Affective and Autonomic Reactivity During Parent–Child Interactions in Depressed and Non-Depressed Mothers and Their Adolescent Offspring

Abstract

Depression presents risks that are profound and intergenerational, yet research on the association of depression with the physiological processes that might be associated with impaired mental and physical health has only recently been contextualized within the family environment. Participants in this multi-method case–control study were 180 mother-adolescent dyads (50% mothers with a history of depression treatment and current depressive symptoms). In order to examine the association between maternal depression and affective and autonomic reactivity amongst these mothers and their adolescent offspring we collected self-reported measures of positive and negative affect, as well as measures of cardiovascular and electrodermal autonomic activity, during mother-adolescent interaction tasks. Findings indicated that depressed mothers and their adolescent offspring exhibited greater self-reported negative affect reactivity during a problem-solving interaction and blunted (i.e., low) sympathetic activity as measured via skin conductance level across both interaction tasks. These effects remained significant after controlling for a range of potential covariates, including medication use, sex, age, adolescents own mental health symptoms, and behavior of the other interactant, along with correcting for multiple comparisons. Findings indicate that depressed mothers and their adolescent offspring both exhibit patterns of affect and physiology during interactions that are different from those of non-depressed mothers and their offspring, including increased negative affect reactivity during negative interactions and blunted sympathetic activity across both positive and negative interactions. These findings have potential implications for understanding the role of family processes in the intergenerational transmission of risk for depressive disorders.

Introduction

Depression is associated with significant impairment, morbidity, and mortality, which presents risks that are profound and intergenerational. When those who are affected are parents, it adversely impacts family interactional processes and relationships, creates transdiagnostic psychiatric risk amongst offspring, and has adverse physical health outcomes for both depressed individuals and their offspring (Nelson et al., 2021; Nemeroff & Goldschmidt-Clermont, 2012; Raposa et al., 2014). Importantly, research on the association of depression with physiological processes that might be associated with impaired mental and physical health has only recently been contextualized within the family environment. This limits our understanding of how risk is embedded in potentially modifiable daily processes such as family interactions (Kudinova et al., 2019; McMakin et al., 2011; Woody et al., 2016). It is particularly important to study these effects when offspring are adolescents, as this is a developmental period associated with increased risk for onset of mental disorders such as depression (Zisook et al., 2007), increased sensitivity to social and emotional learning (Dahl et al., 2018) and heightened physiological reactivity to stressors (Stroud et al., 2009). In this study, we examined the association between maternal depression and affective and autonomic reactivity amongst mothers and their adolescent offspring —variables with significance for both mental and physical health—during parent-adolescent interactions. However, before turning to the relationship between maternal depression and adolescent affective and psychophysiological outcomes, we first review the affective and underlying autonomic psychophysiological processes associated with depression more generally.

Association of Depression with Affective and Autonomic Stress Reactivity

Affective Reactivity.

Depression has been generally found to be associated with higher levels of baseline negative affect and lower baseline positive affect (Davidson et al., 2002; Forbes et al., 2004; Watson, 2000), but the literature is mixed when considering affective reactivity. In a systematic quantitative meta-analysis, Bylsma and colleagues (2008) concluded that the experimental evidence supported an emotion context insensitivity (ECI; Rottenberg, 2007; Rottenberg et al., 2005) model whereby depressed individuals show reduced reactivity to both positive and negative affective stimuli. However, most studies included in this meta-analysis used stimuli of low personal relevance, such as generic pleasant or unpleasant pictures or film clips and some research indicates that this ECI profile may be limited to lab-based research (Nelson et al., 2018; Thompson et al., 2012) and may not apply to real-world personally relevant contexts, particularly in regard to negative affect. For example, research shows that those with depression have greater NA reactivity to negative life circumstances (Bylsma et al., 2011; Myin-Germeys et al., 2003; van der Stouwe et al., 2019; van Winkel et al., 2015). Overall, there are remaining questions regarding the role that depression plays in affective reactivity, including whether reactivity differs between specific types of affect (i.e., positive and negative affect), and whether emotional valence of the interpersonal interaction (e.g., positive interaction or negative interaction) influences affective reactivity.

Autonomic Reactivity.

Depression is also associated with greater psychophysiological stress reactivity (Bylsma et al., 2008; Koval & Kuppens, 2012; Koval et al., 2012; Nelson et al., 2017) of the autonomic nervous system (ANS), which is comprised of both the energy-expending sympathetic nervous system (SNS) and the energy-conserving parasympathetic nervous system (PNS). The ANS can be activated in response to real or imagined threat and challenge, including psychosocial stressors. Most research has found blunted (i.e., low) autonomic nervous system (ANS) responses as depressive symptoms increase (Carroll et al., 2007; Phillips, 2011; Salomon et al., 2009; Schwerdtfeger & Rosenkaimer, 2011; Stange et al., 2017; York et al., 2007), although this is not found in all studies (Kibler & Ma, 2004; Light et al., 1998; Matthews et al., 2005; Stange et al., 2017). As Schwerdtfeger and Rosenkaimer (2011) point out, these mixed findings may be due to type of stressor (active vs. passive). Active stressors, including interpersonal interactions (e.g., public speaking tasks), result in a beta-adrenergic response that yield blunted autonomic reactivity (Carroll et al., 2007; Phillips, 2011; Salomon et al., 2009; York et al., 2007). Such a blunted pattern of biological disengagement has been confirmed in recent reviews of depression and autonomic reactivity (Ginty, 2013; Schiweck et al., 2019), specifically for sympathetic activity (Sarchiapone et al., 2018), and is in line with a hypothesized motivational deficit in those with depression (Salomon et al., 2013; Schwerdtfeger & Rosenkaimer, 2011). Furthermore, Schwerdtfeger and Rosenkaimer (2011) point out that passive stressors (e.g., mirror tracing task) may engender an alpha-adrenergic response that results in increased autonomic reactivity (Matthews et al., 2005). Overall, more research is needed to replicate these patterns of blunted ANS reactivity during active interpersonal stressors and to examine reactivity during positive interactions.

Maternal Depression as a Potential Risk Factor for Offspring Affect and Autonomic Reactivity

As mentioned above, depression not only presents profound risks for the individual, but also intergenerational risks, with adverse health outcomes for both depressed persons and their offspring. There are various potential pathways through which maternal depression may act as a risk factor for offspring affective and autonomic reactivity. These include both genetic mechanisms, which may account for up to 40% of the variance in depression (Goldberg, 2006; Levinson, 2006), as well as adverse environmental factors, including exposure to higher rates of negative parenting behaviors displayed by depressed mothers (Beck, 1995; Leinonen et al., 2003; Norcross et al., 2017)— all of which are likely to be experienced as stressful by offspring, potentially leading to affective and autonomic dysregulation.

Adolescence may be a particularly important developmental period during which to examine these effects as it is characterized by increased sensitivity to social information (Dahl et al., 2018) and increased physiological reactivity to stressors when compared to childhood and adulthood (Stroud et al., 2009). Together these psychobiological alterations may assist in explaining why the adolescent years are associated with heightened risk for the first onset of depression (Zisook et al., 2007). Few studies have examined the association between maternal depression and affective or ANS reactivity in offspring (Bleker et al., 2018), however extant findings have reported that prenatal and early postnatal exposure to maternal depression is associated with greater autonomic (e.g., systolic blood pressure) reactivity in middle childhood (Fan et al., 2016), and adolescence (Vedhara et al., 2012), although consistent findings are not observed in all studies (Rash et al., 2015). Furthermore, some recent research has shown that exposure to prenatal maternal depression is associated with blunted child sympathetic (i.e., electrodermal) activity (Buthmann et al., 2019). Research on offspring exposed to maternal depression later in life has demonstrated mixed findings with some studies showing greater physiological reactivity to stress (Burkhouse et al., 2014; Gump et al., 2009) and some not (Waugh et al., 2012), though the latter study did show greater NA reactivity. Overall, such autonomic reactivity may have a role in the transmission and generation of depression symptoms across time (Carnevali et al., 2018). The broad range of ages, stimuli, and outcome measures, in conjunction with the relatively small evidence base, moreover, makes it difficult to draw clear conclusions. More research is needed that examines both branches of the ANS as well as positive and negative affect during both positive and negative interactions in order to tease apart prior mixed findings.

The Current Study

A deeper understanding of the patterns of affective and autonomic reactivity in depressed mothers and their offspring is necessary not only to understand the impact of depression on affective functioning, but also for understanding the mechanisms of intergeneration transmission of risk for psychopathology and identifying targets for prevention and early intervention. The current study adds to this understanding by examining cross-sectional differences in affective and autonomic reactivity in mothers with and without depression (Depressed and Non-Depressed Group) and their adolescent offspring (High Risk Group for those with mothers experiencing depression and Lower-Risk Group for those with mothers not experiencing depression). We examined reactivity during two mother-adolescent interaction tasks designed to differentially elicit positive (Event-Planning Interaction; EPI) and negative (Problem-Solving Interaction; PSI) interpersonal processes, and which we have found to elicit differential autonomic responses in past research (Nelson et al, 2017). We examined tasks that elicit both PA and NA, given that both positive and negative contexts may be stressful for depressed persons (Forbes et al., 2004; Watson, 2000), and that our prior work has revealed that harsh parental behavior during putatively positive affective contexts have adverse associations with adolescent emotional health (Schwartz et al., 2013).

Affect Hypotheses.

We hypothesized that during both positive and negative interactions, Depressed mothers and their High-Risk offspring would report blunted PA reactivity compared to Non-Depressed and Lower-Risk Groups in accordance ECI theory and past research. We hypothesized that, across interactions, Depressed mothers would show greater NA reactivity compared to Non-Depressed mothers based on research indicating greater stress sensitivity and reactivity to social stimuli. In addition, we hypothesized that across interactions, High-Risk offspring would show greater NA reactivity as maternal depression may act as a form of social threat for adolescents that results in increased affective reactivity during mother–child interactions.

Autonomic Hypotheses.

We hypothesized that across interactions, Depressed mothers would be less reactive – as indicated by lower overall ANS reactivity (i.e., heart rate), less SNS reactivity (i.e., shortening of pre-ejection period), less PNS reactivity (i.e., greater heart rate variability and lower skin conductance level) compared to Non-Depressed and Lower-Risk Groups as prior research has shown that active stressors result in blunted autonomic activity in those with depression. Alternatively, we acknowledge the possibility that depressed mothers may be more sensitive to social rejection and may be particularly impacted by negative adolescent behaviors during interactions resulting in greater ANS reactivity. For example, the dramatic changes to parent–child relationships that occur during adolescence pose unique parenting challenges, with most parents finding it to be the most stressful and least rewarding period (Steinberg & Silk, 2002). The nature of these changes—increased argumentativeness and disrespectful behavior, decreased engagement in family life, and increased demands for autonomous decision-making– likely play on the sensitivity to rejection that is characteristic of depression (Gao et al., 2017), potentially, exacerbating the developmentally typical increase in parent–child conflict during this period. We do not have specific hypotheses to how this may differ by type of interaction task (e.g., positive or negative) and therefore, this is an exploratory analysis.

In contrast, we hypothesized that across interactions, adolescents of depressed mothers (High-Risk Group) would have greater heart rate reactivity, more shortening of pre-ejection period (i.e., shortening of interval), less heart-rate variability, and greater skin-conductance reactivity relative to adolescents of non-depressed mothers (Lower-Risk Group) as our adolescent hypotheses are based on the premise that maternal depression acts as a form of social threat for adolescents that results in increased physiological reactivity during mother child interactions. Again, due to a lack of evidence, we do not have specific hypotheses to how this may differ by type of interaction task (e.g., positive or negative) and therefore, this is an exploratory analysis.

Methods and Materials

Participants

Participants were 180 low-income women (see Table 1) and their adolescent children, aged 11–14 (see Table 2). Two groups of women were recruited: a Depressed Group, selected for currently elevated depressive symptoms (PHQ-9 cut-off score > 10; mean = 12.32, SD = 5.84) and a history of treatment for depression at any point in time and with any treatment type (note this group was not selected on depressive diagnosis, but rather current elevated depressive symptoms as literature shows that elevated depressive symptoms effect parenting adversely regardless of diagnostic status),¹ and a Non-Depressed Group, selected for no or low levels of current depressive symptomatology (PHQ-9 cut-off score < 8; mean = 2.57, SD = 2.70), no history of treatment for depression, and no current (i.e., past month) mental health treatment for any mental health disorder. Adolescents had to be between the ages of 11–14 and those of mothers in the Depressed Group were placed in a High Risk Group, while adolescents of mothers in the Non-Depressed Group were placed in the Lower Risk Group (note that we named this “Lower Risk,” rather than “Low Risk” as both groups came from a low-income sample). Exclusion criteria for participants of both groups included psychosis, physical illness, or cognitive impairment that would interfere with participation (e.g., substance use that would render abstinence for the assessment difficult to tolerate). There were no statistically significant group differences in adolescent biological sex, adolescent age, race, and ethnicity or maternal education, employment, or income.

Table 1.

Maternal Characteristics by Group

	Depressed			Non-Depressed			Group Difference
Variable	N	Mean (SD)	Percentage	N	Mean (SD)	Percentage	Test statistic
Group	90		100%	90		100%
Age	90	40.6 (6.5)		90	40.0 (6.4)		t(177.94) = 0.70, p = 0.48, Hedge’s g = 0.10
Mental Health
PHQ-9	90	12.32 (5.84)		89	2.57 (2.70)		t(125.75) = 14.37, p < 0.001. Hedge’s g = 2.14
Depressive Dx	64		71.11%	0		0.0%
Anxiety Dx	36		40.0%	3		3.3%	t(135.36) = 11.36, p < 0.001, Hedge’s g = 1.69
Income							χ2(7) = 6.48, p = 0.48, Cramer’s V = 0.00
<$17,000	25		27.78%	16		18.18%
$17,000—$19,999	7		7.78%	13		14.77%
$20,000—$24,999	14		15.56%	8		9.09%
$25,000—$29,999	11		12.22%	10		11.36%
$30,000—$34,999	5		5.56%	6		6.82%
$35,000—$39,999	6		6.67%	9		10.23%
$40,000—$49,999	8		8.89%	10		11.36%
> = $50,000	14		15.56%	16		18.18%
Don’t Know	0		0.00%	2		2.22%

GAD-7 Generalized Anxiety Disorder-7, PHQ-9 Patient Health Questionnaire-9

Table 2.

Adolescent Characteristics by Group

	High Risk			Low Risk			Group Difference
Variable	N	Mean (SD)	Percentage	N	Mean (SD)	Percentage	Test Statistic
Group	90		100%	90		100%
Age	90	12.90 (1.30)		90	12.90 (1.20)		t(177.91) = 0.04, p = 0.97, Hedge’s g = 0.01
YSR Total Symptoms	88	54.30 (10.10)	97.78%	87	47.90 (9.80)	96.67%	t(172.95) = 4.24, p < 0.001. Hedge’s g = 0.64
Sex							χ2(1) = 0.80, p = 0.37, Cramer’s V = 0.00
Male	45		50%	51		56.67%
Female	45		50%	39		43.33%
Gender							χ2(2) = 1.64, p = 0.44, Cramer’s V = 0.00
Male	45		50.00%	52		57.78%
Female	44		48.89%	36		40.00%
Other	1		1.11%	2		2.22%
Race							χ2(5) = 5.72, p = 0.33, Cramer’s V = 0.06
White or Caucasian	75		83.33%	67		74.44%
More than One Race	15		16.67%	18		20.00%
American Indian/Alaska Native	0		0.00%	1		1.11%
Native Hawaiian/Pacific Islander	0		0.00%	1		1.12%
African American	0		0.00%	1		1.11%
No Response/Unknown	0		0.00%	2		2.22%

Recruitment

In order to recruit a low-income sample, the majority of participants (n = 132) were recruited through the organization that administers the Oregon Health Plan (Medicaid) in the county where data were collected. The remainder of the sample (n = 48) were recruited through online advertisements; those recruited online were screened to ensure that their incomes would have rendered them eligible for Medicaid. This low-income sample was selected, because mental health problems such as depression are more prevalent in these groups. There were significant group differences in recruitment source, such that more mothers in the depressed group were recruited online, X² (1) = 12.53, p < 0.001. Mothers and adolescents provided informed consent and assent, respectively, prior to assessment. All procedures were approved by the Institutional Review Board at Oregon Research Institute.

Assessment Procedures

After an informed-consent meeting, mothers completed an online questionnaire through a secure link and diagnostic interviews were conducted over the phone by research assistants. Subsequently, mothers and adolescents participated in a laboratory assessment where participants completed questionnaires and were outfitted with ambulatory electrocardiography (ECG) and impedance cardiography (ICG) devices to record psychophysiological indices during a 2-min baseline and two 15-min interaction tasks. One task was an Event-Planning Interaction (EPI) in which participants were asked to plan a vacation they would like to take together. The second task was a Problem-Solving Interaction (PSI), in which dyads were asked to discuss and try to resolve one or two areas of conflict selected from the Issues Checklist (Prinz et al., 1979). Topics chosen for discussion were those with the highest mean frequency by intensity ratings across mother and adolescent reports. These tasks have been shown to differentially elicit observed positive and negative affect, respectively (Nelson, et al., 2017), and did so in the current study for both mothers and adolescents (see supplementary materials). The ordering of tasks was counterbalanced, and separated by a puzzle task to reduce affect contagion or carry-over effects from one task to the next.

Measures

Diagnostic Measure.

Mothers completed the Structured Clinical Interview, non-patient version (First et al., 1996) in order to characterize the sample and ensure that participants in the Non-Depressed Group did not meet criteria for depressive disorders. Interrater reliability was kappa = 0.80. See Table 1 for rates of depressive, anxiety, and trauma/stress diagnoses.

Self-Report of Mental Health Symptoms.

Mothers completed the Patient Health Questionnaire-9 (PHQ-9; Kroenke et al., 2001), which had acceptable reliability (α = 0.92). Adolescents completed the Youth Self Report (Achenbach, 1991), which had had acceptable reliability (α = 0.95).

Affect Assessment.

At the start and conclusion of each interaction, mothers completed the Positive and Negative Affect Schedule (Watson et al., 1988) and adolescents completed the Positive and Negative Affect Schedule for Children (PANAS-C; Laurent et al., 1999), respectively, to provide indices of state level positive affect (e.g., energy, interest, and engagement) and negative affect (e.g., emotional distress) in order to measure affective reactivity across interaction tasks. See supplementary materials for descriptive affect statistics for each task. For adolescents Cronbach’s Alpha was 0.64 for pre-EPI NA, 0.90 for pre-EPI PA, 0.85 for post-EPI NA, 0.90 for post-EPI PA, 0.59 pre-PSI NA, 0.90 pre-PSI PA, 0.57 post-PSI NA, and 0.92 post-PSI PA. For mothers Cronbach’s Alpha was 0.67 for pre-EPI NA, 0.92 for pre-EPI PA, 0.71 for post-EPI NA, 0.93 for post-EPI PA, 0.77 for pre-PSI NA, 0.92 for pre-PSI PA, 0.82 for post-PSI NA, and 0.91 for post-PSI PA.

Biological Measures

Psychophysiological Assessment.

ECG and ICG data were acquired using Vrije Universiteit Ambulatory Monitoring System (VU-AMS), which uses a 3-lead ECG and 4-lead ICG. Electrodes were placed in line with the VU-AMS manual instructions. Data were scored using the Data Analysis and Management Software (DAMS) program (http://www.vu-ams.nl/). To obtain the resting baseline assessment, participants were asked to sit quietly for a 2-min period. Data were averaged over this 2-min period.

Heart Rate was calculated based on time (in milliseconds) between successive R waves (R-R intervals) on the ECG to calculate average beats per minute. The root mean square of successive R-R interval differences (RMSSD), which has been shown to reflect vagal tone (Thayer & Lane, 2007; Kleiger et al., 2005; Laborde, Mosley, & Thayer, 2017) was calculated to measure parasympathetic nervous system activity. This measure of HRV was chosen as it provides reliable estimates of HRV across different duration of recordings (Laborde et al., 2017), such that 1 min recordings of the natural log of RMSSD has good reliability relative to 5 min RMSSD (Esco & Flatt, 2014) and because recordings longer than 120 s have been shown to be unnecessary to record accurate measures of RMSSD (Munoz et al., 2015) allowing this measure to accurately capture HRV during the brief 2-min baseline. Furthermore, while RMSSD is highly correlated with high-frequency heart rate variability (HF-HRV; Kleiger et al., 2005) and is relatively free of respiratory influences, unlike high frequency parameters (Hill et al., 2009). This may be particularly important during interaction tasks like ours that require speech. Cardiac pre-ejection period (PEP), a marker of sympathetic nervous system (SNS) activity, was automatically calculated for each cardiac cycle as the time interval in msec between the onset of ventricular depolarization (Q wave onset of the ECG) and the opening of the aortic valves (B point in the ICG dZ/dt signal). Skin conductance level (SCL), a second measure of SNS activity, was automatically calculated based on the electrodermal activity of the skin on the pointer and middle fingers, which was measured using direct current utilizing a 16 bit A/D converter and we did not quantify phasic changes. The sampling rate was 10 Hz with a signal range of 0–95 micro Siemens (μS). The SCL signal was pre-processed to remove the noise and power-line interference in the signal. A low-pass filter with cut-off frequency of 2 Hz was used to filter the signal. In order to avoid shifting of peaks, filtering was done both in forward and reverse directions. The signal was then visually inspected for artifacts and averaged over the 2 min period.

Covariates.

We collected the following measures as potential covariates: age, self-reported biological sex, ethnicity, race, body mass index, and medication use as these variables have been shown to influence psychophysiology measures (Hill et al., 2015; Koenig et al., 2014; Quintana et al., 2016). Only covariates that were significantly associated with outcome variables were included in analyses in order to prevent overfitting. In addition, each model controlled for baseline measures of psychophysiology and affect in order to index reactivity or change from baseline.

Statistical Analyses

All statistical analyses were conducted with R Studio, version 4.0.2. Statistical significance was defined using 95% confidence intervals and p-values. Exploratory analyses including histograms as well as skew and kurtosis statistics were run for each variable to check for normality. All variables were winsorized to ± 3 SD to correct for outliers and any variable that then had a skew of ± 2 was log transformed. All tables for unadjusted and adjusted models as well as figures for null results can be found in supplementary material.

In all analyses we define reactivity as change in construct of interest from baseline or pre-task to task-level. To assess differences in reactivity a series of multilevel models were performed for each cardiovascular measure (HR, RMSSD, PEP, SCL) and affect measure (positive and negative) separately for adolescents (High-Risk and Lower-Risk Groups) and mothers (Depressed and Non-Depressed Groups). For each model, intercepts were allowed to vary by participant. For each of the adolescent analyses, three separate models were run for each outcome adding more stringent covariates, which included 1) unadjusted model with only baseline physiology or affect, group, task, and group x task as fixed effects, 2) adjusted model with age, biological sex, and medication entered as fixed effects, and 3) adjusted model with adolescent’s own mental health symptoms added as a fixed effect in addition to prior covariates in order to covary out any influence adolescents’ own mental health had on outcome measures. For each of the mother analyses, two separate models were run, and for each more stringent covariates were successively added, which included 1) unadjusted model with only baseline physiology or affect, group, task, and group x task as fixed effects and 2) adjusted model with age and medication entered as fixed effects in addition to prior covariates. Dependent variables included measures of affect (positive and negative) and psychophysiology (heart rate, RMSSD, PEP, and SCL) averaged within each condition. Significant interactions were explored descriptively using graphical presentation of the data and post-hoc statistical tests of differences between conditions with simple contrasts. All reactivity model tables can be found in supplementary materials.

We corrected for multiple comparisons by hypothesis construct, such that we ran Benjamini–Hochberg Correction (Benjamini & Hochberg, 1995; Benjamini & Yekutieli, 2001) for pre-ejection period and skin conductance level analyses (sympathetic nervous system hypotheses) and PA and NA.

Results

Affect

Mother Affect.

There were no overall group differences for mother NA reactivity across interactions (b = −0.68, p = 0.08, 95% CI [−1.44 – 0.08]). There was a significant task effect, such that there was greater NA reactivity during the PSI compared to the EPI (b = 0.76, p = 0.04, 95% CI [0.05 – 1.47]), but this was not significant after correcting for multiple comparisons (p = 0.08). This effect was moderated by group such that Depressed Mothers demonstrated greater reactivity of NA during the PSI task (b = −1.50, p = 0.004, 95% CI [−2.50 – −0.50]), as shown in Fig. 1a, and this was significant after correcting for multiple comparisons (p = 0.008). In contrast, there were no overall group differences in PA reactivity across interactions (b = −0.72, p = 0.44, 95% CI [−2.53 – 1.09]), no task effect (b = −0.27, p = 0.75, 95% CI [−1.98 – 1.43]), and no group by task effect for PA reactivity (b = 0.55, p = 0.65, 95% CI [−0.09 – 0.12]).

Fig. 1.

PSI Negative Affect Reactivity. A Maternal Negative Affect Reactivity. B Adolescent Negative Affect Reactivity. Note: Plots were made using code from Allen et al. (2019) and show raw data, density plots, boxplots, and standard error

Adolescent Affect.

There were no overall group differences for adolescent NA reactivity across interactions (b = 0.25, p = 0.20, 95% CI [−0.13 – 0.63]), but there was a task effect such that there was greater NA reactivity during the PSI as compared to the EPI (b = 0.58, p = 0.002, 95% CI [0.22 – 0.95]) and this was significant after correcting for multiple comparisons (p = 0.002). Again, as shown in Fig. 1b, this effect was moderated by group such that High-Risk Adolescents demonstrated greater reactivity in reported NA during the PSI task (b = −1.19, p < 0.001, 95% CI [−1.71 – −0.68]). In contrast, there were no overall group differences in PA reactivity across interactions (b = −0.43, p = 0.48, 95% CI [−1.60 – 0.75]), no task effect (b = −0.36, p = 0.54, 95% CI [−1.49 – 0.78]), and no group by task effects for PA reactivity (b = 0.64, p = 0.44, 95% CI [−0.97 – 2.26]). After controlling for multiple comparisons, the task effect for NA reactivity persisted and the NA reactivity by task persisted (p < 0.0001).

Psychophysiology

Mother Psychophysiology.

There were no overall group differences for mother heart rate reactivity across interactions (b = 1.19, p = 0.09, 95% CI [−0.20 – 2.57]), no task effects (b = 1.19, p = 0.09, 95% CI [−0.20 – 2.57]) and no significant group by task effect (b = −0.15, p = 0.78, 95% CI [−1.16 – 0.87]). There were no overall group differences for mother RMSSD reactivity across interactions (b = 0.55, SE = 1.83, p = 0.76, 95% CI [−3.03 – 4.13]), no effect of task (b = −1.13, p = 0.24, 95% CI [−3.02 – 0.76]), and no significant group by task effect (b = −0.76, p = 0.58, 95% CI [−3.41 – 1.90]). There were no overall group differences for mother PEP reactivity across interactions (b = 2.04, p = 0.47, 95% CI [−3.46 – 7.53]), no between task effects (b = 0.10, p = 0.94, 95% CI [−2.38 – 2.58]), and no significant group by task effect (b = −2.31, p = 0.18, 95% CI [−5.71 – 1.08]). There was an overall group difference for mother SCL reactivity across both interactions (see Fig. 2a), such that Depressed mothers exhibited less SCL reactivity across tasks (b = 0.50, p = 0.02, 95% CI [0.10 – 0.89]), which was still significant after correcting for multiple comparisons (p = 0.03). In addition, there were no significant task effect (b = 0.15, p = 0.08, 95% CI [−0.02 – 0.32]) or significant group by task effects (b = −0.01, p = 0.92, 95% CI [−0.25 – 0.22]),

Fig. 2.

SCL Reactivity. A Maternal SCL Reactivity. B Adolescent SCL Reactivity. Note: Plots were made using code from Allen et al. (2019) and show raw data, density plots, boxplots, and standard error. SCL = Skin Conductance Level

Adolescent Psychophysiology.

There were no overall group differences for adolescent heart rate reactivity across interactions (b = 0.25, p = 0.76, 95% CI [−1.38 – 1.89]), but there was a task effect, such that adolescents exhibited greater heart rate reactivity during the negative task (b = 1.49, p < 0.001, 95% CI [0.76 – 0.2.23]). In addition, there was no significant group by task effect (b = −0.63, p = 0.23, 95% CI [−1.66 – 0.40]). There were no overall group differences for adolescent RMSSD reactivity across interactions (b = −1.16, p = 0.70, 95% CI [−7.01 – 4.68]), no task effect (b = −2.18, p = 0.18, 95% CI [−5.32 – 0.96]), and no significant group by task effect (b = 2.81, p = 0.21, 95% CI [−1.58 – 7.20]). There were no overall group differences for adolescent PEP reactivity across interactions (b = 1.89, p = 0.30, 95% CI [−1.64 – 5.43]), no task effect (b = −0.83, p = 0.12, 95% CI [−1.88 – 0.22]) and no significant group by task effect (b = −0.00, p = 1.00, 95% CI [−1.48 – 1.47]). There was an overall group difference for adolescent SCL reactivity across interactions (see Fig. 2b), such that High-Risk adolescents exhibited less SCL reactivity across tasks (b = 1.02, p = 0.001, 95% CI [0.42 – 1.61]), which was still significant after correcting for multiple comparisons (p = 0.002). In addition, there was no task effect (b = 0.212, p = 0.12, 95% CI [−0.05 – 0.49]) or significant group by task effects (b = −0.23, p = 0.24, 95% CI [−0.61 – 0.15]),

Post-Hoc Analyses

One potential concern in interpreting the above results, is that between group effects in reactivity could be attributable to between group differences in dyadic interactional behavior. This is a particular concern because it is well established that families of depressed women demonstrate more adverse family interactions (Beck, 1995; Leinonen et al., 2003; Norcross et al., 2017); in fact, in this study we found that High-Risk adolescents had significantly greater dysphoric behavior during the PSI as compared to the EPI (p = 0.03) tasks, while depressed mothers had significantly greater aversive behavior during the PSI as compared to the EPI (p < 0.001) tasks (see Supplemental Materials for behavioral observation methods). Therefore, it is difficult to discern whether greater reactivity reflects within person characteristics or the fact that persons in families of depressed women are responding to more emotionally evocative interactions. Hence, we conduced post-hoc between group analyses, controlling for interactants’ behavior (adolescent behavior in maternal models and maternal behavior in adolescent models). These models can be found in supplemental materials. Findings remained consistent after controlling for interaction behaviors, suggesting that observed between group differences do reflect differential reactivity as a function of maternal depressive status and adolescent risk status, rather than interactional behaviors.

Discussion

The current study was designed to examine the association of maternal depression with both mother and adolescent offspring’s self-reported affect and autonomic reactivity during positive and negative interaction tasks. The results provide partial support for the hypotheses. Both Depressed mothers and their High-Risk adolescents displayed significantly greater reactivity in self-reported NA during the negative interaction task. This finding is consistent with 1) research showing that depression is associated with greater NA reactivity to negative life circumstances (Bylsma et al., 2011; Myin-Germeys et al., 2003; van der Stouwe et al., 2019; van Winkel et al., 2015), 2) research showing that depressed persons exhibit greater reactivity to emotional stimuli that are social in nature and personally relevant (Allen & Badcock, 2003; Price et al., 1994), 3) research showing that during negative interactions mothers and their depressed adolescents exhibit greater levels of NA behavior (Sheeber et al., 2012), 4) female adolescents with depressed mothers exhibit greater NA reactivity to stress than female adolescents without depressed mothers (Waugh et al., 2012), and 5) the theory of stress sensitivity by which the tendency to react stronger to stress has been shown to predict increased depressive symptomatology and development of depression (Wichers et al., 2007, 2009). Importantly, these findings contrast with prior research showing that those with depression exhibit blunted affective reactivity in laboratory studies using less personally relevant stimuli (Bylsma et al., 2008). These findings also suggest that NA reactivity may be context dependent, such that depressed mothers and their offspring exhibit typical levels of NA reactivity during positive interactions, with differences from non-depressed mothers and Lower-Risk peers only arising in more challenging situations. The finding regarding adolescents is even more compelling when noting that the model controlled for adolescents’ own mental health symptoms and post-hoc analyses also controlled for maternal behaviors during the interaction – indicating that maternal depression may have effects on adolescent NA reactivity that are independent of adolescents’ own mental health or maternal behaviors.

Also in line with our hypotheses, Depressed Mothers showed blunted sympathetic (i.e., SCL) reactivity across tasks when compared to Non-Depressed Mothers, even after controlling for covariates (medication use, age, and adolescent behaviors in post hoc analyses) and multiple comparisons. This finding is consistent with research indicating that those with depression and those with increasing depressive symptoms have blunted autonomic reactivity (Carroll et al., 2007; Phillips, 2011; Salomon et al., 2009; Schiweck et al., 2019; York et al., 2007), particularly during interpersonal interactions that elicit a beta-adrenergic response (Salomon et al., 2009, 2013). This blunted or biological disengagement of sympathetic activity is in line with decreased motivation in those with depression (Salomon et al., 2013).

In contrast to our hypotheses, which were based on the theoretical assumption that maternal depression may acts as a form of social threat for adolescents that results in increased physiological reactivity, our findings indicated that High-Risk Adolescents exhibited a similar pattern of blunted sympathetic reactivity (as measured via SCL) across both positive and negative interaction tasks as displayed by their Depressed Mothers. This result is therefore more consistent with the alternative theoretical approach, which predicts that that adolescents of depressed mothers would exhibit a more blunted autonomic reactivity profile associated with depression (Salomon et al., 2013). These findings extend prior literature by 1) replicating findings of blunted autonomic reactivity in those with depression, 2) extending these findings to the adolescent offspring of depressed mothers, and 3) elucidating that this process may be somewhat context independent, in that it was consistent across tasks with varying emotional demands. Again, the finding among adolescents is even more compelling when noting that the model controlled for adolescent’s own mental health symptoms, medication use, age, gender, and in post-hoc models maternal behaviors during the interaction tasks – indicating that maternal depression was associated with sympathetic reactivity independent of those other factors. Furthermore, this blunted sympathetic process exhibited by adolescents of depressed mothers and depressed mothers themselves may indicate withdrawal and disengagement during the interaction tasks, which would be consistent with a pattern of learned helplessness during which there is a blunting of acute stress reactivity across time (Maier & Seligman, 2016).

Finally, there were no differences between the groups in either PA reactivity nor any measure of other putative autonomic reactivity, which may reflect that the interaction tasks didn’t elicit the type of PA reactivity that elicits autonomic arousal, as suggested by the fact that there was no direct task effect on PA reactivity in adolescents or mothers (p > 0.05). Perhaps an interaction task that involved more high arousal PA, such as those displayed more commonly among friends than between parents and adolescents, would have elicited a difference (Dahl et al., 2018).

Limitations and Future Directions

While the present study had significant strengths such as using a multimethod assessment of reactivity across affective self-report and autonomic psychophysiology within an ecologically valid social interaction, there were a number of limitations that should be noted. First, this was a cross-sectional study, which precluded the ability to study developmental timing effects within adolescence (i.e., early, middle, late adolescence). Future studies would benefit from longitudinal designs to extend these findings. Second, although the EPI and PSI tasks significantly discriminate between groups in affective behavior, tonic levels of psychophysiology (Nelson et al., 2021) and some indices of reactivity, the tasks capture a relatively narrow behavioral context. As we have argued previously, future studies might address this limitation by utilizing passive sensing technologies (i.e., wearable, smartphone, and smart home devices) to increase the range of environmental contexts examined (Allen et al., 2019; Nelson & Allen, 2018, 2019) to bring this type of research out of the lab and into real world contexts. Third, this study investigated overall levels of physiological activity across the interactions. Future studies should consider other analytic approaches that allow for more nuanced understanding of autonomic reactivity during the interaction tasks. Furthermore, although we have observed some descriptive similarities between the patterns of reactivity in the maternal and adolescent samples, we have not statistically tested this dyadic similarity. Approaches that explicitly examine the physiological synchrony of reactivity patterns using time-lagged mixed effects analyses could be used to for explicitly test the degree of similarity in reactivity patterns in a more temporally fine-grained way (Woody et al., 2016). Fourth, this study did not assess lifetime depression in order to identify age of depression onset for mothers or duration of exposure to maternal depression, which may be critical variables influencing adolescents’ underlying stress response systems. Future studies should collect more detailed lifetime depression in mothers to better understand the developmental timing effects of exposure to maternal depression on adolescents’ underlying stress response systems. Fifth, and related to the prior point, the diagnostic status of offspring was not assessed in the current study. It is likely that adolescents in the high-risk group may have a higher prevalence of psychopathology than those in the lower-risk group, similar to observed differences in YSR scores, which may have influenced findings, though the fact that controlling for youth YSR did not affect results, somewhat mitigates this concern.

Conclusion

Overall, our findings indicate that both depressed mothers and their adolescent offspring exhibit similarly exaggerated NA reactivity, specifically during negative interactions, while they also show blunted sympathetic activity across positive and negative interactions. These findings indicate that dysregulated NA during negative interpersonal interactions and blunted sympathetic activity across positive and negative interactions may be two mechanisms that are influenced by depression within and across generations. These findings have implications for mechanistic models of the intergenerational transmission of risk, suggesting that some aspects of affective and autonomic reactivity show similar profiles in depressed mothers and their high-risk adolescents. As noted above, longitudinal research is needed to clarify whether these profiles are prospectively associated with the onset of psychopathology in offspring. If such a relationship is confirmed, then it would strongly support the development of interventions designed to enhance the regulation of negative affect and increase social engagement during interpersonal interactions as a mechanism driven approaches to prevention and early intervention amongst adolescents with a maternal history of depression.

Availability of Data and Material

Data is available upon request.