Maternal Socioeconomic Disadvantage, Neural Function During Volitional Emotion Regulation, and Parenting

Abstract

The transition to becoming a mother involves numerous emotional challenges, and the ability to effectively keep negative emotions in check is critical for parenting. Evidence suggests that experiencing socioeconomic disadvantage interferes with parenting adaptations and alters neural processes related to emotion regulation. The present study examined whether socioeconomic disadvantage is associated with diminished neural activation while mothers engaged in volitional (i.e., purposeful) emotion regulation. 59 mothers, at an average of 4 months postpartum, underwent fMRI scanning and completed the Emotion Regulation Task (ERT). When asked to regulate emotions using reappraisal (i.e., Reappraise condition; reframing stimuli in order to decrease negative emotion), mothers with lower income-to-needs ratio exhibited dampened neural activation in the dorsolateral and ventrolateral PFC, middle frontal and middle temporal gyrus, and caudate. Without explicit instructions to down-regulate (i.e., Maintain condition), mothers experiencing lower income also exhibited dampened response in regulatory areas, including the middle frontal and middle temporal gyrus and caudate. Blunted middle frontal gyrus activation across both Reappraise and Maintain conditions was associated with reduced maternal sensitivity during a mother-child interaction task. Results of the present study demonstrate the influence of socioeconomic disadvantage on prefrontal engagement during emotion regulation, which may have downstream consequences for maternal behaviors.

New mothers, particularly first-time mothers, encounter novel, highly stressful situations related to parenting (Deater-Deckard, 1998; Law et al., 2019; Parkes et al., 2015; Vismara et al., 2016). A common challenge for mothers is the ability to regulate their emotions while caring for a distressed child. For instance, infant distress cues (e.g. infant crying) elicit caregiving behaviors (Soltis, 2004), but may also elicit aversive feelings (Leerkes et al., 2011; Lin & McFatter, 2012; Soltis, 2004). Failures of maternal emotion regulation in this context may result in the expression of negative parenting behaviors (Firk et al., 2018; Hajal et al., 2019; Lovejoy et al., 2000; J. Martini et al., 2017). Thus, the transition to motherhood might be an especially critical time to regulate emotions well, given the downstream effects of negative emotion on parenting and infant development (Feldman, 2009; Granat et al., 2017; Rutherford et al., 2015; Tronick & Reck, 2009). The previous literature suggests that mothers experiencing socioeconomic disadvantage are at increased risk of emotion regulation difficulties and of exhibiting less sensitive caregiving toward offspring (Chaudron et al., 2010; Goyal et al., 2010; van Doesum et al., 2007). Furthermore, non-parent populations experiencing socioeconomic disadvantage have exhibited altered neural processing in brain regions related to emotion regulation (for review see Farah, 2017; Gianaros et al., 2008; Hackman et al., 2010; Kim et al., 2013; Muscatell et al., 2012). To date, the associations among socioeconomic disadvantage, neural activation during emotion regulation, and parenting behaviors have not been examined among new mothers. The current study aims to investigate the association between maternal socioeconomic disadvantage (indexed by maternal income-to-needs ratio) and maternal neural function during a volitional emotion regulation task (i.e., the Emotion Regulation Task involving the purposeful regulation of negative emotions). We further investigated the association between neural activation and maternal mood and parenting behaviors during a mother-child interaction task.

Emotion Regulation, Poverty, and Parenting

The ability to volitionally regulate emotions is critical in the context of parenting. Volitional emotion regulation refers to one’s ability to purposefully manipulate or influence the experience and expression of emotions (Gross, 1999; Ochsner et al., 2012). To appropriately respond to infant distress cues, mothers must regulate their own distress and negative emotions (Rutherford et al., 2015). Failure in maternal emotion regulation in this context may result in less sensitive behavior toward the child, including difficulties perceiving, correctly interpreting, and responding appropriately and promptly to infant signals (Ainsworth et al., 1978; Lovejoy et al., 2000; Martini et al., 2004). Indeed, volitional processes and cognitive control capacities have been associated with more positive parenting behaviors (for review see Crandall et al., 2015; Lorber, 2012; Lorber & O’Leary, 2005). Furthermore, deficits in volitional emotion regulation have been related to greater vulnerability to affective disorders such as depression (Aker et al., 2014; Gross & Muñoz, 1995; Joormann & Gotlib, 2010), which can further impair a mother’s ability to appropriately care for her child (Matijasevich et al., 2015; Paulson et al., 2006).

Exposure to poverty predicts deficits in emotion regulation, marked by decreased tendency to engage in and less success in using adaptive emotion regulation skills (Elsayed et al., 2021; Herd et al., 2020). The cumulative experience of chaotic environments, financial stressors, and limited social support and resources overwhelms the emotion regulatory system. This further contributes to higher rates of maternal depression (Belle & Doucet, 2003; Heflin & Iceland, 2009; Reading & Reynolds, 2001) and limits mothers’ capacity for sensitive caregiving (Finegood et al., 2016). Living in a low-income household is a significant risk factor for negative parenting behaviors, including hostile behaviors and child abuse, and reduced positive behaviors, including less affection, animation, and stimulation of infant development (Coulton et al., 1995; Eckenrode et al., 2014; Finegood et al., 2016). Previous literature suggests that volitional forms of emotion regulation are especially important in parenting to help mothers maintain emotional health, promote sensitive caregiving, and potentially protect children from the downstream effects of socioeconomic risk (Herd et al., 2020; Milojevich et al., 2020). To date, however, engagement of emotion regulation brain regions in new mothers, particularly among mothers experiencing socioeconomic disadvantage, has not been used to address questions about emotion regulation, mood, and parenting.

Maternal Brain Response to Infant Cues

Neurobiological adaptations supporting regulatory demands of parenting occur during the postpartum period (Atzil et al., 2011; Hipwell et al., 2015; Kim, 2016; for review see Kim et al., 2016; Pereira, 2016). Mothers experience pronounced structural changes during the postpartum period in prefrontal regulatory regions (Hoekzema et al., 2017; Kim, Leckman, Mayes, Feldman, et al., 2010). Functionally, parents show heightened activation to infant cues in “parental brain” networks that overlap considerably with regions involved in emotional response (e.g., amygdala, insula, striatum) and regulation (prefrontal cortex, anterior cingulate cortex) (Rutherford et al., 2015; Swain, 2011). For example, new mothers exhibit heightened activation in middle and lateral prefrontal cortices in response to infant distress cues, which are associated with more sensitive parenting behaviors (Gholampour et al., 2020; Kim et al., 2016; Kim, Leckman, Mayes, Newman, et al., 2010; Musser et al., 2012; Swain et al., 2008). Maternal neural response is also observed to change across the postpartum period (e.g. increased prefrontal, orbitofrontal cortex response over time), perhaps reflecting developing parenting and regulatory capacities (Gingnell et al., 2015; Parsons et al., 2017). Taken together, neurobiological functions that support emotion regulation are considered critical for optimal caregiving behaviors and positive infant development.

What factors might determine which mothers show these neurobiological adaptations? Environmental factors, such as socioeconomic disadvantage, may influence parenting behaviors and neural activation in regions implicated in emotion regulation (Kim, 2021; Kim et al., 2020). Socioeconomic disadvantage has been shown to interfere with neural adaptations critical to parenting (Gianaros et al., 2008; Kim et al., 2016, 2017). For instance, while passively listening to infant cries, low-income first-time mothers exhibit diminished neural activity in medial and middle prefrontal regions (Kim et al., 2016) - brain regions which are involved in regulating negative emotions (Buhle et al., 2014; Goldin et al., 2008; Ochsner et al., 2012; Ochsner & Gross, 2005). It is important to note, however, that Kim and colleagues (2016) did not specifically examine neural activation while mothers were explicitly engaged in emotion regulation. Thus, it remains unclear whether socioeconomic disadvantage is associated with altered neural activation during volitional emotion regulation, and whether altered neural activation is further associated with negative parenting outcomes.

Neural Correlates of Volitional Emotion Regulation

The current study focuses on cognitive reappraisal – a volitional and adaptive emotion regulation strategy that involves reframing or changing the meaning of a stimulus, which can change the emotion elicited by it, for example, making it less aversive (Gross, 1999). One candidate for examining volitional emotion regulation is the use of the Emotion Regulation Task (ERT). Variations of the ERT paradigm have been systematically and extensively used in non-parent populations (Kim et al., 2013; McRae et al., 2008, 2010; Ochsner et al., 2002; Phan et al., 2005). In this task, participants are shown a series of negatively valenced images and are instructed to either passively view the images, or reinterpret (i.e., reappraise) the affective content of the images to make them less aversive. Using the ERT paradigm, a number of studies have characterized the brain areas engaged when passively viewing images (i.e., unaltered emotional response and/or implicit forms of emotion regulation) and brain areas necessary when deliberately changing an evaluation of an emotional stimulus (i.e., cognitive reappraisal ). Existing studies have shown the importance of prefrontal control regions including the dorsolateral PFC (DLPFC) and ventrolateral PFC (VLPFC) in implementing cognitive strategies to modulate emotion reactive regions such as the amygdala (Cutuli, 2014; Goldin et al., 2008; Ochsner et al., 2012; Ochsner & Gross, 2008).

Recent findings suggest that exposure to childhood socioeconomic disadvantage is associated with a reduced capacity to volitionally regulate emotions in adulthood. Lower childhood family income was associated with dampened neural activation in DLPFC and VLPFC regions during reappraisal and increased activation in emotion processing regions (i.e., amygdala) among non-parent adults (irrespective of adult income; Kim et al., 2013). In the current study, we examined whether socioeconomic disadvantages may be associated with similar alterations in neural patterns during volitional emotion regulation, which may further be associated with the dependent variables of parenting behaviors and depressed mood among new mothers.

Current Study

The current study investigated whether maternal socioeconomic disadvantage experienced by first-time mothers was linked to diminished neural activation in regions involved in volitional emotion regulation while performing an Emotion Regulation Task. We hypothesized that the magnitude of socioeconomic disadvantage experienced by mothers (indexed using maternal income-to-needs ratio) would be associated with blunted activation in prefrontal regions (DLPFC, VLPFC) associated with volitional emotion regulation. We selected maternal income-to-needs ratio (INR) as an indicator of socioeconomic disadvantage; this approach captures concurrent changes in risk, rather than more static indicators such as maternal education. We investigated the association between INR and neural activation in prefrontal regions during volitional emotion regulation, and explored potential relations with self-reported maternal depressive mood and parenting behaviors observed during a mother-infant interaction task.

Methods

Participants

Mothers with their first biological infants, at an average of 4.4 months postpartum, were recruited through flyer distribution and collaborations with Denver and Boulder county postnatal clinics and programs (e.g., Denver Health, Women, Infant, Clinic (WIC), and Prenatal Plus programs). The study targeted mothers experiencing low and middle income (maternal income-to-needs ratio between 0 and 6; see Maternal Income-to-Needs section). The first year postpartum can be a challenging time for many first-time mothers, with elevated rates of maternal depressed mood and parenting stress exhibited across the first year (Goodman, 2004; Oddi et al., 2013; Patel et al., 2012); thus, we included mothers from a range of postpartum months (0.89–10.65 months). Interested participants contacted the research team and were screened for eligibility. Illicit psychoactive drug use, current psychiatric/neurological illness (other than mood disorders), IQ scores below 70, reports of infant chromosomal anomalies and serious medical issues (i.e., premature birth, low birth weight, neonatal intensive care unit (NICU) stay) were grounds for exclusion. Sixty-five participants completed MRI scanning. 5 participants were excluded due to excessive motion (above 20% TRs removed; motion cut-off was framewise displacement in any direction exceeding 0.5 mm) and 1 was excluded because she completed only 1 run of the emotion regulation task (ERT), resulting in 59 participants in the final analyses. Demographic characteristics are described in Table 1¹.

Table 1.

Participant characteristics

	N(%)	Mean ± SD	Range
Maternal age (years)	--	25.98±5.52	18–37
Maternal Race/Ethnicity
Hispanic/Latina	26 (44.1)	--	--
Caucasian/White	25 (42.4)	--	--
African-American/Black	3 (5.1)	--	--
Asian	1 (1.7)
Other	4 (6.8)	--	--
Maternal education (years)	--	14.10±2.46	9–20
Postpartum month at the time of fMRI scans	--	4.58±1.98	0.89–10.65
Income-to-needs ratio	--	2.55±1.53	0.43–6.24
Interval between home and fMRI visits (weeks)	--	4.52±4.49	0.29–27.14
Infant sex (female)	37(62.7)	--	--
History of psychiatric disorder (Yes)	25 (42.4)
Psychiatric medication use (Yes)	6 (10.2)
BDI Depressive symptoms	--	7.83±5.49	0–26
Maternal IQ	--	99.37±11.83	73–125
Ever Breastfed (Yes)	59(100)	--	--
Right Handedness (Yes)*	53(89.8)	--	--

^*missing 1 participant’s handedness data

Procedures

Participant eligibility was assessed during an initial phone screen with a trained research assistant. If eligibility criteria were met, a home visit was scheduled. At each phase of the study, mothers provided written consent in accordance with the University of Denver’s Institutional Review Board guidelines.

First, trained researchers administered printed survey measures in the participants’ home. Demographic data (see Table 1) and several questionnaires that measured mood were collected during a structured interview with the mother. IQ was assessed using the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler et al., 1997) and the mother-infant interaction was video-recorded.

Second, participants completed the fMRI portion of the study. During the second phase, participants were asked to visit the Intermountain Neuroimaging Center at the University of Colorado – Boulder. The average time interval between the two phases was approximately one month. Participants were provided $100 compensation for their participation in the home visit and $100 for participation in the fMRI visit. This amount was based on an hourly rate of $25. In multiple projects with low-income populations, we found this amount to be respectful of participants’ time and effort, but not overly high to have the risk of being coercive.

Measures

Maternal Income-to-Needs Ratio

Maternal socioeconomic disadvantage was assessed based on the mother’s current income-to-needs ratio (INR). The ratio of maternal income-to-needs was computed by dividing the total mother-reported family income by the federal poverty threshold, adjusted for the number of individuals in the home. The INR was calculated based on the mother-reported family income within the last 12 months from the date of the home visit. In the present study, INR ranged from 0.43 to 6.24 (M = 2.55, SD = 1.53). Although INR was used as a continuous measure, for descriptive purposes it is important to note that an INR of less than 2 is considered low income, whereas an income-to-ratio between 2 and 6 is considered middle income (Cauthen & Fass, 2008). Thus,, 45.8% percent of the current sample were considered as experiencing low income.

Depressive Mood

Depressive symptoms were measured using the Beck Depression Inventory (BDI; Beck et al., 1988). The BDI is a widely used self-report questionnaire that includes 21 items assessing depressive symptoms in the past week, including low mood, anhedonia, feelings of guilt, irritability, fatigue, and poor appetite. Each item is answered on a 4-point Likert scale (0 = symptom absent to 3 = symptom severe). Higher total scores suggest increased depressive symptom severity (<10= none or minimal depression, 10–18 = mild to moderate depression, 19–29 = moderate to severe depression, 30–63 = severe depression). 39 participants (66.1%) within our study reported none/minimal depression, 18 participants (30.5%) reported mild to moderate depression, 2 participants (3.4%) reported moderate to severe depression, and no participants reported severe depression.

Maternal Behaviors

Maternal behaviors were rated based on a videotaped mother-child interaction task that occurred in the child’s home. To assess maternal behaviors, the study employed a 15-minute videotaped mother-child interaction task, which involved unstructured and natural play between mothers and infants. Mothers were instructed to interact with their child as they normally would without the use of toys. Parenting behaviors were rated using the Emotional Availability (EA) scales (Biringen et al., 1998; Saunders et al., 2015). Two researchers were trained and certified by Dr. Biringen, creator of the Emotional Availability scales. The two researchers coded the videos, with 24% overlap. The average intraclass correlation (ICC) was 0.91. The EA includes 4 parenting scales: sensitivity, structuring, non-intrusiveness, and non-hostility. The current study focused on the sensitivity subscale, which refers to a mother’s displays of authentic and congruent interest and pleasure while playing with her infant. This can be demonstrated through warm smiles, laughter, eye contact, affectionate words, and comforting and playful physical contact. Highly sensitive mothers accurately read and flexibly respond to her child’s signals. Maternal sensitivity on the EA scales is scored from highly insensitive (1) to highly sensitive (9). In the present study, sensitivity ranged from 3 (somewhat insensitive) to 7 (generally sensitive). Somewhat insensitive parenting typically describes either an overly active and overbearing style or a passive and noninteractive style. By contrast, generally sensitive parenting describes very positive interactions, which could be down-marked slightly due to mother’s subtle preoccupation with her thoughts at brief moments or reduced creativity in the interaction.

fMRI Paradigm

Emotion Regulation Task

Whole-brain BOLD signal was measured while the participants were engaged in the Emotion Regulation Task (ERT) (Kim et al., 2013; Phan et al., 2005). All images in the Emotion Regulation Task were selected from the International Affective Picture System (IAPS) (Lang et al., 1997; see Appendix for corresponding IAPS image numbers). The task includes three conditions, which were presented in a randomized order. During the Look condition, participants were asked to simply look at a series of emotionally neutral images. During the Maintain condition, participants were instructed to look and maintain their emotional responses to a series of negatively valenced images. During the Reappraise condition, participants were asked to cognitively reappraise (e.g., reinterpret or reframe) aspects of a series of negatively valenced images in ways that make them less aversive. For example, an image of a crying woman could be interpreted as tears of joy as opposed to tears of sadness. During fMRI scanning, each block was followed by a self-report 4-point scale (1=least negative, 4=extremely negative) asking participants to rate the intensity of their negative affect. The task included 3 blocks. At the start of each trial, an instruction was presented (i.e., LOOK, MAINTAIN, or REAPPRAISE) for 5 seconds, followed by a series of images (a total of 4 images) presented at 5 seconds each. Each trial was followed by the affective rating scale presented for 5 seconds and a fixation cross presented for 20 seconds. Each instruction was presented 6 times (18 blocks total). Total task duration was 15 minutes and 45 seconds.

Prior to entering the scanner, participants received instruction in cognitive reappraisal and practiced with a research assistant. The research assistant gave explicit instructions and a number of examples of how to transform a depicted scenario into a less negative or more positive situation. In order to confirm that participants fully understood the reappraisal process, participants were presented with negative images and were instructed to narrate their reappraisal out loud to the research assistant. The fMRI sessions were conducted only when participants fully understood each condition.

fMRI Data Acquisition and Processing

Scanning was conducted using two different scanners due to a scanner update; scanners were Siemens Trio and Siemens Prisma. 36 participants were scanned on the Siemens Trio and 23 on the Siemens Prisma. Both were 3.0 T Siemens magnet scanners using a standard 32-channel head coil. Images were acquired using 405 T2*-weighted echo-planar-imaging (EPI) volumes. The parameters of T2* functional sequences were matched across the scanners (TR = 2300ms, TE = 27ms; flip angle = 73; field of view = 192mm; matrix size, 64 × 64; 36 axial slices; voxels = 3 mm³). For normalization and localization, high resolution anatomical T1-weighted images were also acquired using the 3D magnetization-prepared rapid gradient echo (MPRAGE) protocol. For the Siemens Trio, high resolution T1 - weighted magnetization prepared rapid gradient -echo (MPRAGE) images were acquired with the following parameters: 192 sagittal slices, TR = 2530 ms, TE = 1.64 ms, flip angle = 7°, FOV = 256 mm 2 and voxel size 1 × 1 × 1 mm. For the Siemens Prisma, T1 sequence parameters were 224 sagittal slices, TR = 2400 ms, TE = 2.07 ms, flip angle = 8°, FOV = 256 mm 2 and voxel size 0.8 × 0.8 × 0.8 mm.

Analysis of Functional Neuroimages software (AFNI; Cox, 1996) was used for preprocessing and statistical analysis. The first four pre-steady-state volumes (two dummy TRs and two additional TRs) for each run were discarded. Data preprocessing included slice time correction, (i.e., images were realigned to the last volume of the last run to correct for head movement) and motion correction (images with motion greater than 0.5 mm and 0.5 degrees in any direction were censored). Participants with more than 20 percent of TRs removed were excluded from data analyses. Within the data included in the analysis, the range of number of volumes censored was 0–80 (M = 13.56 ± 20.14; median = 6; ≤ 20% of the total volumes). Thus, at least 90 volumes for each condition were included in the present analysis. Anatomical and functional images were normalized to Talaraich space (Talairach & Tournoux, 1988). Realigned functional images were co-registered to anatomical images. Images were spatially smoothed with a 6-mm root-mean-square deviation Gaussian blur.

Analysis

Covariate Selection

The following sociodemographic variables were tested to see if they were associated p < .05 with the independent variable (income-to-needs ratio): maternal age, education, IQ, race and ethnicity, handedness, maternal psychiatric medication use, maternal history of depression, anxiety or other psychiatric disorder, breastfeeding status, infant age, and infant sex. After examining correlations, the following covariates were selected to include in the whole brain analysis: maternal age, IQ, and scanner type.

Behavioral Ratings Analysis

Affective ratings were analyzed using SPSS. An independent samples t-test was conducted to test the mean affective rating difference between negative (Maintain and Reappraise images) and neutral (Look) images. A repeated measures ANCOVA was conducted with maternal income-to-needs ratio as a between-subject continuous variable, and affective rating conditions (Look, Maintain, Reappraisal ratings) as within-subject factors, with covariates IQ and maternal age.

fMRI Analysis

Analysis of Functional Neuroimages software’s 3dLME was utilized to create a whole-brain linear mixed effects model with maternal income-to-needs scores as a between-subject quantitative variable, and condition (Look, Maintain, Reappraise) as within-subject factors, controlling for scanner type, IQ, and maternal age (see Sample Characteristics section). A whole brain mask was created based on 90% EPI coverage. The cluster extent threshold was set to k ≥ 33 with a height threshold of p < .001, equivalent to a whole brain corrected false positive probability of p < .05, as calculated by 3dClustSim using the spatial autocorrelation function (ACF) coordinates. The new -acf option was used in 3dClustSim, which provides a more accurate control of false positives (Cox et al., 2017). Using the extracted average beta values from significant clusters, posthoc regression analyses were conducted using SPSS statistical software (Statistical Package for the Social Sciences version 26, Chicago, IL).

The association between self-reported depressive symptoms (using the Beck Depression Inventory), maternal sensitivity scores, and neural activation (using posthoc data pulled from the whole brain analysis) within PFC regions were further examined using repeated measure ANCOVA in SPSS. We focused on PFC regions given its consistent involvement in volitional emotion regulation (Cutuli, 2014; Goldin et al., 2008; Ochsner et al., 2012; Ochsner & Gross, 2005). A mediation analysis among maternal income-to-needs ratio, brain activation, and maternal sensitivity and depressive symptoms was not performed due to the cross-sectional nature of the data.

Results

Sample Characteristics

Table 1 includes descriptive statistics for all participants. Maternal income-to-needs ratio was not significantly correlated to maternal sensitivity or depressive symptoms. Demographic variables correlated with maternal income-to-needs ratio were maternal age (r = .50, p < .001), maternal education (r = .51, p < .001), and maternal IQ (r = .37, p = .005), such that higher maternal income-to-needs ratio is associated with older age, higher education, and higher IQ. Maternal sensitivity was associated with higher maternal education (r = .35, p = .007) and higher IQ (r = .44, p < .001). Maternal age, education, and IQ were all highly correlated with one another (ps < .01). Maternal age was selected as a covariate rather than maternal education; this was because mothers’ age was more relevant to the postpartum period, rather than a static indicator of risk. Maternal age, IQ, and scanner type were included as covariates in the whole brain analysis.

Behavioral Data

Greater negative emotion was reported for the negative images (M = 2.66, SD = 0.50) compared to neutral images (M = 1.09, SD = 0.15) (t(58) = 25.26, p < .001). Greater negative emotion was reported for the Maintain trials (M = 2.94, SD = 0.58) compared to the Reappraise trials (M = 2.38, SD = 0.55) (t(58) = 8.23, p < .001), which suggests that the reappraisal instruction reduced negative emotion. To test the influence of maternal income-to-needs ratio on behavioral ratings of negative emotion, an ANCOVA was conducted for the dependent measure of emotion ratings, with maternal income-to-needs ratio as a between-subject factor and Condition (Look, Maintain, Reappraise) as within-subject factors, controlling for IQ and maternal age. No significant interactions or main effects were observed, p’s > .05. We additionally tested the influence of maternal depressive symptoms and demographic factors (i.e., maternal age, IQ, education) on behavioral ratings of negative emotion; findings were not significant.

Maternal Income-to-Needs-Ratio and Neural Emotion Regulation

A two-way maternal income-to-needs ratio x Condition (Look, Maintain, Reappraise) interaction with IQ, maternal age, and scanner type as covariates revealed several significant clusters (Table 2; Figures 1–3) – left caudate, right and left inferior frontal gyrus (including right and left ventrolateral PFC), right and left precentral gyrus, right middle temporal gyrus, and left superior frontal gyrus (including left dorsolateral PFC), p< .05, corrected for multiple comparisons.. Parallel analyses were conducted with only scanner type as a covariate (without the covariates maternal IQ and age); results were similar when these covariates were excluded from the analysis. Analyses also remained significant when maternal education was added as an additional covariate to the model (in addition to IQ, maternal age, and scanner type).

Table 2.

Brain areas showing INR x Condition (Look, Maintain, Reappraise) interactions

Regions	BA	Side	X	Y	Z	Voxels	F
Caudate	--	L	−10	5	8	196	13.03
Inferior Frontal Gyrus	47	R	56	23	−1	76	12.80
Precentral Gyrus	6	R	44	−1	47	50	14.40
Middle Temporal Gyrus	39	R	38	−70	20	44	12.86
Superior Frontal Gyrus	8	L	−31	23	50	41	13.74
Precentral Gyrus	6	L	−46	−4	50	41	14.11
Inferior Frontal Gyrus	47	L	−46	23	−1	34	11.82

(Note: x, y, z are Talaraich coordinates)

Figure 1.

Region showing a significant maternal income-to-needs x condition (Look, Maintain, Reappraise) interaction: left caudate (x, y, z = −10, 5, 8, k = 196, voxel-wise threshold p < .001). The right panel includes scatterplots describing the association between maternal income-to-needs ratio and neural activity during the following conditions: (a) Maintain (r = .28, p = .035), and (b) Reappraise (r = .48, p < .001).

Figure 3.

Region showing a significant maternal income-to-needs x condition (Look, Maintain, Reappraise) interaction: Right precentral gyrus (BA 6; x, y, z = 44, −1, −47, k = 50, voxel-wise threshold p < .001). The left panel includes scatterplots describing the association between maternal income-to-needs ratio and neural activity during the following conditions: (a) Maintain (r = .41, p = .001), and (b) Reappraise (r = .46, p < .001). The right panel includes scatterplots describing the association between maternal sensitivity and right precentral gyrus activation during (a) Maintain (r = .26, p =.045) and (b) Reappraise conditions (r = .35, p =.007).

Post-hoc Analyses

Using the extracted beta values, post-hoc regression analyses were conducted with maternal income-to-needs ratio. Post-hoc regression suggests that during the Look condition, lower income-to-needs ratio was associated with increased activations in the left inferior frontal gyrus (b = −1.12, t = −2.13, p = .038).

During the Maintain condition, lower maternal income-to-needs ratio was associated with reduced activation in the following regions: left caudate (b = 3.01, t = 2.16, p = .035), right precentral gyrus (b = 2.20, t = 3.35, p = .001), and right middle temporal gyrus (b = 2.93, t = 2.70, p = .009).

During the Reappraise condition, lower maternal income-to-needs ratio was associated with reduced activation in the following regions: left caudate (b = 4.06, t = 4.09, p < .001), right inferior frontal gyrus (b = 1.79, t = 2.87, p = .006), right precentral gyrus (b = 1.87, t = 3.88, p < .001), right middle temporal gyrus (b = 3.61, t = 3.58, p = .001), left superior frontal gyrus (b = 1.28, t = 2.59, p = .012), left precentral gyrus (b = 1.06, t = 2.92, p = .005), and left inferior frontal gyrus (b = 1.00, t = 2.21, p = .031).

Maternal Depressive Mood and Neural Emotion Regulation

The associations between maternal depressive mood and the identified PFC clusters were investigated. A posthoc repeated measure ANCOVA analysis (Look, Maintain, Reappraise conditions) was conducted with maternal depression as a between subject-variable. No significant associations were identified, p’s > .05.

Maternal Behaviors and Neural Emotion Regulation

The associations between maternal sensitivity and identified PFC clusters were investigated. A posthoc repeated measure ANCOVA analysis (Look, Maintain, Reappraise conditions) was conducted with maternal sensitivity as a between subject-variable. A significant association was indicated in the right precentral gyrus [F(2,114)=3.54, p =.032]. Posthoc regressions further suggest that lower sensitivity was associated with reduced activation in the right precentral gyrus during the Maintain (b = 1.19, t = 2.05, p = .045) and Reappraise conditions (b = 1.19, t = 2.79, p = .007).

Discussion

The present study examined whether socioeconomic disadvantage experienced by first-time mothers is associated with dampened neural activation in regions involved in volitional emotion regulation. When participants were instructed to engage in cognitive reappraisal (i.e., Reappraise condition), lower maternal income-to-needs ratio was associated with dampened activation in the following areas: superior frontal gyrus (including the dorsolateral PFC (DLPFC)), inferior frontal gyrus (including the ventrolateral PFC (VLPFC)), precentral gyrus (including the middle frontal gyrus), middle temporal gyrus, and caudate. When asked to naturally respond to negative images (i.e., Maintain condition), mothers experiencing lower income exhibited dampened neural activation in regulatory areas, including the caudate, precentral gyrus (including the middle frontal gyrus), and middle temporal regions. Blunted neural activation was in turn associated with fewer expressions of positive parenting behaviors. Dampened activation in the precentral gyrus, during Maintain and Reappraise conditions, was associated with lower maternal sensitivity (i.e., fewer displays of maternal interest, pleasure, and warmth during an infant play interaction).

Volitional Emotion Regulation

Results showed that lower maternal income was associated with reduced activation in prefrontal regions previously associated with volitional, instructed emotion regulation. VLPFC regions are implicated in the selection of salient information and response inhibition (Badre & Wagner, 2007; Levy & Wagner, 2011) and the DLFPC is implicated in working memory, selective attention, and goal directed behaviors (Curtis & D’Esposito, 2003; Johnson et al., 2007). These regions may be recruited to maintain the goal to reappraise, direct attention to relevant stimuli, and select appropriate appraisals. Further, increased activation in prefrontal control regions is associated with successful regulation of negative emotions (Buhle et al., 2014; Kalisch, 2009; McRae et al., 2010; Ochsner et al., 2012). The DLPFC and VLPFC are included in neural networks of maternal behavioral response to infant distress, and play a particularly important role in evaluating the emotional value of infant cues (Kikuchi & Noriuchi, 2015). Although the current study did not utilize infant cue stimuli, reduced responses in these prefrontal regions have been interpreted as lower sensitivity to infant cues and lower motivation to attend to these cues in a prior, partially overlapping sample (Kim et al., 2016).

In addition to altered prefrontal functioning, lower maternal income-to-needs ratio also predicted dampened activation in the caudate during reappraisal. Previous studies have shown an increase in caudate activity during reappraisal, including associated decreases in self-reported negative emotion (McRae et al., 2008; Opialla et al., 2015; Wager et al., 2008). The caudate is part of the striatum and has been implicated in executive processes and goal-directed action (Grahn et al., 2008; Monchi et al., 2006), suggesting the importance of this region in effective emotion regulation. Furthermore, in the context of parenting, activations in striatal (i.e., caudate) regions have been associated with reward-related and motivation responses. Caudate activation is heightened in response to infant distress cues, and may reflect reward detection and the initiation of emotion-induced action (Kikuchi & Noriuchi, 2015; Swain et al., 2008). By contrast, depressed mothers are less likely to engage striatal regions in response to their own infant cries (Laurent & Ablow, 2012), which may underlie affiliative bonding difficulties and reduced motivation to respond to infant cues among depressed mothers (Moehler et al., 2006).

To our knowledge, the current study is the first to examine neural engagement while mothers are actively engaged in a volitional emotion regulation task with non-infant stimuli. Previous studies that used overlapping samples with the present study (see Methods section) showed that mothers experiencing lower income and mothers exposed to severe stress showed reduced prefrontal activations when passively responding to infant cues (i.e., infant cries, infant faces) (Kim et al., 2017, 2020). Findings suggest that first-time mothers with lower income may experience greater amounts of stress related to both socioeconomic disadvantage and the transition to parenthood; the combination of these stressors may have an impact on neural regulatory processes (Troy et al., 2010, 2017).

Uninstructed Emotion Regulation

Although not initially hypothesized, when viewing negative images in the absence of explicit instructions to regulate, lower income-to-needs ratio was associated with reduced neural activation in the caudate, precentral gyrus (including the middle frontal gyrus), and middle temporal regions. By contrast, mothers experiencing middle-income exhibited increased activation, suggesting possible uninstructed regulation of negative affect even during a non-regulation condition (Etkin et al., 2015; Gyurak et al., 2011; Koole & Rothermund, 2011; Silvers et al., 2015; Wang et al., 2017). Middle and superior temporal gyri have been implicated in reappraisal studies and may be involved in the detection of emotion-relevant information that assist in emotion regulation (Buhle et al., 2014). The medial frontal cortices such as the precentral gyrus have also been associated with inhibitory control, which is necessary to inhibit natural responses to negative stimuli (Li et al., 2006; Padmala & Pessoa, 2010).

While emotion regulation via reappraisal involves deliberate transformation of negative affect, an increasing body of evidence suggests a more uninstructed process of emotion regulation (Etkin et al., 2015; Gyurak et al., 2011; Wang et al., 2017). Although neural studies of uninstructed emotion regulation remain scarce, literature suggests a significant overlap between brain areas typically implicated in both deliberate and uninstructed emotion regulation (e.g., prefrontal control areas including DLFPC and VLPFC, temporal gyrus, and precentral gyrus; Silvers et al., 2015; Wang et al., 2017), which is consistent with our findings. Uninstructed forms of emotion regulation are said to be less cognitively costly and decrease the physiological responses to emotion-eliciting stimuli (Yuan et al., 2015). Thus, uninstructed emotion modulation may be adaptive in the context of parenting, because it helps mothers to down-regulate negative affect and physiological responses without the cognitive costs of explicit forms of regulation. Previous studies have shown that healthy mothers exhibit reduced physiological responses to stressors, which is in turn associated with better emotional health (i.e., lower depressive mood) (Slattery & Neumann, 2008). It is possible that reduced stress responsivity and better maternal mood result from reappraisal (both uninstructed and instructed) by the maternal brain. Given the scarcity of uninstructed emotion regulation studies, however, these findings remain exploratory and should be interpreted cautiously.

Parenting

Results further showed that reduced activation of the precentral gyrus (including the middle frontal gyrus) during Reappraise and Maintain conditions was associated with reduced expressions of maternal sensitive behavior during a mother-infant interaction task. The precentral gyrus has been implicated in response inhibition and attentional control (Japee et al., 2015; Li et al., 2006; Padmala & Pessoa, 2010; Yamasaki et al., 2002). Research suggests that the middle frontal gyrus is also involved in the downregulation of emotional responses (Grecucci et al., 2013; Wang et al., 2020). Perhaps decreased activity in the precentral gyrus suggests mothers’ reduced capacity to regulate their own emotions, which further impairs their ability (or motivation) to attend to their child’s needs. Neural models of parenting also show that the engagement of the middle frontal gyrus is related to maternal empathy and positive mother-infant play (Bak et al., 2021; Kim, Leckman, Mayes, Newman, et al., 2010; Wan et al., 2014). Consistent with this, dampened engagement of the middle frontal gyrus was related to fewer expressions of sensitive parenting behaviors in the current study, which may have downstream effects on child development.

Limitations and Future Directions

The present findings should be considered in light of several limitations. First, it is difficult to infer the causality of present findings, as all measures were assessed concurrently. It is possible that neural findings may prospectively predict psychological vulnerability or greater difficulties in parenting. Future investigation using a longitudinal design would be important to investigate whether mothers who exhibit aberrant neural activation during emotion regulation would later develop affective disorders or exhibit more negative mother-infant relationships. It is also important to note that income-to-needs ratio was not directly associated with maternal sensitivity or depressive symptoms in the current study. It is possible that brain activation may be a more direct measure of individual differences, rather than more proximal factors (e.g., income-to-needs ratio). Alternatively, the current study may have been under-powered to detect associations between income-to-needs ratio and maternal mood and parenting outcomes. Future longitudinal research with larger samples is needed to examine the mediating role of mother’s neural processes in the link between socioeconomic disadvantage and postpartum outcomes.

Second, the range of maternal depressive symptoms and observed sensitivity was constrained in the current study. Surprisingly, no significant association was observed between the identified neural recruitment during emotion regulation and maternal depressive symptoms in the present study. This may have been because the majority of mothers exhibited minimal current depressive symptoms, and less than 5% reported moderate to severe symptoms. Mothers in our sample also did not exhibit highly insensitive behaviors toward their child (e.g., extreme insensitivity to the child’s signals or little knowledge of child-rearing techniques). The current study underlines the influence of maternal socioeconomic disadvantage even among a generally healthy community population; however, it is unclear how findings would present in more high-risk samples. Future studies should investigate volitional emotional regulation among mothers with heightened risk factors for depression and child maltreatment.

Third, the current study assessed maternal emotion regulation within the context of non-infant relevant stimuli. The Emotion Regulation Task (ERT) utilizes images selected from the International Affective Picture System (IAPS). The ERT and IAPS pictures are widely used and well-validated, which is a strength of the current study. Additionally, the present findings are consistent with prior research showing that maternal emotion regulation assessed within general (i.e., not parenting-specific) contexts can still have important implications for parenting (Brenning et al., 2020; Lorber, 2012b; Morelen et al., 2016). Although the present study helps to build the foundation for future research in this area, it is important that subsequent studies utilize infant-relevant stimuli in order to more directly extrapolate findings to the parenting context. Future research could examine mothers’ neural regulation of emotion in response to negative infant stimuli, such as infant cry sounds, sad baby pictures, or challenging parenting situations.

Fourth, it is important to note that prefrontal regions such as the DLPFC and VLPFC engage in domain general processes that also support executive functioning in non-emotional contexts (Barbey et al., 2013; Levy & Wagner, 2011). It is unclear the extent to which general executive functioning and emotion regulation processes recruit the same or merely similar neural regions (Ochsner & Gross, 2008). In this study we controlled for IQ, one indicator of executive ability, in order to discern whether the results were due to differences in general executive functioning (Blair, 2006; Salthouse & Pink, 2008). To our knowledge, neuroimaging studies directly comparing cognitive control in the context of emotional and non-emotional stimuli remain scarce (Cromheeke & Mueller, 2014; Song et al., 2017). Future neuroimaging reappraisal studies should incorporate measures that assess general executive processes (e.g., Stroop Task, Go/NoGo tasks) to directly compare neural regions involved in emotion regulation and non-emotional forms of cognitive control (Giuliani et al., 2019).

Fifth, it is important to note that no behavioral level differences (ratings of negative emotion and self-reported depressive symptoms) were observed as a function of maternal income-to-needs ratio. This is consistent with previous individual differences studies of reappraisal that showed discrepancies between subjective ratings and neural findings (Kim et al., 2013; McRae et al., 2008). However, it is possible that self-reported ratings of negative emotion may not be sensitive enough to detect subtle differences in emotional state. Future studies should incorporate physiological measures (e.g., heart rate, physiological arousal) that may better capture subtle differences in affective state.

Finally, it is unclear which aspect of maternal socioeconomic disadvantage (e.g., family turmoil, chaos, etc.) most strongly correlates with the recruitment of regions associated with emotion regulation. We also assessed mothers’ current, rather than past, exposure to socioeconomic disadvantage. Childhood poverty exposure has been associated with diminished neural activation in adulthood (e.g., dampened DLPFC and VLPFC during reappraisal; Kim et al., 2013). It is unclear if adulthood socioeconomic disadvantage may be correlated with childhood socioeconomic disadvantage in the current study. We cannot rule out the potential influence of childhood poverty experience on neural function observed in new mothers in the present study. Additionally, the current study limited the sample to mothers in low and middle-income environments, in order to focus investigation on this relatively under-studied population (Bornstein et al., 2013; Jang & Vorderstrasse, 2019). Findings may differ in samples with greater levels of income disparity, such as when comparing mothers living in economic disadvantage vs. economic wealth. Further research examining the role of both childhood and adulthood socioeconomic status across levels is needed to address these limitations.

In summary, the current study showed that maternal income-to-needs ratio was associated with diminished neural activation during volitional emotion regulation, as well as when naturally responding to negative stimuli (potentially indicating less uninstructed engagement of regulation processes). Blunted activation in the precentral gyrus, across both Reappraise and Maintain conditions, was related to reduced observed maternal sensitivity. Findings suggest neural risks for difficulties in emotional regulation and positive mother-infant relationships among economically disadvantaged mothers. Research is needed to improve financial, social, and mental health supports that can help to reduce economic stress and enhance emotion regulation among postpartum mothers.

Figure 2.

Region showing a significant maternal income-to-needs x condition (Look, Maintain, Reappraise) interaction: right inferior frontal gyrus, including the VLPFC (BA 47; x, y, z = 56, 23, −1, k = 76, voxel-wise threshold p < .001) and left superior frontal gyrus, including the DLPFC (BA 8; x, y, z = −31, 23, 50, k = 41, voxel-wise threshold p < .001) The right panel includes scatterplots describing the association between maternal income-to-needs ratio and neural activity during Reappraise in the inferior frontal gyrus (r = .36, p = .006) and superior frontal gyrus (r = .33, p = .012).