Affective Neural Mechanisms of a Parenting-Focused Mindfulness Intervention

Abstract

Objectives:

Behavioral evidence suggests that parenting-focused mindfulness interventions can improve parenting practices and enhance family wellbeing, potentially operating through altered emotional processing in parents. However, the mechanisms through which parent mindfulness interventions achieve their positive benefits have not yet been empirically tested, knowledge which is key to refine and maximize intervention effects. Thus, as part of a randomized controlled trial, the present study examined the affective mechanisms of an 8-week parenting-focused mindfulness intervention, the Parenting Mindfully (PM) intervention, versus a minimal-intervention parent education control.

Methods:

Twenty highly stressed mothers of adolescents completed pre- and post-intervention behavioral and fMRI sessions, in which mothers completed a parent-adolescent conflict interaction, fMRI emotion task, and fMRI resting state scan. Mothers reported on their mindful parenting, and maternal emotional reactivity to the parent-adolescent conflict task was assessed via observed emotion expression, self-reported negative emotion, and salivary cortisol reactivity.

Results:

Results indicated that the PM intervention increased brain responsivity in left posterior insula in response to negative affective stimuli, and altered resting state functional connectivity in regions involved in self-reference, behavioral regulation, and social-emotional processing. Changes in mothers’ brain function and connectivity were associated with increased mindful parenting and decreased emotional reactivity to the parent-adolescent conflict task.

Conclusions:

Findings suggest that mindfulness-based changes in maternal emotional awareness at the neurobiological level are associated with decreased emotional reactivity in parenting interactions, illuminating potential neurobiological targets for future parent-focused intervention.

Parent emotional functioning can impact a wide range of family processes. High levels of parent negative emotion and poor emotion regulation may negatively affect the strength of the parent-child bond, the extent to which youth internalize adaptive emotional coping strategies, and a parent’s ability to carry out developmentally appropriate and positive parenting practices (for review, Bariola et al. 2011; Crandall et al. 2015). Parents may be particularly at risk for showing dysfunctional patterns of emotion—such as high emotional reactivity, emotional distancing, and poor emotion regulation—during their child’s adolescence, as adolescence represents a sensitive developmental period for the family system (Laursen et al. 1998; Steinberg and Silk 2002). Accordingly, there is a need for interventions to target parent emotional functioning, particularly for parents of adolescent children (Maliken and Katz 2013). A growing body of evidence suggests that mindfulness-based interventions may be effective in targeting the neurobiological processes supporting emotional functioning in adults (Hölzel et al. 2011; Tang et al. 2015), and thus parenting-focused mindfulness interventions may me ideally suited in targeting parent emotion in the family context. Despite the promise of mindfulness interventions for parents, no controlled study to date has evaluated the affective mechanisms of parenting-focused mindfulness training, information that is key for understanding intervention effects and optimizing future interventions.

Parenting-focused mindfulness interventions teach parents mindfulness practices with a focus on mindfulness in the family context, promoting nonjudgmental awareness, self-regulation, and compassion in parenting and in the parent-child relationship (Duncan et al. 2009). To date, empirical studies show evidence for the positive benefits of parent mindfulness intervention—including increased parent mindfulness, reduced parent stress, enhanced parent-child bonding, and reduced mental health symptoms in youth—across diverse populations in both non-controlled (Bögels et al. 2014; Singh et al. 2006, 2007, 2010; Van der Oord et al. 2012) and randomized controlled trials (Benn et al. 2012; Coatsworth et al. 2010, 2015; Felver et al. 2017; Neece 2014). An accumulating body of literature investigating the mechanisms of mindfulness-based interventions in general adult populations suggests mindfulness training achieves it positive benefits in part through its impact on affective neurobiological systems (Chambers et al. 2009)—including processes involved in emotional arousal, regulation, awareness, and social-emotion—with many of these changes overlapping with networks considered key in caregiving (e.g., for review, Kim et al. 2016; Swain et al. 2011). However, the neural mechanisms of parenting-focused mindfulness programs have not yet been investigated in controlled research. Given that mindfulness training may operate through a number of emotion-related pathways, greater knowledge of which affective mechanisms are critical to the effects of the intervention on parenting is needed to refine the effective implementation of these interventions.

First, research investigating the effects of mindfulness-based interventions and meditation has found that mindfulness is associated with lower levels of reactivity in brain regions involved in emotional arousal such as the amygdala (Desbordes et al., 2012; Goldin and Gross 2010; Lutz et al. 2014; Taylor et al. 2011). Moreover, in a separate literature, research on parents of infants has also shown that aberrant amygdala reactivity in response to distressed infant cues is associated with poorer parenting and poorer relationship quality (Atzil et al. 2011; Kim et al. 2017; Laurent and Ablow 2012). Thus, especially with respect to emotional processing of negative stimuli, altered amygdala function may represent a mechanism of mindfulness training that is particularly relevant to parenting. With its key roles in emotional processing, threat-detection, and stress, reduced amygdala responsivity may be associated with parents’ better ability to tolerate challenging events in parent-adolescent interactions and respond to their adolescents’ behaviors without over-reactivity or avoidance.

In addition, studies on mindfulness interventions have also found that mindfulness training may increase involvement of regulatory cortical regions in response to negative emotional stimuli (Allen et al. 2012; Haase et al. 2015; Johnson et al. 2014; Lutz et al. 2014), as well as strengthen cortical-limbic functional connectivity (Doll et al. 2015; Hölzel et al. 2013; Taren et al. 2015). More specifically, it has been proposed that changes in dorsal anterior cingulate cortex (dACC) following mindfulness training may be related to non-judgmental awareness and acceptance, processes that are specifically encouraged in mindfulness interventions and that may allow for greater cognitive control of or regulation of emotion (for review, Tang et al. 2015; Young et al. 2018). Interestingly, neuroimaging studies of maternal brain response to distressed infant stimuli have also shown that engagement of cognitive control systems may contribute to individual differences in parenting, as greater recruitment of ACC and prefrontal cortical regions is related to higher parenting sensitivity and positive parenting (Michalska et al. 2014; Musser et al. 2012). Thus, changes in brain regions involved in emotion regulation may represent another important mechanism of parenting-focused mindfulness interventions, as strengthened regulatory systems in the brain may allow parents the capacity to accept and regulate the experience and expression of strong negative emotions, thereby reducing over-reactive or automatic behaviors in interactions with youth.

Based on theory and empirical evidence, it has also been posited that mindfulness training may reduce self-referential processing and instead increase present-moment, objective awareness of emotion. That is, mindfulness training is thought to cultivate awareness of physical sensations of emotions without attaching one’s self to an emotion’s meaning, evaluative context, or associated cognitions (Chambers et al. 2009). In support of this notion, several studies show that mindfulness training increases posterior insula activation, a region of insular cortex thought to support interoceptive awareness (Allen et al. 2012; Farb et al. 2007, 2010, 2012) and increases functional connectivity among posterior insula and reward/attention regions (Farb et al. 2007; Kirk et al. 2014, 2016). Moreover, research on parenting also implicates insular cortex in caregiver neurobiology (for review, Swain et al., 2011), with for example functional neuroimaging studies showing medial-posterior insula activity in parents in response to distressed child stimuli (Lenzi et al., 2009, Swain et al. 2017). For parents, increasing sensory awareness of negative emotions and decreasing self-referential focus may lead to less avoidance (e.g., experiential avoidance), cognitive over-engagement (e.g., worry, rumination), and reflexive responding (e.g., punitive parenting) based on previous self-narratives and experiences. Therefore, changes in emotional awareness may reflect an additional process relevant for intervention-related changes in parent mindfulness.

Finally, emerging evidences suggests that mindfulness practice may also enhance compassion and empathy-related processes (e.g., Dekeyser et al. 2008; Laneri et al. 2017; May et al. 2016), with several studies showing altered mindfulness-related responsivity in anterior insula and dACC (Laneri et al. 2017; Lutz et al. 2008; May et al. 2016), key nodes of the salience network (Menon 2015) which have been implicated in empathic responding to other’s pain (Fan et al. 2011). In the only investigation to our knowledge investigating the neural mechanism of a mindfulness intervention for parents, May et al. (2016), found that mothers showed increased activation in anterior insula during a mindful awareness of breathing task in a non-controlled study of eighteen parents of adolescents. The authors interpreted these findings as evidence of enhanced empathy and emotion regulation. Of note, given that this study was a non-controlled investigation of mindfulness training for parents, controlled research is needed to replicate these findings. Still, consistent with this work, studies examining maternal brain response to infant stimuli also support the importance of brain networks involved in social-emotion, with neuroimaging studies linking increased activation and connectivity in social processing networks with maternal sensitivity (Elmadih et al. 2016; Wan et al. 2014) and parent-infant synchrony (Abraham et al. 2016). Thus, the processing of social-emotion at the neurobiological level may reflect an additional mechanism of parenting-focused mindfulness interventions.

The current randomized controlled pilot study uniquely evaluated changes in parents’ affective brain function and resting state functional connectivity following a Parenting Mindfully (PM) intervention and Parent Education (PE) minimal control intervention in a sample of highly stressed mothers of adolescents. First, it was hypothesized that the PM intervention would alter parent brain response to negative emotional stimuli and resting state functional connectivity from pre to post-intervention, reflecting decreased emotional arousal, and increased emotional awareness, regulation, and social-emotional processing. Second, we explored associations between intervention-related changes in brain function and changes in parenting-related behaviors from pre to post-intervention, including changes in mindful parenting practices and parent emotion-related responsivity to a parent-adolescent conflict interaction. Given the sample size of the present study, correlations between changes in brain function/connectivity and parent behavior were considered exploratory in order to inform larger future studies. It was hypothesized that the aforementioned changes in brain function and connectivity would be associated with increased mindful parenting and decreased negative emotional reactivity to the parent-adolescent conflict interaction. Of note, the present study’s small-scale pilot investigation represents an important first step in treatment development and evaluation. Initial stages of treatment development, which primarily rely on small sample sizes, are needed to specify theoretical rationale and change mechanisms of interventions and establish feasibility for larger-scale investigations (Rounsaville et al. 2001).

Method

Participants

Participants included 20 female primary caregivers (referred to throughout as mothers), drawn from a larger study of a parenting-focused mindfulness intervention. As part of the full intervention study, 96 primary caregivers and their adolescents (ages 12–17 years) were recruited through mailings, community advertisements, and community mental health centers in the mid-Atlantic U.S and randomized to an intervention condition. For the MRI subsample, the mean age of participants was 48.5 years (SD = 7.62) and caregivers were biological mothers with the exception of one adoptive mother. Adolescent race was 55% Non-Hispanic White, 10.0% Mixed-Race, 5% Hispanic, 20% African-American, 10% Other/Not Reported. Most participants had family household annual incomes above $100,000 (60%; 5% between $75,000–100,000; 5% between $60,000–74,999; 10% between $45,000–59,999; 15% < $45,000, n = 19). The sub-sample of 20 mothers did not differ significantly from the larger sample on age, race/ethnicity, and income.

Inclusion criteria for the larger study included primary caregivers with a child between 12–16 years, adequate English proficiency, and elevated parent self-reported stress (mean score of at least 3 [on a 1–5 scale] for two questions adapted from stress and parenting stress measures: “In the last month, how often have you felt stressed?” and “In the last month, how often have you felt stressed by parenting your teenager or worried about your teenager?”). Exclusion criteria for the larger study included a history of psychotic disorders or intellectual disability for adolescents and a medical/physical condition that would preclude participation in gentle yoga for parents. Of those 96 primary caregivers that were interested in and eligible for MRI scans, 10 from each intervention condition were randomly selected to complete an additional fMRI pilot study session at pre- and post-intervention. MRI inclusion criteria were MRI safety eligibility (e.g., no metal in body), not currently pregnant, right-handed, and not taking psychiatric medications.

Procedure

As part of the larger study, participants were randomly assigned to either the PM intervention (8 sessions over 8 weeks) or the PE minimal control intervention (3 sessions over 8 weeks) using a computer-generated random numbers sequence. Mothers and adolescents in the present study sample (n = 20) completed pre-intervention behavioral and fMRI assessment sessions (spaced approximately 1–4 weeks apart from one another), then the PM or PE intervention (8 weeks in duration), and finally the post-intervention behavioral and fMRI assessment sessions (spaced approximately 1–4 weeks from one another).

fMRI Session.

Mothers completed fMRI scan sessions using a Siemens 3T Allegra MRI scanner. Functional images of the BOLD response were collected using T2*-weighted gradient EPI) [TR/TE: 2350/30ms; flip = 70°; FOV: 192mm; matrix size: 64 × 64; 40 axial 3mm thick/1mm gap slices]. For structural imaging, a T1-weighted MPRAGE anatomical image was acquired (TR/TE = 2300/3ms; FOV = 260mm; matrix size = 256 × 256; 160 1mm thick slices).

Mothers completed two functional scans including the emotion image task and 6-minute resting state scan. The emotion image task was a rapid, event-related design that included 27 general negative, 27 neutral, and 27 negative adolescent images. General negative and neutral images were drawn from the IAPS (Lang et al. 2008). Each trial included the image (4.70 seconds) and a fixation cross (jittered between 2 and 12 seconds). Trials were presented in a pseudo-randomized order across three 5-minute runs, and participants were instructed to push a single button with the index finger of their right hand when each image appeared to ensure sustained attention to the stimuli. In order to isolate mothers’ general negative emotional processing, the present study focused on the general negative (referred to as “negative” throughout) > neutral image contrast. For the resting state scan, mothers were instructed to keep their eyes open and focus on a fixation cross.

Behavioral Assessment Session.

For the behavioral assessment sessions, mothers and adolescents completed self-report assessments, cortisol assays, and the parent-adolescent interaction task (PAIT). Adolescents and mothers refrained from eating during the session and were asked to refrain from alcohol or drug use in the day before the session. The PAIT was a 10-minute conflict interaction based on conflict tasks performed in prior research (e.g., Sheeber et al. 1997). At each behavioral session, mother and adolescent completed the IC (Prinz et al. 1979), a checklist of common family conflict topics. Then, mothers and adolescents separately completed a 10-minute adaption period, in which participants listened to a 5-minute guided breathing and muscle relaxation recording and were told to relax. Immediately following the adaptation period, baseline salivary cortisol measurements and self-reported ratings of emotion were collected. Next, the mother and adolescent were seated in the same room and completed the conflict interaction, during which the mother and adolescent were asked to discuss their mutually highest-rated conflict topic from the IC. They were asked to “use the next 10 minutes to discuss the issue and to try to reach a solution that you think will work for you.” After the discussion task, mothers returned to their rooms and measures of mothers’ salivary cortisol and self-reported ratings of emotion were taken immediately post-task, 15 minutes following, and 30 minutes following the discussion task. Then, adolescents and mothers completed self-report questionnaires and interviews.

Intervention

The PM intervention consisted of 8 weekly group sessions (2 hours per session) focused on cultivating mindfulness in the context of parenting. The PM Intervention was based on existing mindfulness interventions—Kabat-Zinn’s (1990); Segal, Williams, and Teasdale’s (2012)—but focused on parenting interactions. The intervention incorporated in-session meditation practices, parenting-focused mindfulness exercises (e.g., child-directed activities, mindful listening), assigned daily meditation, and guided discussion of mindfulness practices in parenting (e.g., compassion, reducing automatic pilot in parenting interactions). For further details on PM intervention content, please see Chaplin et al. (2018).

The PE intervention was an active minimal-control group that met three times (weeks 1, 4, and 8) for 30 minutes each. Information on adolescent social and emotional development, risk behaviors, and parenting advice (e.g., supervision, parent-child communication) was presented. Parents were also given 3-page pamphlets based on information pamphlets created by NIDA for parents (e.g. NIDA 2012) and given the opportunity to ask questions and discuss content. Although the PE control did not match the PM intervention with regard to the frequency and duration of sessions, this control was intended to model other existing brief prevention programs for parents and control for some nonspecific factors, including attendance in group sessions with other parents and expectations for improvements.

Measures

Mothers’ emotional reactivity to the PAIT was examined across three domains of emotion functioning, including observed emotion expression, self-reported subjective emotion experience, and physiological arousal.

PAIT emotion expression.

Mothers’ negative and positive emotion expression and parent-adolescent shared positive emotion expression during the PAIT interaction were coded using the PAIT Coding System (Chaplin, 2010, unpublished manual). Emotion expression coding assessed facial, vocal, gestural, and postural cues of negative and positive emotion based on emotion coding systems in the literature (Cole, Barrett and Zahn‐Waxler 1992; Ekman and Friesen 1978; Izard 1979). Emotion expression codes were rated on a scale from 1 to 5 (“none” to “high”). Negative emotion coding was based on cues for sadness, anger, fear, contempt and aggression (e.g., furrowed brows, crying). Positive emotion coding was based on cues for happiness (e.g., smiling with crinkling around eyes, laughing). Shared positive emotion coding was based on the proportion of positive emotion events in which both the parent and adolescent expressed positive emotion at the same time while looking at one another. This proportion was represented by the total number of simultaneous parent-adolescent positive emotion expressions divided by the total number of positive emotion expressions of the partner (parent or adolescent with the fewest positive emotion expressions during the conflict task). Coders were trained on the PAIT coding system for 6 hours and attended bi-monthly coding meetings to discuss coding questions. Nineteen percent of videos were chosen at random, double-coded and checked for inter-rater reliability. The intraclass correlation coefficients (ICC’s) were acceptable for negative, positive, and shared positive emotion expression (ICC = .83, ICC = .80, ICC = .89, respectively).

PAIT self-reported emotion.

Mothers’ self-reported emotions in response to the PAIT were measured using the Differential Emotions Scale-Revised short form (DES-R; Izard 1972). The DES-R demonstrates good psychometric properties (Izard 1972) and has been applied in previous parent-adolescent interaction tasks (Chaplin et al. 2012). For this report, we examined two emotion subscales of the DES-R that we expected the conflict interaction to elicit—sadness and anger. Emotion subscales are comprised of five adjectives describing each emotional state. The present study examined self-reported emotions immediately following PAIT as they were thought to best reflect mothers’ subjective experience of emotion during the PAIT.

PAIT cortisol reactivity.

Mothers’ HPA axis activation in response to the PAIT was examined as a measure of emotion-related physiological arousal. Saliva was collected using a cotton swab, which was placed between each participant’s tongue and cheek for approximately 2 minutes. Saliva samples were stored at −20 degree Celsius. Saliva samples were assayed in duplicate using standard radioimmunoassay kits with no modifications (intra-assay coefficients of variation from 3.0 to 5.1%). An overall cortisol reactivity score was calculated by subtracting baseline cortisol from each individual’s peak reactivity score (highest salivary cortisol value at any time point after the discussion task), which has been used as a valid measure of reactivity to an acute stressor (Buss et al. 2005; Chaplin et al. 2014; Rudolph et al. 2010).

Mindful parenting.

Mothers’ report of her own mindful parenting was assessed through the Interpersonal Mindfulness in Parenting scale (IM-P, Duncan 2007). The IM-P scale is an 8-item questionnaire designed to assess awareness and present-centered attention, nonjudgment, and nonreactivity in parenting. Items are rated from 1 to 5 (“never true” to “always true”), such as “I find myself listening to my child with one ear because I am busy doing or thinking about something else at the same time” (reverse-scored). The present study utilized the IM-P total score, reflecting overall levels of mindful parenting behavior. The IM-P has demonstrated concurrent and discriminant validity (Duncan 2007). In the present study’s sample, alpha was .82 for the IM-P total score.

Data Analyses

Task-based analysis.

Task-based fMRI data analysis was conducted using FMRI Expert Analysis Tool as part of FSL (Jenkinson et al. 2012). Standard preprocessing steps included rigid-body motion correction, slice-time correction, and spatial smoothing (6mm FWHM Gaussian kernel). Functional images were co-registered to each participant’s MPRAGE structural image using FLIRT and normalized to the MNI template using FLIRT non-linear transformations. FSL’s motion outlier function was utilized to identify and regress out motion-affected scans in first-level analyses. Runs with extreme motion outliers (motion greater than 3mm in any direction for one TR or motion >1.5 mm for > 80% of the run) were excluded from further analyses.

First-level linear regression at each voxel was conducted, using generalized least squares with a voxel-wise, temporally and spatially regularized autocorrelation model, drift fit with Gaussian-weighted running line smoother (96 sec FWHM).

The mean beta coefficients from each participant’s first-level regression analyses were extracted from a priori ROIs, selected based on empirical research and theory in both mindfulness and parenting literatures: amygdala (e.g., Atzil et al. 2011; Desbordes et al. 2012), dACC (e.g., Haase et al. 2015; Michalska et al. 2014), posterior insula (e.g., Farb et al. 2007, 2010, 2012; Swain et al. 2017), anterior insula (e.g., May et al. 2016). ROIs were created using FSL’s Harvard Oxford Atlas (http://www.fmrib.ox.ac.uk/fsl/). For non-midline regions, left and right ROIs were extracted separately. Consistent with studies demonstrating an anterior-posterior insula gradient (e.g., Cauda et al. 2011), the anterior insula mask included the atlas’ insular cortex structure at y > 6, with remaining insular cortex defined as posterior. The dACC ROI included the atlas’ anterior division of the cingulate gyrus at z > 25, consistent with reported functional divisions of the anterior cingulate (e.g., Etkin et al. 2011).

Intent-to-treat analyses were utilized, such that all participants were analyzed in the groups in which they were randomized regardless of session attendance. This approach is considered conservative, as it is less likely to provide biased treatment effects compared to approaches in which only fully compliant participants are analyzed (Gupta 2011). A series of analysis of covariance models (ANCOVAs) were conducted, with post-intervention ROI estimates (negative > neutral scenes) entered as the dependent variable, intervention group as the independent variable (PM = 1, PE = 0), and pre-intervention ROI estimates entered as covariates. Alpha was set at .05 for these analyses. Given that ROIs were carefully selected based on theory and prior empirical work on mindfulness, we did not correct for multiple comparisons for this analysis. This approach is also consistent with recent pilot studies on mindfulness-based treatments with similar sample sizes (e.g., Braden et al. 2016; Li et al. 2016; Opialla et al. 2015).

Resting state functional connectivity.

Resting state preprocessing steps were completed in CONN (Whitfield-Gabrieli and Nieto-Castanon 2012). Standard preprocessing included realignment, slice-timing correction, coregistration, spatial normalization to the MNI template, spatial smoothing using an 8 mm FWHM Gaussian kernel, and temporal band-pass filtering of 0.008-.09 Hz. Scans with extreme motion outliers (5 standard deviations about the mean global intensity or motion exceeding .9 mm) were identified, and sessions with greater than 15% of affected volumes were excluded from further analyses (n = 2). Noise components from white matter and cerebrospinal fluid were extracted by principal component decomposition of BOLD signal in these regions following the CompCor approach (Behzadi et al. 2007). Six rigid-body motion parameters, motion outliers, CSF, and white matter noise components were included in the model as nuisance covariates. Resting state fMRI time course from a priori seed ROIs in the amygdala, posterior insula, anterior insula, and dACC and were used in a general linear model (GLM) regression to examine functional connectivity between seed ROIs and every other brain voxel. Seed-to-voxel correlation coefficients were subjected to a Fisher r-to-z transformation. To identify group differences in intervention-related connectivity changes, group-level random effects t-tests were conducted assessing intervention group (PM vs PE) effects on post-intervention > pre-intervention z-scores. Results were thresholded at z > 2.6 with a whole-brain cluster level false discovery rate (FDR) correction of p < .05.

Brain-behavior correlations.

Associations between changes in brain activity from pre- to post-intervention and changes in parenting behaviors from pre- to post-intervention were also explored. In separate linear regressions, parenting behaviors (i.e., mindful parenting, PAIT emotion reactivity) at post-intervention were regressed on ROI change scores (post – pre-intervention values), controlling for pre-intervention parenting behaviors.

Results

Preliminary Data Analysis

Across intervention groups, no significant differences emerged on adolescent gender, adolescent and mother age, adolescent race, current therapy status, or any primary study variable at pre-intervention (p’s > .10). Eighty percent of mothers attended at least one intervention session (80% PM, 80% PE). Additionally, consistent with previous studies of group-based interventions (Gillham et al. 2007), the average percentage of sessions attended was 41.25% for PM and 50% for PE in the MRI sample (of note, there were fewer PE sessions than PM).

Of the 20 participants in the present study’s MRI sample, one participant at post-intervention was excluded due to excessive motion and one participant did not complete the post-intervention MRI scan and thus was excluded in further analyses. For exploratory analyses of brain-behavior correlations, two participants were excluded, as they did not complete the post-intervention behavioral assessment session. Two additional participants were missing cortisol data due to an inadequate amount of saliva to assay at post-intervention. Thus due to this combined data loss for this variable (n = 15 for emotion task, n = 14 for resting state), for analyses involving cortisol reactivity, a last observation forward approach was utilized as an assumption of non-response to the intervention to maintain the present study’s sample size. Mothers with missing data however did not differ significantly (p’s > .05) on parent age, adolescent race/ethnicity, adolescent gender, therapy enrollment status, income levels, and pre-intervention levels of PAIT emotion reactivity variables and mindful parenting.

Task-Based Analyses

It was hypothesized that PM mothers would show increased posterior insula, anterior insula, and dACC BOLD response and decreased amygdala BOLD response to the negative > neutral image contrast compared to PE controls. Consistent with hypotheses, results indicated that mothers in the PM intervention showed significantly greater BOLD response in the negative > neutral image contrast compared to PE in left posterior insula (F[1,15] = 5.76, p < .05, cohen’s d = 1.28) and right posterior insula at a level approaching statistical significance (F[1,15] = 4.42, p = .053, cohen’s d = 0.94) at post-intervention, covarying for pre-intervention BOLD response, with large effect sizes. Figure 1 shows the means values for each group at pre and post-intervention for left posterior insula, and suggests that PE mothers showed reduced activity from pre- to post-intervention; whereas, PM mothers showed an overal maintainence or slight increase in activity in posterior insula from pre- to post-interevention. Contrary to expectations however, no significant intervention effects emerged for amygdala (L: F[1,15] = 0.14, p = .71, R: F[1,15] = 0.86, p = .37), anterior insula (L: F[1,15] = 0.10, p = .76, R: F[1,15] = 0.39, p = .94), or dACC (F[1,15] = 0.10, p = .76) activation.

Fig. 1.

Posterior insula BOLD response to negative > neutral images at pre- and post-intervention. PE = Parent Education, PM = Parent Mindfulness, L = Left, R = Right, ns = nonsignificant. *p < .05, ^†p <.10

Resting State Functional Connectivity

First, it was posited that PM mothers would show stronger connectivity between posterior insula and salience network nodes (i.e., anterior insula, dACC), reflecting greater monitoring and attention to interoceptive cues. Seed-to-voxel resting state connectivity indicated that mothers in the PM intervention showed reduced connectivity between right posterior insula and a cluster encompassing superior parietal lobule (SPL) and precuneus (peak = −16, −48, 66; 810 voxels, size FDR-corrected p < .001). The group difference in mean right posterior insula-SPL/precuneus connectivity at post-intervention reflected a large effect size (PM mean β = 0.01; PE mean β = 0.34; cohen’s d = 1.98). Results of seed-to-voxel analyses are shown in Figure 2.

Fig. 2.

Resting state functional connectivity with a) dACC seed; b) R posterior insula seed; c) R anterior insula seed. R = right, L = left

Second, it was anticipated that PM mothers would show greater intra-connectivity between salience network nodes (anterior insula and dACC) as well as greater inter-connectivity with social processing regions (e.g., STS). Results suggested that mothers in the PM intervention showed reduced connectivity between right anterior insula and left posterior middle temporal cortex (peak = −60, −60, 4; 323 voxels, size FDR-corrected p < .05), in a cluster encompassing middle temporal gyrus and STS as compared to PE mothers, reflecting a large effect size (PM mean β = 0.07; PE mean β = 0.42; cohen’s d = 1.70). No results emerged for left anterior insula.

For the dACC seed, results showed that mothers in the PM intervention exhibited reduced connectivity between dACC and a cluster encompassing precentral gyrus (preCG) and middle frontral gyrus (MFG) (peak = −40, −4, 52; 356 voxels, size FDR-corrected p < .05) as compared to PE mothers with a large effect size (PM mean β = 0.40; PE mean β = 0.03; cohen’s d = 2.53).

Lastly, it was anticipated that PM parents would show greater increases in amygdala – frontal cortical connectivity, reflecting greater monitoring and regulation of negative emotion. Contrary to expectations, however, no results emerged for either right or left amygdala seed.

Brain-Behavior Associations

Associations between intervention-related neural changes in brain function and functional connectivity and maternal behavior were explored. Specifically with respect to task-based function, associations between changes in left posterior insula activation (post minus pre-intervention BOLD response) and self-reported mindful parenting and maternal PAIT emotional reactivity were examined. Results showed that increased left posterior insula response to the negative > neutral image contrast was significantly associated with lower levels of cortisol reactivity to the PAIT at post-intervention, controlling for pre-intervention maternal cortisol response (β = −0.44, p < .05). Additionally, increased left posterior insula BOLD response predicted higher levels of observed shared positive emotion between parent and adolescent during the PAIT (β = 0.52, p < .05), controlling for pre-intervention shared positive emotion. No significant associations emerged between left posterior insula BOLD response and maternal self-reported negative emotion in response to the PAIT, observed parent negative and parent-only positive emotion, and mindful parenting.

Next for resting state functional connectivity, each participants’ connectivity values at pre and post-intervention were extracted from significant clusters from seed-based resting state analyses, including posterior insula-SPL/precuneus, anterior insula-posterior temporal cortex, and dACC-preCG/MFG connectivity, and associations with mindful parenting and maternal PAIT emotion reactivity at post-intervention were examined. Results indicated that decreased right posterior insula-to-SPL/precuneus connectivity from pre to post-intervention was associated with greater mindful parenting at post-intervention (β = −6.2, p < .05), controlling for pre-intervention mindful parenting levels. No results emerged between anterior insula-poster temporal cortex and dACC-preCG/precuneus connectivity and mindful parenting.

Furthermore, increases in anterior insula-posterior temporal cortex connectivity from pre to post-intervention were associated with higher levels of self-reported sadness following the PAIT (β = 0.43, p < .05), controlling for pre-intervention self-report ratings. No significant results emerged for either dACC-to-preCG/MFG and right posterior insula-SPL/precuneus with self-reported sadness. Further, intervention-related changes in resting state functional connectivity were not significantly associated with mother’s self-reported ratings of anger following the PAIT. Moreover, no results emerged between resting state functional connectivity and self-reported observed maternal emotion and cortisol reactivity in response to the PAIT.

Discussion

Empirical research to date suggests that parenting-focused mindfulness interventions hold promise for improving parenting practices, strengthening the parent-child bond, and improving youth outcomes. However, no study to date has examined the neurobiological mechanisms of a parenting-focused mindfulness intervention in a randomized controlled study. Thus, the current investigation examined changes in affective maternal brain function and connectivity following a mindfulness intervention for parents in a controlled trial. Findings indicated that the PM intervention increased mothers’ posterior insula response to negative emotional stimuli from pre to post-intervention compared to controls and altered posterior insula, anterior insula, and dACC resting state functional connectivity. PM intervention increases in emotion-related brain function and resting state functional connectivity were associated with increases in mindful parenting behavior and decreased negative emotional reactivity to a challenging parenting interaction. Together, findings from this investigation highlight that mindfulness interventions may affect maternal emotional awareness at a neurobiological level, which in turn may reduce emotional reactivity and allow for more mindful parenting behaviors in challenging parenting interactions. These findings provide important insight for future parenting interventions aimed to reduce parent emotional reactivity to improve parenting, by suggesting that maternal emotional awareness rather than general emotional reactivity is a key treatment target for parent behavioral change.

Emotional Awareness

Mothers’ task-based functional brain responses to emotional stimuli were first examined, with expectations that the PM intervention would result in reduced amygdala responsivity and increased posterior insula, anterior insula, and dACC responsivity. Hypotheses were confirmed in part, as mothers participating in the PM intervention exhibited increased left posterior insula response to the negative > neutral image contrast compared to PE controls. Given the posterior insula’s role in interoceptive processing, this finding suggests that parenting-focused mindfulness training may result in increased self and emotion-related awareness. That is, mothers trained in mindfulness may be more aware of bodily sensations and internal sensory cues of emotional events. Indeed, theoretical conceptualizations and reports from mindfulness practicitioners underscore the increase in attention to and nonjudgmental acceptance of internal body-based experiences, and previous neuroimaging studies of mindfulness have shown increased posterior insula activity during meditation (Sze 2010) and following mindfulness-based training (e.g., Farb et al. 2007, 2010).

Furthermore, analysis of brain-behavior associations showed that increases in left posterior insula response to negative images was associated with decreased maternal cortisol reactivity to the parent-adolescent conflict discussion. This finding is an important one as it suggests that affective brain changes from parent mindfulness training have implications for downstream emotion and stress-related physiological reactivity during parenting. Stress reactivity is considered a key risk factor for child maltreatment (Deater-Deckard 1998), and physiological stress responses, including high levels of cortisol reactivity, have been linked to harsh parenting (Martorell and Bugental 2006). Consistent with models of mindfulness-based intervention mechanisms, increased nonjudgmental monitoring of emotions, perhaps evidenced by increased posterior insula processing, may help individuals in their capacity to avoid automatic over-reactivity to stressors, thus reducing physiological arousal to parenting interactions.

Findings also revealed that increases in posterior insula response were also associated with increased positive emotion shared between parent and adolescent during the conflict discussion, suggesting that increased intervention-related posterior insula activity may have benefits not only for physiological components of emotional reactivity but also for changes in parent emotional behavior. It is possible that increases in nonjudgmental observation of the sensory aspects of emotion may allow parents the space to make more intentional choices about the way they relate to their adolescent, thus increasing opportunities to share in positive emotion exchanges and affection.

Intervention-related changes in maternal resting state functional connectivity indicated that mothers participating in the PM intervention showed decreased functional connectivity between posterior insula and a midline parietal cluster consisting of SPL and precuneus. This finding may be consistent in part with a previous study showing altered insula connectivity following mindfulness training, with participants exhibiting a decrease in connectivity between insula and medial PFC, a node of the default mode network implicated in self-referential processing (Farb et al., 2007). The authors proposed that this shift in insula-medial PFC connectivity represented a decrease in self-referential evaluation of internal cues and sensations. This finding in the current study may suggest a similar pattern, in that the precuneus is also considered to be involved in self-referential processing. Thus perhaps, reduced posterior insula-SPL/precuneus connectivity may represent a decrease in evaluative, narrative-based processing of interoceptive cues.

Moreover, reduced posterior-SPL/precuneus resting state connectivity was related to increases in levels of self-reported mindful parenting at post-intervention. In other words, this decreased connectivity predicted mother’s reports of their ability to bring nonjudgmental present-centered awareness to parent-adolescent interactions following the intervention. Perhaps, mothers who are able to avoid patterns of self-related processing of emotions, particularly when emotional processing is evaluative in nature rather than objective, may be able to bring greater moment-to-moment attention and acceptance to their parenting interactions with their adolescent. Altogether, findings suggest that posterior insula function and connectivity may represent key mechanisms of parent-focused mindfulness interventions, with implications for both mindful parenting and emotional reactivity in parenting.

Emotion Regulation

PM mothers showed increased connectivity between dACC and a dorsal lateral region of frontal cortex, including precentral and middle frontral gyrus, compared to PE mothers. More specifically, dACC connectivity increased in a region likely encompassing premotor cortex, a functional area involved in selecting motor responses and behavior. Thus, this increased integration between dACC, an emotion regulation region, and preCG/MFG, a region involved in action-selection, may represent a strengthened control mechanism of the parent-focused mindfulness, which may have implications for improved regulation of emotion in parents. Parents with more highly integrated regulatory and motor control may be better able to manage negative emotions in parenting interactions and consciously selected subsequent behavior in accordance with their parenting goals.

Social-Emotion

While PM mothers did not show changes in anterior insula activity in response to negative emotional stimuli as anticipated, PM mothers did demonstrate changes in anterior insula resting state functional connectivity. Specifically, the PM group relative to the PE control exhibited reduced functional connectivity between right anterior insula and left posterior temporal cortex (posterior MTG extending into posterior STS). This finding was unexpected as we anticipated regions implicated in empathy (i.e., anterior insula and dACC) to show greater connectivity with social-cognitive regions such as posterior temporal cortex/STS, potentially reflecting enhanced integration of empathy or social-emotional processing circuitry. Anterior insula is thought to have a role in integrating multisensory emotional inputs with cognition, with empirical research showings its involvement in the subjective experience of emotion and feelings of empathy (Fan et al. 2011; Menon, 2015). Posterior STS is a multisensory processing region involved in diverse social-cognitive functions (e.g., biological motion, faces, semantic memory) as well as audition and language (Deen et al. 2015). Thus, although contrary to the direction of hypotheses, this finding may reflect a reduced propensity for parents trained in mindfulness to engage in social-cognitive processes concerned with the internal states of others, perhaps because mindfulness meditation practices predominantly focus on observing internal feeling states. Alternatively, some research has suggested that empathy is comprised of affective and cognitive components, with affective dimensions referring to one’s visceral emotional reactions to others’ negative emotional states and cognitive dimensions referring to one’s ability to make cognitive inferences about others’ emotional states and mental perspectives. In their meta-analysis on the functional neural correlates of empathy, Fan et al. (2011) found that affective-perceptual aspects of empathy were more likely to involve right anterior insula. Thus perhaps, right anterior insula activity supports the negative emotional arousal component of empathy, similar to processes of emotional contagion. In this light, the present study’s findings may suggest that mindfulness enables individuals to distance themselves from automatic negative emotional reactions, even those involved in response to social-emotional displays by another.

Interestingly, reduced anterior insula-posterior temporal cortex functional connectivity was associated with lower levels of self-reported ratings of sadness in response to the parent-adolescent conflict discussion. This association may lend support for the notion that anterior insula-posterior temporal cortex connectivity represents a decrease in emotional reactivity to social-emotional cues, thereby resulting in lower levels of subjective experiences of negative emotion during parent-adolescent interactions. Undoubtedly, however, future studies are needed to replicate this finding to shed further light on social processing changes related to parent mindfulness training.

Emotional Arousal

Finally, the present study did not detect intervention-related changes in amygdala responsivity, both in mothers’ task-based brain function and resting state functional connectivity. It is possible that in the current study, we were underpowered to detect such subcortical effects in emotion-related responses; however, it may be that the current parent-mindfulness intervention more strongly impacted emotional functioning through altered emotional awareness and presumably one’s acceptance toward such internal states, rather than an individual’s emotional reactivity in and of itself. Future parent-focused mindfulness studies with larger sample sizes will be useful in further investigating changes in amygdala and subcortical affective networks.

Limitations and Future Directions

While the findings from the current study provide important initial insight into the neural mechanisms of a parent-mindfulness study, the findings should be interpreted in light of several limitations. First, the small sample size of the present study may limit our ability to capture smaller or more moderate effect sizes, thus affecting our ability to assess more subtle or nuanced intervention effects and associations with parent behavior. Second, the present study focused on a sample of highly-stressed mothers, who were predominantly biological mothers of adolescents. Because of the specificity of this sample, findings may not generalize to other samples. Future studies are needed to test the effects of parent-mindfulness interventions with other caregivers, including, for example, fathers or other guardians. Furthermore, the present study focused in large part on emotion-related aspects of mindfulness mediated changes in neurobiology; however, other cognitive mechanisms have been proposed, including enhanced attentional control, working memory, and other executive functions not necessarily involved in emotional experiences and functioning. Therefore, future studies investigating such potential mechanisms of parent mindfulness interventions are also needed, as these neural changes may have unique effects on parenting behaviors.

Despite these limitations, the present study examined the neural mechanisms of a parent mindfulness intervention within a randomized controlled trial, and suggests that parent mindfulness intervention may affect neurobiology related to increased emotional awareness, with implications for changes in parent behavior. Specifically, intervention-related changes in brain function and connectivity tracked changes in mindful parenting and mothers’ emotional reactivity to stressful parent-adolescent interactions. Altogether, results of the present study suggest that future parent-focused interventions targeting caregivers’ emotional reactivity should focus on body-based emotional awareness (e.g., through meditative practices such as the mindful body scan), in order to decrease emotion arousal and increase positive emotional exchanges between parents and their adolescent children.