Parenting and addictions: Current insights from human neuroscience

Abstract

Purpose.

A growing body of human research has documented associations between the maternal brain and maternal substance use and addictions. This neuroscience-informed approach affords the opportunity to unpack potential neurobiological mechanisms that may underscore challenges in maternal caregiving behavior among mothers with addictions and provide new directions for parenting interventions.

Findings.

Consistent with theoretical models of parenting and addictions, five studies evidence both hypo- and hyper-reactivity to infant affective cues across neuroimaging methods and tasks that incorporate both infant face and cry stimuli. Three structural and resting-state brain studies as a function of maternal substance use are also reported.

Conclusions.

While human neuroimaging research converges in showing that maternal substance use is associated with differential reactivity to infant affective cues, further multi-level/multi-modal, longitudinal, and dimensional research is critically needed to advance this area of investigation.

Introduction

Substance use during pregnancy and the postpartum period can have a detrimental impact on both the mother and her developing child¹. In addition to the teratogenic consequences of maternal substance use (1, 2), maternal behavior during the postpartum period (and beyond) may also be negatively impacted by substance use. In particular, two seemingly opposing patterns of disrupted maternal behavior have been observed relative to non-substance-using mothers: first, substance-using mothers may be more passive and disengaged during their interactions with their child (3); and second, substance-using mothers may be more hostile and intrusive (4). Consequently, increasing interest has focused on the measurement of maternal sensitivity to her infant in the prediction of child developmental outcomes in the context of substance use and addictions (5).

Although behavioral methods have been routinely employed to measure maternal sensitivity, recent approaches have adopted neuroscience methods to provide objective and mechanistic insight into the cascade of sensory, attentional, and cognitive processes underscoring maternal behavior (6–8). In particular, functional magnetic resonance imaging (fMRI) and event-related potentials (ERPs) have been employed to study the maternal brain in the presence, and absence, of substance use. These approaches are typically implemented by having mothers view photographs or videos of infants or listen to audio-recordings of infant cries and examining the neural response elicited by these salient infant cues. fMRI has excellent spatial resolution in the identification of neural circuitry implicated in processing infant cues with millimeter resolution, but limited temporal resolution relative to ERPs. ERPs have strong temporal precision and have been utilized in unpacking, with milli-second accuracy, the varying stages of infant cue processing. The combined strengths of these techniques have provided a critical lens through which to understand potential mechanisms that may underscore disrupted maternal behaviors in the presence of substance use, and thus, generating important directions to optimize interventions to support mothers and their developing children.

The Reward-Stress Dysregulation Model of Addictions and Parenting

A decade ago, we reviewed the literature examining neural circuits implicated in maternal behaviors and addictions (9). Notably, there were significant overlaps in neural regions and circuits implicated in reward processing, stress reactivity and regulation, and maternal behaviors. Strikingly, these reward and stress neural processes were also those evidencing dysregulation in addictions. This led to a working neurobiological hypothesis that caregiving challenges observed in parents with substance use disorders may reflect dysregulation secondary to addictions of reward and stress neural circuits implicated in parenting (10, 11). Specifically, in the presence of addictions, infant cues may be less rewarding and more stress-inducing. Paralleling behavioral work, we therefore proposed that the previously observed passive and disengaged maternal behavior may reflect disruptions in reward-related neural circuitry; while intrusive and hostile maternal behaviors may reflect disruptions to stress-related neural circuitry.

When this prior literature review was published, there were no human neuroimaging studies focused on mothers with addictions. Instead, this neurobiological hypothesis was driven largely by preclinical data and non-parent human studies. Since this time, only eight papers to our knowledge have been published probing the maternal brain in the presence of substance use and addictions (Table 1). Importantly, these studies vary with respect to whether mothers met criteria for substance use disorders or were substance-using with or without diagnostic validation of addictions, and we describe the maternal samples accordingly.

Table 1.

Summary of human neuroscience studies of maternal neural functioning and substance use and addictions, including mode, sample details, stimuli, and overview of results.

Authors	Mode	Sample Details	Substance Use (SU) Definition	Stimuli	Hyper-reactivity	Hypo-reactivity
Kim et al. (2017)	fMRI	* 36 mothers previously poly-substance-using in residential treatment * Recruited at 6 months postpartum * Race/Ethnicity: African American, European American, & Hispanic/Latino	* All participants were recruited from an inpatient SU treatment program; current or past year substance dependence as per MINI and reported history of substance use during the most recent pregnancy	* Own and Unknown * Happy and sad faces	* Own happy * Own sad	* Own happy
Landi et al. (2011)	fMRI	* 26 poly-substance-using mothers * 28 non-substance-using mothers * Recruited at 2 months postpartum * Race/Ethnicity: African American, European American, Asian American, African American and European American, & Hispanic/Latino	* Self-report and interview of prenatal and postnatal SU; positive urine toxicology	* Unknown * Happy, sad, and neutral faces * High- and low-distress cries	*Happy face	* All faces * All cries
Lowell et al. (2020)	ERP	* 29 poly-substance-using mothers * 29 non-substance-using mothers * Recruited at 2 months postpartum * Race/Ethnicity: African American, European American, & Hispanic/Latino	* Self-report and interview of current / past problematic SU; prenatal SU; current / past SU treatment	* Unknown * Happy, sad, and neutral faces * High- and low-distress cries	* Neutral faces (P300)	* All faces (N170) * Cries (N100, P300)
Rutherford et al. (2015)	MRI	* 30 poly-substance-using mothers * 34 non-substance-using mothers * Recruited at 3 months postpartum * Race/Ethnicity: African American, European American, Asian American, & Hispanic/Latino	* Self-report and interview of prenatal and postnatal SU; positive urine toxicology	* None - MRI	N/A	N/A
Rutherford et al. (2017)	ERP	* 35 tobacco-smoking mothers * 35 non-smoking mothers * Recruited at 2 months postpartum * Race/Ethnicity: African American, European American, Asian American, African American and European American, & Hispanic/Latino	* Self-report of tobacco-smoking (cigarettes) and no other illicit SU (including cannabis); alcohol use allowed and assessed.	* Unknown * Happy, sad, and neutral faces * High- and low-distress cries	* Neutral faces (P300)	*All faces (N170)
Rutherford et al. (2020)	fMRI	* 32 poly-substance-using mothers * 22 non-substance-using mothers * Recruited at 8 months postpartum * Race/Ethnicity: African American, European American, & Hispanic/Latino	* Self-report and interview of current / past methadone maintenance or SU treatment; current nicotine dependence (FTND); positive urine toxicology; current or past year drug abuse or dependence on the MINI	*Own and unknown * Happy and sad faces * Cries	* Own faces	* Sad faces * Unknown faces
Swain & Ho (2019)	fMRI	* 7 mothers receiving buprenorphine for opioid use disorder * 25 non-substance-using mothers * Recruited at 1 month postpartum * Race/Ethnicity: African American, European American, & Native American	*Receiving buprenorphine treatment	*None – Resting State	* Greater connectivity	* Less connectivity
Wilbanks et al. (2016)	EEG	* 33 tobacco-smoking mothers * 35 non-smoking mothers * Recruited at 3 months postpartum * Race/Ethnicity: African American, European American, Asian American, African American and European American, & Hispanic/Latino	* Self-report of tobacco-smoking (cigarettes) and no other illicit SU (including cannabis); alcohol use allowed and assessed.	*None – Resting State	N/A	* Less aroused state as indicated by EEG frequency bands

Note. Sample size reflects data points included in analyses (if removed across all analyses) rather than sample size recruited. All studies were conducted in the USA. Maternal race / ethnicity included where specifically detailed. Recruitment postpartum is approximate in months. FTND - Fagerstrom Test for Nicotine Dependence. MINI - Mini-International Neuropsychiatric Interview. Unless specific substances are stated, all illicit and legal substances are included in the SU definitions.

While most studies presented in Table 1 have examined neural responses to infant cues, two have examined resting-state neural activity, and one has studied brain structure. With respect to the single structural brain report, poly-substance-using mothers, relative to non-substance-using mothers, had decreased gray matter (GM) volume, particularly in the frontal lobes (12). GM volume changes have been implicated in prior studies of mothers (13, 14), and decreased GM volumes have been observed in studies of individuals with addictions irrespective of parental status (15). With respect to resting-state activity, one electroencephalography study reported an overall decreased spectral power of neural oscillations in tobacco-smoking mothers relative to non-smoking mothers in frontal regions, consistent with a less aroused cortical state in the tobacco-smoking group (16).

An fMRI study examined resting-state functional connectivity in 7 mothers receiving buprenorphine for opioid use disorder (OUD) and 25 non-substance-using mothers across the initial months postpartum (17, 18). In the latter study, functional connectivity between the periaqueductal gray region and hypothalamus at 1 and 4 months postpartum was assessed. Although differences in patterns of connectivity existed between mothers in the OUD group and non-substance-using mothers at 1 month postpartum, the differential functional connectivity patterns had decreased by 4 months postpartum, perhaps reflecting benefits (biologically and/or psychosocially) of continued buprenorphine treatment. Notably, in a study of adults with OUD, changes in ventral striatal reactivity to infant faces have been observed after two weeks of naltrexone treatment (19). Taken together, these two studies highlight the potential for considering how varying pharmacological treatments for OUD may impact the maternal brain.

The remaining studies listed in Table 1 have presented mothers with photographs of infant faces and/or audio-recordings of infant cries to examine neural activity to these typically salient infant cues. Infant face stimuli have varied in emotional expression as well as familiarity (i.e., own vs. unknown); infant cry stimuli have also varied in their intensity (i.e., high-distress vs. low-distress) and familiarity. We have organized the review of these studies by evidence of hypo-reactivity to infant cues, potentially implicating disruption to reward-related neural circuitry, and hyper-reactivity to infant cues, potentially implicating disruption to stress-related neural circuitry. Importantly, the studies reported here overlap with evidence of hyper- and hypo-reactivity to infant cues, highlighting the importance of considering multiple mechanistic pathways in understanding the impact of substance use on maternal brain and behavior.

Evidence of Hypo-Reactivity to Infant Cues

Hypo-reactivity to infant cues in substance-using mothers may be interpreted in neurobiological terms as these cues holding decreased salience or emotional / motivational value (11). In fMRI studies, this may be indicated by decreased responding to infant cues in substance-using mothers as compared to non-substance-using mothers; in ERP studies, this may be reflected in decreased amplitudes and / or delayed neural responses to infant cues in substance-using mothers. In the first fMRI study of substance-using mothers, participants viewed photographs of unknown happy, sad, and neutral infant faces and listened to unknown high- and low-distress infant cries (20). Substance-using mothers evidenced decreased responsivity to infant emotional faces across prefrontal cortical regions, including the ventromedial, dorsolateral, and dorsomedial prefrontal cortex (PFC), visual processing areas, and limbic regions including the hippocampus and amygdala. With respect to infant cries, substance-using mothers evidenced relatively decreased reactivity to high- and low-distress cries in the amygdala, frontal gyri, and insula. Taken together, these findings converge to support the hypothesis that infant cues, including faces and cries, may hold less salience to substance-using mothers.

Two ERP studies have examined the association between maternal substance use and unknown infant cue processing (happy, sad, and neutral faces; high- and low-distress cries). One study focused specifically on mothers who were either currently tobacco-smoking or non-smoking, with both groups of women being demographically-matched and comparable in their levels of depression and anxiety (21). The face-specific N170 component was delayed in tobacco-smoking mothers as compared to non-smoking mothers, consistent with the potentially decreased salience of infant face cues. There were no differences in ERPs elicited by high- or low-distress infant cries as a function of tobacco-smoking status.

The second ERP study employed the same unknown infant cues, examining ERP responses in demographically-matched substance-using and non-substance-using mothers (22). Again, N170 latency was implicated; at a neural level, N170 latency was modulated by emotional expression in non-substance-using mothers, being temporally earlier when elicited by neutral infant faces as compared to sad infant faces. This differential sensitivity to the emotional content of infant faces was absent in substance-using mothers, and this absence may reflect less sensitivity to encoding the infant emotional signals in the presence of substance use and addictions. Furthermore, relative to non-substance-using mothers, substance-using mothers also evidenced a delayed response to infant cries as indexed by the N100 component, an early marker of auditory processing, consistent with decreased salience of these stimuli early in auditory processing. Finally, while non-substance-using mothers evidenced a heightened attentional P300 response to high-distress as compared to low-distress infant cries, substance-using mothers evidenced a comparable P300 response to both cry types, suggesting that high-distress infant cries are no more salient than low-distress ones. Taken together, these two ERP studies converge in evidencing that maternal substance use is associated with a decreased and delayed neural response to otherwise salient infant facial and cry cues.

Advancement of this work has been to assess whether the familiarity of infant affective cues may generate differential neural responses among mothers with and without current substance use (23). The first study to examine this recruited a cohort of mothers in residential treatment (with their infants) for substance use disorders and examined neural responses to own and unknown happy and sad infant faces. The central finding of this work was that these mothers had a decreased response when viewing photographs of their own happy infant, as compared to an unknown happy infant, in the hypothalamus, ventral striatum, and ventromedial PFC, areas previously implicated in maternal brain networks and reward responsivity (7). Building on these findings by including a maternal control group, a second fMRI study employing own and unknown infant cues also evidenced hypo-responsivity to infant faces in reward-related neural regions (24). Specifically, substance-using mothers had a decreased response to sad infant faces, irrespective of familiarity, in the ventral striatum relative to non-substance-using mothers. The ventral striatum has been widely implicated in motivational and reward processes (e.g., 25, 26), is innervated by dopamine, and is thought to underscore approach behavior towards rewards (27). Unlike prior work employing unknown infant cries (i.e., 20), no differences in reactivity to infant cries were observed as a function of substance-use status in this latter study.

Taken together, these ERP and fMRI findings provide initial evidence of an attenuated response to infant affective cues in the presence of substance use and addictions.

Evidence of Hyper-Reactivity to Infant Cues

Critically, studies of maternal neural functioning and substance use have also demonstrated increased reactivity to infant cues. When exposed to unknown infant faces, substance-using mothers, as compared to non-substance-using mothers, evidenced increased reactivity to happy faces in the parahippocampal gyrus (20), a region that has previously been implicated in memory and emotion (28). Furthermore, tobacco-smoking mothers evidenced a greater P300 response to neutral infant faces relative to non-smoking mothers (21), with a similar P300 enhancement to neutral infant faces also observed in poly-substance-using mothers (22). Although infant neutral faces are typically employed as a potential baseline condition to valenced expressions (i.e., happy, sad), increasing attention has focused on neutral infant faces as holding a negative valence, particularly in the context of maternal anxiety, given the ambiguity of those faces (29, 30). This interpretation is consistent with a broader literature suggesting that neutral stimuli may be interpreted more negatively in the presence of anxiety (31–33). Thus, hyper-reactivity to neutral infant faces may reflect increased anxiety in maternal addictions, and this possibility should be examined in future research.

Studies employing own and unknown infant faces have also evidenced increased reactivity to own infant cues. In mothers in residential treatment for substance use disorders (23), increased reactivity to own, versus unknown, happy and sad faces occurred in numerous overlapping areas, including the cingulate, inferior frontal gyrus, thalamus, and cerebellum. Increased reactivity to own, versus unknown, happy faces was also observed in the amygdala, hippocampus, and thalamus in this maternal sample. Differential patterns of increased reactivity to infant cues has also been observed in substance-using mothers and non-substance-using mothers when viewing own and unknown infant faces (24). Specifically, substance-using mothers evidenced increased reactivity to own as compared to unknown infant faces, irrespective of emotional expression, with non-substance-using mothers instead evidencing increased reactivity to unknown as compared to own infant faces across multiple cortical regions, including the medial and superior frontal cortex, insula, temporal gyrus, inferior parietal lobule, and occipital cortex. Increased activity in overlapping cortical regions has also been observed when participants view pictures of emotionally arousing content (positively and negatively valenced (34)) – perhaps suggesting own infant faces hold greater affective intensity to substance-using mothers. Furthermore, increased activity in the insula has previously been implicated in mothers with insecure attachment style while viewing sad infant faces (35), and greater insular activity may reflect heightened pain or distress responses (36). Consequently, the increased insula response to own infant cues observed in substance-using mothers may be consistent with social distress, perhaps reflecting disruptions to mothers’ own early attachment experiences. However, future research is needed to examine this possibility.

In sum, increased neural reactivity to own and unknown infant affective cues has been observed in substance-using mothers (relative to non-substance-using mothers), suggesting that further research is needed to understand the implications of both decreased and increased neural responding to infant cues within the context of maternal substance use and addictions. These preliminary data indicate that neural responses to infant cues might be a clinical target that may help improve parenting approaches in mothers with substance use behaviors or disorders.

Parenting and Addictions: Next Steps for Human Neuroscience Research

The reward-stress dysregulation model of addictions and parenting has provided a unifying framework for neurobiological studies of maternal neural responses to infant cues. Here, we advance this model by emphasizing the role of both hypo- and hyper-reactivity to infant cues, implicating both reward- and stress-related neural circuitry, allowing a broader consideration of the mechanisms that may be compromised by maternal substance use and addictions. Key findings are presented in Figure 1.

Figure 1.

Summary of human neuroscience findings relating to maternal neural functioning and substance use and addictions organized by hypo-reactivity and hyper-reactivity to own and unknown infant cues (top panel) and by stimulus type and methodology (left panel). PFC - prefrontal cortex; VM, ventromedial, DM, dorsomedial, DL, dorsolateral, and IFG, inferior frontal gyrus.

Although convergent findings are emerging with respect to hyper- and hypo-reactivity to infant cues, a critical next step for this research is establishing how and whether neural responses to infant cues are associated with different aspects of maternal behavior. This is key to understanding the functioning of the maternal brain as related to addictive behaviors and the utility of findings to both our understanding of caregiving and the development of interventions. Prior research in non-substance using mothers has evidenced associations between neural responses to infant cues and maternal behavior (37); however, extension of this work specifically to maternal substance use is strongly needed. Only one study reviewed here considered any measurement of parenting, wherein patterns of maternal brain functional connectivity were associated with self-reported postpartum bonding (17). In advancing this area of maternal brain-behavioral links, it may be valuable to consider the nature of experimental stimuli employed in neuroimaging tasks and examine associations with behaviors, including assessing the ecological validity of employing only photographs or short cry recordings, and instead employing videos or extended cry recordings to be more consistent with parenting experiences outside of the research setting. Furthermore, moving beyond infant cue reactivity tasks to capture other critical facets of parenting, such as emotion regulation (38) and attachment (39), may also prove valuable.

To date, the research examining maternal neural responses to infant cues has focused on the categorical distinction of mothers being either substance-using or not, overlooking multiple factors that may also differentiate these maternal samples and potentially moderate neural responses to infant cues irrespective of substance use. Although some studies demographically match samples or control for demographic or other caregiving differences in statistical analyses, adopting a more dimensional approach may prove valuable to identifying at-risk non-substance-using mothers and allow for careful consideration of other variables that may be targeted clinically. Consistent with this notion, adopting a more developmental perspective in understanding pathways to motherhood and addictions could prove valuable (40). Although studies to date are cross-sectional and allow for understanding current associations between maternal substance use and neural responses to infant cues, converging measures of early experiences, including attachment security, adversity, and trauma, could prove insightful (35, 39, 41). An important question currently being addressed by our group is whether pre-existing conditions, such as disturbed attachment due to early life neglect or trauma, may predispose to both altered maternal brain responses and substance use and addictive behaviors. These approaches require larger maternal sample sizes to be appropriately powered to probe such dimensional and developmental questions, where the recruitment of such samples can be difficult given the continued stigma associated with substance use and maternal fears of child removal following disclosure of substance use.

Equally challenging is the nature of substances to which mothers are exposed and how substance use is operationalized across research studies. The vast majority of studies in Table 1 concentrate on poly-substance-using mothers, including varying illicit and legal substances, reflecting the typicality of maternal substance use not being limited to one specific substance. Indeed, poly-substance use in these studies referred not only to the sample of substance-using mothers generally, but also to individual mothers specifically – where utilization of more than one substance was often endorsed. Consequently, there has been little research addressing the impact of one specific type of substance on the maternal brain. Even in those studies focusing on tobacco-smoking (16, 42), mothers (in both the smoking and control groups) still reported some alcohol use. While substances may vary in their impact at neurochemical levels, it has been hypothesized that at a systems level of reward-stress dysregulation in the maintenance of addiction, varying substances may exert comparable impacts on the maternal brain (11). However, given different mechanisms of action of each substance, there may be differential effects of different substances. Likewise, differing patterns of early life trauma (abuse vs. neglect, for example) may lead to a preference for different substance groups (e.g., depressants, opioids, or stimulants) (39). The studies presented in Table 1 also vary in how substance use was operationalized, where approaches included self-report, diagnostic assessment, treatment program enrollment, and positive urine toxicology, which might add further variability to the reported findings.

Taken together, in human models, it is therefore difficult to parcellate the contribution of specific substances on the maternal brain and caregiving. Consequently, the identification of behaviors that can be observed across species would allow greater precision in the types and quantities of substance exposures and understanding their impact on maternal behavior would be valuable. Furthermore, studies examining substances that vary in their legality, and in ongoing changes in legality (e.g., cannabis use in the USA), may also be important to investigate given evolving patterns of use. Additionally, considering behavioral addictions (e.g., gambling and gaming disorders) may be helpful in potentially disentangling substance-related and addictive impacts (43, 44). In studies to date, there has also been less measurement of the stage in the addictive cycles at which data are being collected – including the length of time abstinent prior to data collection. More consideration is needed of the impact of abstinent and satiated states (and states in between these extremes) on maternal brain and behavior. There may not be a consistent neural response to infant cues, but instead more dynamic reactivity to infant cues reflecting physiological variability as function of abstinence and satiation (and perhaps other non-substance-related factors too).

While multiple neuroimaging studies incorporate mothers engaging with cues from their own infant presented as experimental stimuli, characteristics of the infant have not been fully taken into consideration. In utero exposure to substances may impact infant temperament, and more generally, variability across infant temperament could be an important missing piece to consider in understanding differences in maternal neural responses to infant cues. Similarly, given changes that take place in infant temperament across the initial months postpartum, variability in infants’ capacities to express emotions should be considered. In addition to a greater consideration of the developing child, there is a dearth of neuroimaging research on fathers generally (45), and no neuroimaging research, to our knowledge, specifically on substance-using fathers. Therefore, it is unknown whether the same hypo- and hyper-reactivity to infant cues observed in substance-using mothers may also be present for substance-using fathers. Additionally, as the field matures, replication of findings and use of advanced analytic approaches that consider networks and their segregation and interactions may help increase the rigor of existing studies.

Conclusions

A growing body of research has begun to document associations between maternal substance use and addictions and the neural correlates of infant cue processing. The complexity of these relations incorporate both increased and decreased reactivity to infant cues within and across studies in varying neural regions and circuitries. As this work continues, there are multiple directions for future research, including a strong need for multi-level/multi-modal measurement, longitudinal designs, and dimensional approaches. These additions will be critical to the identification of mechanisms that can be targeted by intervention programs in optimally supporting parents with substance use and addictions.