cortisol data were excluded, but data from 3 infants were excluded, leaving 123 mothers and 119 infants in the final data set (46 female, and 73 male infants). The Social Network Map study variable means of those who returned for Session 2 did not differ significantly from those who did not return (p’s > .10).

The ethnic composition of the initial sample of mothers (N = 133) who provided data was primarily Hispanic or Latino (50%) and Non-Hispanic Caucasian (41%) (9% ‘other’). The racial composition was 40% White, 3% Black/African, 1% American Indian/Alaskan, 1% Asian, and 55% ‘other’ (including ‘more than one race’). Mothers averaged 24.8 years of age (SD = 6.2 years) and 14.9 years of education (SD = 2.9 years). They were paid $20 for each of two sessions for their participation.

All infants were full-term at birth (37–42 weeks gestational age and >5 lb birth weight; M = 7.1 lb, SD = 0.97), and the majority (77%) experienced vaginal deliveries, with uncomplicated perinatal histories. The percentage of mothers who were solely breastfeeding their infants at 3 months was 59.3%, solely bottle feeding was 35.0%, and using a mixture of bottle feeding and breastfeeding was 5.7%. Mothers and infants were all in good health at both study sessions. The procedures complied with the standards of the New Mexico State University Internal Review Board. Informed consent was obtained from all mothers in the study.

Sociodemographic and Health-Related Questions

The purpose of Session 1 was to give general information about all of the sessions in the longitudinal study, collect demographic and health information, and administer the Social Network measure (Tracy & Whittaker, 1990). Demographic questions included whether the mother smoked cigarettes, and whether the infant or mother was taking medication (and what type).

Social Network Measure

The Social Network Map was designed by Tracy and Whittaker (1990) to assess the size and make-up of the mother’s social network, the duration and frequency of contact of each individual in her network, and the various types of support received from network members. The map first requires respondents to sort their network members by name into 7 different categories based on which area of life they are from, including household, formal services, neighbors, friends, clubs/organizations, work/school, and other family. They are then asked to list each network member and respond to questions specific to that particular member. The Social Network Map provides a variety of information including network size, which is equal to the total number of individuals in the network, and stability, or the average length of time that the respondent has known each member. Stability is reported on a 3-point scale with 1 = known less than a year, 2 = known one to five years, and 3 = known more than 5 years. The map also assesses the frequency of contact on a 4-point scale with 1 = few times per year, 2 = monthly, 3 = weekly, and 4 = daily. Concrete and emotional support, information/advice, and criticalness are also examined with respondents reporting whether they almost always, sometimes, or hardly ever receive these types of support. The level of closeness with each member is also reported (1 = not very close, 2 = sort of close, and 3 = very close), as well as the direction of help (1 = help goes both ways, 2 = help is from you to them, and 3 = help is from them to you).

With the exception of network size, a percentage is calculated for each variable indicating how many of the total network members fall into each category. For example, it may be reported what percentage of the total network is from the household, or what percentage of the total network is reported as being critical almost always. (Please see the Appendix for a complete description of the Social Network Map instructions.)

Sequence of Events for Session 2; Saliva Collection and Cortisol Analysis

The timeframe for Session 2, when saliva was collected, was between 7:00 and 11:30 a.m. Mothers were instructed not to feed their infants immediately prior to testing while at the lab. The first saliva sample (T1) was considered a baseline, and was collected from both mothers and infants 20–25 min after entering the lab. The second saliva sample (T2) was collected 20–25 min after the start of the habituation experiment that the infants participated in at age 3 months. During the time that infants participated in the experiment, mothers sat quietly in a chair behind a curtain that separated them. The timing of T2 saliva samples was intended to reflect HPA activity at the beginning of the habituation experiment for infants, and at the beginning of a restful period for mother, that would result in cortisol levels measured in saliva 20–25 min later (Ramsay & Lewis, 2003; Thompson et al., 2015). The 10-min videotaped interaction period used to derive the mother-infant behavioral synchrony measures occurred immediately after the habituation experiment, and immediately prior to T2 saliva collection.

To collect saliva from infants, a cotton dental roll was applied to the tongue, cheeks, and gums (Lewis & Ramsay, 1995); mothers chewed on the dental roll until it was saturated. Rolls were then placed into tubes, centrifuged, and the saliva stored in a freezer at −20°C until large batches could be shipped to Salimetrics, Inc., for radioimmunoassay analysis (Fleming et al., 1997). Assays were performed in duplicate, and the results averaged. Salimetrics reports average inter-assay coefficients of variation of 5.1% (Salimetrics, 2012); intra-assay coefficient of variation for dual measures in this study was 4.41%.

Habituation Experiment

The present study used a standard habituation paradigm procedure to investigate whether 3-month-old infants could discriminate changes in audiovisual test stimuli, and whether their success at discriminating these changes would be related to their patterns of cortisol response. The association between cortisol response patterns and infant learning was reported in an earlier monograph (Thompson et al., 2015), thus the details of the experiment are not included here. The important characteristics of this paradigm for the purpose of the present study are: (a) infants were separated from their mothers during the experiment behind a partition, (b) mothers could view the infant participating in the experiment from a viewing window on the other side of the partition if she chose to, (c) the experiment was a novel event for the infants, and (d) mothers had an opportunity to rest while remaining in the same room but were physically separated from their infants. Infants sat in a seat alone, in front of the large booth containing the computer monitor. One experimenter was situated outside a testing booth and viewed the infant through a window above the monitor displaying the stimuli, while the mother sat outside a curtain separating her from her infant. The duration of the experiment was approximately 10 min.

Mother-Infant Behavioral Synchrony Measures

Following the habituation experiment, and prior to the T2 saliva sample, mothers and infants were videotaped for 10 min in the lab while they were alone together. Mothers were told to ‘play with’ their infants in any way they ‘felt to be comfortable and appropriate’. The videotapes were digitized and imported into the Observer (Noldus Corporation) software program. All maternal and infant behaviors were coded in 10-s intervals according to a system initially developed by Isabella, Belsky, and von Eye (1989; Isabella & Belsky 1991), which we modified only slightly and reported in several studies (Thompson et al., 2015; Thompson & Trevathan, 2008, 2009). Eleven specific maternal behaviors (i.e., no behavior, leisure activity, leisure and attend to infant, attend to infant, vocalize to infant, stimulate/arouse, stimulate/arouse and vocalize to infant, en face interaction, response to infant vocalization, soothe, response to vocalization and soothe) and 10 infant behaviors (i.e., sleep/drowsy, no behavior, explore, look at mother, vocalize, vocalize and look at mother, en face interaction, response/explore, fuss/cry, response/explore and en face) were coded by individuals who were extensively trained on the coding system. Coders were trained to a level of 90% reliability with a set of reference videos. Two coders independently coded each video, and each video was coded for maternal and infant behaviors separately. The two coders who coded each video reviewed every interval where their codes differed, and resolved their discrepancies through discussion.

The resulting set of codes for maternal and infant behaviors within each 10-s interval were categorized as to whether the interaction was synchronous, asynchronous, or neutral. Synchronous interactions were those where the mother’s response to her infant seemed sensitive and contingent upon the infant’s behavior; whereas, asynchronous interactions included a mother who was unresponsive, or behaved in a one-sided or intrusive fashion (Isabella et al., 1989). For example, if the infant vocalized and the mother’s response soothed the infant and/or made some other sensitive response, the dyadic interaction was classified as synchronous. If the infant explored a toy and the mother’s response was to pick up a different toy and vocalize, the interaction was coded as asynchronous because mother’s behavior interfered with the infant’s exploration. Mother-infant behavioral combinations not listed in the table were considered neutral. The final measures of maternal sensitivity were the percentages of synchronous interactions (number of synchronous mother-infant interactions divided by the total number of intervals that were assigned synchronous, asynchronous, or neutral codes; for a complete list of mother-infant behavioral combinations that were identified as synchronous and asynchronous, please see the Appendix published in Thompson et al., 2015).

Results

Cortisol Analyses

The raw cortisol values were not normally distributed and thus were transformed to log10 scores (Larson et al., 1998). Cortisol data were excluded from analyses when there was insufficient saliva for analysis in either sample for a participant or if either value was +3 SD higher than the mean for the referent group (n = 0 mothers; n = 3 infants). Research has demonstrated that a circadian rhythm in neuroendocrine functioning is apparent in infants by the age of three months (e.g. deWeerth et al., 2003; Gunnar, 1992), showing generally higher morning cortisol values which decline during the period of our testing window (7:00 – 11:30 a.m.). Due to the possibility of circadian influence on cortisol response values, bivariate correlations were conducted between session start times and infant and maternal cortisol response values; these correlations were nonsignificant (p’s > .10), so they were not used as covariates.

The mean (SE) raw T1 and T2 values are plotted in Figure 1, however, all subsequent analyses used the log10 values. On average, maternal cortisol declined significantly between T1 and T2, t(122) = 10.24, p < .001. However, infant cortisol did not change between T1 and T2, t(118) = −.383, p = .702. These patterns support the assumption that the maternal separation/infant challenge event was less stressful for mothers than it was for infants.

Figure 1.

Figure 1.

Open in a new tab

Average (SE) raw cortisol values for the 3-month-old infants and their mothers in the study at both time points (T1: 20–25 min after entering lab; T2: 20–25 minutes after the start of the infant cognitive task and a restful period for mother).

Baseline T1 cortisol was negatively correlated with cortisol response (ΔT2 - T1); that is, baseline scores that started high tended to result in decreasing response, and those that were lower tended to result in increasing response (all p’s < 0.05), which is consistent with the Law of Initial Values (Wilder, 1956). To compute cortisol response values that accounted for the variance on cortisol response due to T1 values, a regression procedure was used (Albers et al., 2008; Gunnar et al., 1996; Thompson et al., 2017). Specifically, regression analyses were conducted separately for infants and mothers with T2 as the dependent variable and T1 as the independent variable. The standardized residual scores were retained, and cortisol response T-scores were computed by multiplying these residuals by 10 and adding 50. T-scores averaged 50.00 (SD = 9.96) for mothers, and 49.89 (SD = 9.89) for infants.

Social Network Descriptive and Correlational Analyses

In addition to the sociodemographic characteristics reported earlier, very few of the mothers comprising the study population reported smoking (6.5%) or taking medication for depression or anxiety (3.3%), so these variables were not considered in the analyses.

Mothers reported, on average, approximately 9.9 (SD = 3.3) individuals in their network with 17.9% coming from the household, 1.5% from formal services, 3.4% from neighbors, 22.3% from friends, 6.5% from clubs/organizations, 10.6% from work/school, and 37.7% from other family not living in the same household. They also reported knowing, on average, 60.4% of their network members for more than 5 years and seeing 36.3% of their network members daily. As for the support components of the Social Network Map, mothers indicated that, on average, they received concrete support almost always from 61.4% of their network, emotional support almost always from 62.4% of their network, and information/advice almost always from 55.4% of their network. It was reported that 66.8% of the network was critical hardly ever with 10.1% being critical almost always. Mothers reported that they were very close to 65.3% of the individuals in their network and that the direction of help went both ways for 63.6% of their network members. Average scores on all social network measures appear in Table 1.

Table 1.

Average Scores (SE) on All Social Network Measures

Measure Average Score
Number of individuals in network 9.9 (0.3)
% Live in same household 17.9 (1.5)
% Family not in household 37.7 (1.7)
% Giving formal services 1.5 (0.3)
% Neighbors 3.4 (0.6)
% Friends 22.3 (1.3)
% Clubs/organizations 6.5 (1.1)
% Work/school 10.6 (1.2)
Stability
% Known > 5 years 60.4 (2.1)
% Known 1–5 years 31.7 (2.1)
% Known < 1 year 7.6 (1.3)
Frequency
% See daily 36.3 (2.1)
% See weekly 34.5 (2.0)
% See monthly 16.5 (1.5)
% See few times/year 13.3 (1.5)
Concrete support
% Give almost always 61.4 (2.6)
% Give sometimes 26.5 (2.0)
% Give hardly ever 12.0 (1.7)
Emotional support
% Give almost always 62.4 (2.4)
% Give sometimes 27.8 (2.1)
% Give hardly ever 9.9 (1.3)
Receive information/advice from
% Almost always 55.4 (2.5)
% Sometimes 32.0 (2.2)
% Hardly ever 12.4 (1.3)
Are critical of mother
% Almost always 10.1 (1.7)
% Sometimes 24.3 (2.2)
% Hardly ever 66.8 (2.8)
Closeness
% Very close to 65.3 (2.1)
% Sort of close to 27.5 (1.8)
% Not very close to 7.3 (1.0)
Direction of help goes
% Both ways 63.6 (2.3)
% Them to mother 25.3 (2.2)
% Mother to them 10.6 (1.4)
Open in a new tab

Stability of social networks was measured in terms of the percentage of individuals mother identified in her social network that she had known for over 5 years (the higher the percentage the greater the presumed stability). Stability was also measured in terms of the percentage of individuals in her network that she had known for less than one year (the higher the percentage the less the presumed stability). To determine which Social Network Map responses, aside from stability, might predict significant variance in mothers’ cortisol responses when used in the regression models, bivariate correlations were conducted between mothers’ cortisol response values, network size, and the upper-scale (e.g., give concrete support almost always) and lower-scale (e.g., give concrete support hardly ever) measures for each question. The only significant correlations obtained were those for the social network stability questions. Specifically, for mothers, having a greater percentage of individuals in their social network that they had known for over 5 years was significantly correlated with greater decreasing cortisol response, r = −0.201, p = 0.026, and the percentage of people in her network that she had known for less than one year was significantly correlated with increasing cortisol response, r = 0.178, p = 0.048 (all other p’s > .10). Thus, these two measures were used in the regression models to predict variance in cortisol response.

Bivariate correlational analyses were conducted within maternal and infant data sets separately, to examine whether mother-infant synchrony or social network stability measures were correlated with cortisol response. Mother-infant behavioral synchrony and maternal cortisol response were uncorrelated (p > 0.10). For infants, lower cortisol response to the challenge was correlated with higher maternal-infant synchrony scores (r = −0.193, p = 0.037). However, the social network stability measures were uncorrelated with cortisol response in infants (p’s > 0.10). Intercorrelations between all study variables appear separately for mothers and infants in Table 2.

Table 2.

Intercorrelations for Study Variables

Variable 1. 2. 3. 4.
1. Mother-Infant Synchrony - .03 −.04 −.10
2. Known > 5 years .06 - −.34** −.20*
3. Known < 1 year −.03 −.30** - .18*
4. Cortisol response −.19* .07 .11 -
Open in a new tab

Note. The results for mothers (n = 121) are shown above the diagonal. The results for infants (n = 117) are shown below the diagonal.

*

p < 0.05.

**

p < 0.01.

Hierarchical Linear Regression Analyses

Separate 2-step regression models were run for infants and their mothers, predicting their cortisol response during the laboratory event. The models included the variables percent of the network mother had known greater than 5 years, and percent of the network mother had known less than one year (Step 2 in the models), adjusted for potential variance due to mother-infant synchrony scores (Step 1). The models for both infants and their mothers are summarized in Table 3.

Table 3.

Hierarchical Regression Analyses Using Social Network Stability and Mother-Infant Synchrony Measures to Predict Infant and Maternal T-Scores

Model 1 R2
 Model 2 Standardized coefficients
Level variable entered Predictor variables β R 2 ΔR 2
Infant ΔT2 - T1
Step 1 0.04* -0.20*
Step 2 0.06 0.03
known > 5 years 0.12
known < 1 year 0.15
Maternal ΔT2 - T1
Step 1 0.01 -0.09
Step 2 0.07* 0.05*
known > 5 years -0.16
known < 1 year 0.13
Open in a new tab
*

p < 0.05

Note: Mother-infant synchrony was the sole variable entered at Step 1 in the models.

Social variables and infant cortisol response.

As is shown in Table 3, mother-infant behavioral synchrony accounted for significant variance in infant T-scores in Step 1 of the model. Mother-infant synchrony was a unique predictor of variance in Model 2 (β = −0.196, p = .034). After adjusting for mother-infant synchrony in Step 1, the network stability measures did not account for significant variance in infant cortisol response (both p’s > 0.10), and the combined measures did not account for significant variance (p = .058).

Social variables and maternal cortisol response.

In contrast to infants, mother-infant behavioral synchrony did not account for significant variance in maternal cortisol response in Step 1 of the model (p > 0.10), but when both of the social network stability measures were added in Step 2, the model accounted for significant variance (7%) in maternal T-scores (F (3,117) = 2.710, p = 0.048, ƒ2 = 0.054.

Discussion

The present study is the first to investigate how two very different types of social influences, maternal social networks and the degree of synchrony in mother-infant behavioral interactions, are associated with both infants’ and mothers’ cortisol responses to a laboratory event that included maternal-infant separation and infant cognitive challenge. Previous research has demonstrated lowered cortisol response to challenge in infants who are securely attached to their mothers (Gunnar et al., 1996), and in infants exposed to substitute caregivers who interact with them warmly and responsively (Gunnar et al., 1992). Using a different measure of maternal sensitivity, one that assesses maternal sensitivity with a system focusing on the bidirectional nature of mother-infant interactions (e.g. Isabella & Belsky, 1991), correlations between infant cortisol response and mother-infant behavioral synchrony were nonsignificant in several studies (Thompson & Trevathan, 2008; 2009; Thompson et al., 2015). However, there was some indication of this relationship in the 3-month-old infant group of our longitudinal study (Thompson et al., 2015), where we reported that a median split was performed on mothers’ T1 cortisol values, and for infants whose mothers were in the lower T1 group only, infant cortisol response was marginally negatively correlated with behavioral synchrony scores (p = .09). In the present study, the three-month-old infants’ dataset from Thompson et al.’s (2015) longitudinal study was analyzed, however, the cortisol response measure was calculated differently. Specifically, the present study, as well as our most recent longitudinal study (Thompson et al. 2017), used a regression procedure to control for the T1 basal level value on cortisol response, which yields cortisol response measures independent from basal measures. In addition, we investigated associations between cortisol response and behavioral synchrony in models that adjusted these contributions in the context of social network measures. In the present study, we found that mother-infant behavioral synchrony, but not mother’s social network stability, accounted for significant variance in infant cortisol down-regulation of cortisol response. We argue that the regression procedure is the most valid means of calculating cortisol response in research contexts where differential basal values influence the degree of cortisol reactivity.

In contrast to some reported research findings on non-human primates (e.g., Wiener et al., 1987), for mothers in the study, the presence of infants, and the 10-min play period to assess behavioral synchrony that occurred in-between cortisol response measures, did not affect mothers’ cortisol responses. As with the infant models, social network stability measures were included in Step 2 of the maternal models to determine if social network influences could account for variance in cortisol response beyond variance accounted for by mother-infant synchrony, and we found this to be the case. Specifically, having a greater percentage of individuals in her network whom she had known for over five years was significantly correlated with decreases in cortisol response in the laboratory setting, and having a greater percentage of individuals in her network that she had known less than a year was associated with significant increases in maternal cortisol. Hennessy et al. (2009), in their comparative study on social buffering of the stress response, argue that the nature of the relationship between individual members of a species crucially determines what type of impact that relationship will have on the HPA response. Our findings are consistent with this argument, and also demonstrate that the stability of social relationships mothers have with their social network members can have both buffering, as well as exacerbating, effects on their neuroendocrine functioning while in the presence of their infants.

The issue of how to characterize the types of social network support available to family members, and which types best alleviate stress, has been around for several decades (e.g., Unger & Powell, 1980). However, to our knowledge, this is the first study to report on the importance of social network stability for new mothers’ HPA regulation.

Limitations

Our results showed that many factors that are intended to assess the quality of an individual’s social network, including network size, and differing degrees of emotional, concrete, or informational support, were not associated with mothers’ neuroendocrine functioning in a calm laboratory setting during mildly challenging events for their infants. However, it is possible that the importance of these factors would be revealed in life contexts that are far more challenging for mothers, when knowing other people “have your back” can be brought to mind as a means of self-regulation under stress.

Conclusion

As predicted, the results of the model tests showed a developmental trend toward a greater dependence on persons outside of the mother-infant relationship for regulation of HPA response. While mothers and infants are equally committed in terms of time spent in mutual interaction, the mother-infant relationship seems to affect infants’ HPA functioning comparatively more so than it affects mothers. Taylor (2006) notes that a mother’s bond with her infant may gradually become independent of oxytocin involvement, becoming increasingly dependent on neuromechanisms for continued attachment processes. Perhaps, mothers’ reliance on cortisol as a hormone basis for attachment follows a similar developmental trajectory—future research could more directly explore this possibility.

Acknowledgements

Research at Dr. Thompson’s lab during the study period was supported by Award Number S06GM008136 from the National Institutes of Health. We would like to gratefully acknowledge the contributions of Claire Buckingham, Erica Garcia, Cindy Gutierrez-Barraza, the Mother-Infant Study staff and undergraduate students, and the mothers and infants who participated in this study. The authors report no conflicts of interest.

Appendix.

Appendix

Instructions/Script for Social Network Measure

Step One: Developing a Social Network Map

Let’s take a look at who is in your social network by putting together a network map. (Show the network map, which is a pie chart with these categories labeling each “pie” segment: Household, Other Family, Work/School, Clubs/Organizations/Church, Friends, Neighbors, Formal Services). We can use first names or initials because I’m not interested in knowing the particular people. I won’t be contacting any of the people we’ve talked about.

Think back to this past month. What people have been important to you? They may have been people you saw, talked with, or wrote letters to. This includes people who made you feel good, people who made you feel bad, and others who just played a part in your life. They may be people who had an influence on the way you made decisions during this time.

There is no right or wrong number of people to identify on your map. Right now, just list as many people as you come up with. Do you want me to write or do you want do the writing?

First, think of people in your household—that includes who?

Now, going around the map, what other family members would you include in your network?

How about people from work or school?

People from clubs, organizations or religious groups? Who should we include here?

Other friends that haven’t been listed in the other categories?

Neighbors?

Finally, list professional people or people from formal agencies (like CYFD) that you have contact with.

Look over your network. Are these the people you would consider part of your social network this past month? (Add or delete names as needed).

Step Two: Completing the Social Network Grid: (If there are more than 15 people in the network, ask the participant to select the ‘top fifteen’ and then ask the questions about only those network members. For each of the questions use the appropriate sorting guide card. Once the participant has divided up the cards, put the appropriate code number for each person listed on the network grid).

Now, I’d like to learn more about the people in your network. I’m going to write their names on this network grid, put a code number for the area of life, list their ages, and then as ask a few questions about the ways in which they help you. Let’s also write their names on these notecards too; this will make answering the questions a lot easier. We’ll check off the names on this grid as we go through each question.

The first three questions have to do with the types of support people give you.

Who would be available to help you out in real concrete ways—for example, would give you a ride if you needed one, or would pitch in to help you with a big chore, or would look after your belongings for awhile if you were away? Divide your notecards into three piles—those people you can hardly every rely on for concrete help, those you can rely on sometimes, and those you’d almost always rely on for this type of help. Think of hardly ever as meaning less than 10% of the time and almost always as meaning over 90% of the time. Also, base your answers on how available help would be if you were to ask for help.

Now who would be available to give you emotional support—for example, to comfort you if you were upset, to be right there with you in a stressful situation, to list to you talk about your feelings? Again, divide your notecards into three piles—those people you can hardly ever rely on for emotional support, those you can rely on sometimes, and those you almost always can rely on for this type of help.

Finally, who do you rely on for advice—for example, who would give you information on how to do something, help you make a big decision, or teach you how to do something? Divide your notecards into three piles—hardly ever, sometimes and almost always—for this type of support.

Look through your cards and this time select those people, if any, in your network that you feel are critical of you (either critical of you, or your life style, or you as a parent). When I say “critical”, I mean critical of you in a way that makes you feel bad or inadequate. Divide the notecards into three piles—those people who are hardly ever critical of you, sometimes critical of you, and almost always critical of you. Again, we’ll put the code numbers next to their names.

Now look over your notecards, and think about the direction of help. Divide your cards into three piles—those people where help goes both ways (you help them just as much as they help you), those where it’s mostly you helping them, and those where it’s mostly them helping you. Ok, let’s get their code numbers on the grid.

Now think about how close you are to the people in your network. Divide the notecards into three piles—those you are not very close to, those you are sort of lose to, and those you are very close to—and then we’ll put a code number for them.

Finally, just a few questions about how often you have contact with people, and how long you have known the people in your network.

Divide the notecards into four piles—people with whom you have contact just a few times a year, monthly, weekly and daily. (NOTE: If you have contact with someone twice or more than twice a week, count that as “daily”). Ok, we’ll put their numbers on the grid.

This is the last question. Divide the notecards into three piles—those people you have known less than a year, from 1 – 5 years, and over 5 years.

Now we have a pretty complete picture of who is in your social network.

References

  1. Albers EM, Riksen-Walraven JM, Sweep FCG, & de Weerth C. (2008). Maternal behavior predicts infant cortisol recovery from a mild everyday stressor. Journal of Child Psychology and Psychiatry, 49, 97–103. [DOI] [PubMed] [Google Scholar]
  2. Beijers R, Buitelaar JK, & de Weerth C. (2014). Mechanisms underlying the effects of prenatal psychosocial stress on child outcomes: Beyond the HPA axis. European Child & Adolescent Psychiatry, 23, 943–956. [DOI] [PubMed] [Google Scholar]
  3. Bodenmann G, Meuwly N, Germann J, Nussbeck FW, Heinrichs M, & Bradbury TN (2015). Effects of stress on the social support provide by men and women in intimate relationships. Psychological Science, 26, 1584–1594. [DOI] [PubMed] [Google Scholar]
  4. Bronfrenbrenner U. (1979). The ecology of human development. Cambridge, Mass.: Harvard University Press. [Google Scholar]
  5. Bronfrenbrenner U. (1986). Ecology of the family as a context for human development: Research perspectives. Developmental Psychology, 22(6), 723–742. [Google Scholar]
  6. Buchanan TW, Bagley SL, Stansfield RB, & Preston SD (2012). The empathic, physiological resonance of stress. Social Neuroscience, 7, 191–201. [DOI] [PubMed] [Google Scholar]
  7. Crnic KA, Greenberg MT, Ragosin AS, Robinson NM, & Basham RB (1983). Effects of stress and social support on mothers and premature and full-term infants. Child Development, 54, 209–217. [PubMed] [Google Scholar]
  8. Crockenberg SB (1981). Infant irritability, mother responsiveness, and social support influences on the security of infant-mother attachment. Child Development, 52, 857–865. [PubMed] [Google Scholar]
  9. Degnan A, Berry K, Sweet D Abel K, Crossley N, & Edge D. (2018). Social networks and symptomatic and functional outcomes in schizophrenia: A systematic review and meta-analysis. Social Psychiatry and Psychiatric Epidemiology, 53, 873–888. 10.1007/s00127-018-1552-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. deWeerth C, Zijl RH, & Buitelaar JK (2003). Development of cortisol circadian rhythm in infancy. Early Human Development, 3, 39–52. [DOI] [PubMed] [Google Scholar]
  11. Eisenberger NI, Taylor SE, Gable SL, Hilmert CJ, & Lieberman MD (2007). Neural pathways link social support o attenuated neuroendocrine stress responses. NeuroImage, 35, 1601–1612. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Enlow MB, King L, Schreier HMC, Howard JM, Rosenfield D, Ritz T, & Wright RJ (2014). Maternal sensitivity and infant autonomic and endocrine stress responses. Early Human Development, 90(7), 377–385. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Feldman PJ, Dunkel-Schetter C, Sandman CA, & Wadhwa PD (2000). Maternal social support predicts birth weight and fetal growth in human pregnancy. Psychosomatic Medicine, 62, 715–725. [DOI] [PubMed] [Google Scholar]
  14. Feldman R, Weller A, Zagoory-Sharon O, & Levine A. (2007). Evidence for a neuroendocrinological foundation of human affiliation. Psychological Science, 18, 965–970. [DOI] [PubMed] [Google Scholar]
  15. Fleming AS, Steiner M, & Corter C. (1997). Cortisol, hedonics, and maternal responsiveness in human mothers. Hormones and Behavior, 32, 85–98. [DOI] [PubMed] [Google Scholar]
  16. Giesbrecht GF, Poole JC, Letourneau N, Campbell T, & Kaplan BJ (2013). The buffering effect of social support on hypothalamic-pituitary-adrenal axis function during pregnancy. Psychosomatic Medicine, 75, 856–862. [DOI] [PubMed] [Google Scholar]
  17. Gunnar MR (1992). Reactivity of the hypothalamic-pituitary-adrenocortical system to stressors in normal infants and children. Pediatrics, 9, 491–497. [PubMed] [Google Scholar]
  18. Gunnar MR, Brodersen L, Nachmias M, Buss K, & Rigatuso J. (1996). Stress reactivity and attachment security. Developmental Psychology, 29, 191–204. [DOI] [PubMed] [Google Scholar]
  19. Gunnar MR, Larson MC, Hertsgaard L, Harris ML, & Brodersen L. (1992). The stressfulness of separation among nine-month-old infants: Effects of social context variables and infant temperament. Child Development, 63(2), 290–293. [PubMed] [Google Scholar]
  20. Hahn-Holbrook J, Schetter CD, Arora C, & Hobel CJ (2013). Corticotropin-releasing hormone mediates the association between prenatal social support and postpartum depression. Clinical Psychological Science, 1(3), 253–264. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Haley DW, Weinberg J, & Grunau RE (2006). Cortisol, contingency learning, and memory in preterm and full-term infants. Psychoneuroendocrinology, 31, 108–117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Heinrichs M, Baumgartner T, Kirschbaum C, & Ehlert U. (2003). Social support and oxytocin interact to suppress cortisol and subjective responses to psychosocial stress. Biological Psychiatry, 54, 1389–1398. [DOI] [PubMed] [Google Scholar]
  23. Hennessy MB, Kaiser S, & Sachser N. (2009).Social buffering of the stress response: Diversity, mechanisms, and functions. Frontiers in Neuroendocrinology, 30, 470–482. [DOI] [PubMed] [Google Scholar]
  24. Isabella RA, & Belsky J. (1991). Interactional sensitivity and the origins of infant-mother attachment: A replication study. Child Development, 62, 373–384. [PubMed] [Google Scholar]
  25. Isabella RA, & Belsky J, & von Eye A. (1989). Origins of infant-mother attachment: An examination of interactional synchrony during the infant’s first year. Developmental Psychology, 25(1), 12–21. [Google Scholar]
  26. Jewell SL, Luecken LJ, Gress-Smith J, Crnic KA, & Gonzales NA (2015). Economic stress and cortisol among postpartum low-income Mexican American women: Buffering influence of family support. Behavioral Medicine, 41, 138–144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kikusui T, Winslow JT & Mori Y. (2006). Social buffering: Relief from stress and anxiety. Philosophical Transactions of the Royal Society, 361, 2215–2228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kirschbaum C, Klauer T, Filipp S, & Hellhammer DH (1995). Sex-specific effects of social support on cortisol and subjective responses to acute psychological stress. Psychosomatic Medicine, 57, 23–31. [DOI] [PubMed] [Google Scholar]
  29. Kroenke CH, Quesenbeery C, Kwan ML, Sweeney C, Castillo A, & Caan BJ (2013). Social networks, social support, and burden in relationships, and mortality after breast cancer diagnosis in the Life After Breast Cancer Epidemiology (LACE) Study. Breast Cancer Research and Treatment, 137, 261–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Larson MC, White BP, Cochran A, Donzella B, & Gunnar MR (1998). Dampening of cortisol response to handling at 3 months in human infants and its relation to sleep, circadian cortisol activity, and behavioral distress. Developmental Psychobiology, 33, 327–337. [PubMed] [Google Scholar]
  31. Levitt MJ, Weber RA, & Clark MC (1986). Social network relationships as sources of maternal support and well-being. Developmental Psychology, 22, 310–316. [Google Scholar]
  32. Lewis M, & Ramsay DS (1995). Developmental change in infants’ responses to stress. Child Development, 66, 657–670. [DOI] [PubMed] [Google Scholar]
  33. Lin N, Ensel WM, Simeone RS, & Kuo W. (1979). Social support stress life events, and illness: A model and an empirical test. Journal of Health and Social Behavior, 20, 108–119. [PubMed] [Google Scholar]
  34. Nath PS, Borkowski JG, Whitman TL, & Schellenbach CJ (1991). Understanding adolescent parenting: The dimensions and functions of social support. Family Relations, 40, 411–420. [Google Scholar]
  35. Ozbay R, Fitterling H, Charney D, & Southwick S. (2008). Social support and resilience to stress across the life span: A neurobiologic framework. Current Psychiatry Reports, 10, 304–310. [DOI] [PubMed] [Google Scholar]
  36. Ramsay D, & Lewis M. (2003). Reactivity and regulation in cortisol and behavioral responses to stress. Child Development, 74, 456–464. [DOI] [PubMed] [Google Scholar]
  37. Salimetrics (2012). ‘Cortisol determination: Cat No: 1–3002 research kit; Cat No: 1–3102 diagnostic kit’ in High sensitivity salivary cortisol enzyme immunoassay kit. State College, PA: Salimetrics, LLC. [Google Scholar]
  38. Stetler CA, & Miller GE (2008). Social integration of daily activities and cortisol secretion: A laboratory based manipulation. Journal of Behavioral Medicine, 31, 249–257. [DOI] [PubMed] [Google Scholar]
  39. Taylor SE, (2002). The tending instinct: How nurturing is essential to who we are and how we live. New York: Holt. [Google Scholar]
  40. Taylor SE. (2006). Tend and befriend: Biobehavioral bases of affiliation under stress. Current Directions in Psychological Science, 15, 273–277. [Google Scholar]
  41. Taylor SE, Klein LC, Lewis BP, Gruenewald TL, Gurung RAR, & Updegraff JA (2000). Biobehavioral responses to stress in females: Tend-and-befriend, not fight-or-flight. Psychological Review, 107, 411–429. [DOI] [PubMed] [Google Scholar]
  42. Thompson LA, Morgan G, & Jurado KA (2015). A longitudinal study of infant cortisol response during learning events, Monographs of the Society for Research in Child Development, 80(4), vii-viii, 1–122. [Google Scholar]
  43. Thompson LA, Morgan G, Unger CA, & Covey LA (2017). Prenatal maternal cortisol measures predict learning and short-term memory performance in 3- but not 5-month-old infants. Developmental Psychobiology. 59, 723–737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Thompson LA, & Trevathan WR (2009). Cortisol reactivity, maternal sensitivity, and infant preference for mother’s familiar face and rhyme in 6-month-old infants. Journal of Reproductive and Infant Psychology, 27(2), 143–167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Thompson LA, & Trevathan WR (2008). Cortisol reactivity, maternal sensitivity, and learning in 3-month-old infants. Infant Behavior and Development, 31, 92–106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Tracy EM & Whittaker JK (1990). The Social Network Map: Assessing social support in clinical practice. Families in Society, 71(8), 461–470. [Google Scholar]
  47. Unger DG, & Powell DR (1980). Supporting families under stress: The role of social networks. Family Relations, 29(4),566–574. [Google Scholar]
  48. Vandeleest JJ, Capitanio JP, Hamel A, Meyer J, Novak M, Mendoza SP, & McCowan B. (2019). Social stability influences the association between adrenal responsiveness and hair cortisol concentrations in rhesus macaques. Psychoneuroendocrinology, 100, 164–171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Waters SR, West TV, & Mendes WB (2014). Stress contagion: Physiological covariation between mothers and infants. Psychological Science, 25, 934–942. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Wiener SG, Johnson DF, & Levine S. (1987). Influence of postnatal rearing conditions on the response of squirrel monkey infants to brief perturbations in mother-infant relationships. Physiology & Behavior, 39, 21–26. [DOI] [PubMed] [Google Scholar]
  51. Wilder J. (1956). The law of initial value in neurology and psychiatry. Journal of Nervous and Mental Diseases, 125, 73–86. [DOI] [PubMed] [Google Scholar]

RESOURCES