Parental Deprivation- and Threat-Based Factors Associated with Youth Emotion-Based Neurocircuitry and Externalizing Behavior: A Systematic Review

Externalizing Behaviors in Youth

Externalizing behaviors include aggression, defiance, hyperactivity, and impulsivity. These behaviors are most often observed among youth meeting Diagnostic and Statistical Manual of Mental Disorders, 5^th Edition (DSM 5) diagnostic criteria for attention-deficit/hyperactivity disorder (ADHD), conduct disorder (CD), and/or oppositional defiant disorder (ODD; Achenbach, 1966; Frick & Thorton, 2017). Externalizing behavior represents the most common cause of mental health referral among youth (Merikangas et al., 2011), with approximately 7.4% of youth in the United States receiving a diagnosis of an externalizing disorder (Ghandour et al., 2019). Across empirical and epidemiological studies, externalizing diagnoses are more common among male youth (King et al., 2018; Merikangas et al., 2009). Importantly, youth who exhibit externalizing behaviors are at elevated risk for delinquency, substance use, and violent behavior across development (Liu, 2004; Thompson et al., 2011). These outcomes are associated with high financial (e.g., legal fees, treatment) and societal (e.g., lower employment) costs (Foster et al., 2005; Knapp et al., 2011); thus, it is crucial to examine the pathways that lead to development of externalizing behaviors.

Parental Factors and Youth Externalizing Behaviors

It is well established that parental factors, including parenting behaviors (e.g., family conflict, low monitoring, hostility), substance use, and psychopathology (e.g., depression) are associated with externalizing behaviors in youth (Beyers et al., 2003; Georgiou & Symeou, 2018; McKee et al., 2008; Shaw et al., 1994; Zhang et al., 2020). Both parental externalizing and internalizing (e.g., anxiety, depression) diagnoses have been associated with youth externalizing behaviors (McKee et al., 2008; McLaughlin et al., 2012). Further, Patterson’s coercion theory posits that coercive parent-child cycles (initiated by parental hostility and maintained by negative reinforcement) lead to antisocial and delinquent behaviors in youth (Patterson, 2016). One potential mediator of these associations is youth’s ability to regulate emotions (Morris et al., 2007). For example, parental hostility may lead to child difficulties with emotion regulation and consequently higher levels of externalizing behaviors (Siffert & Schwarz, 2011). Extant literature reviews suggest that parental factors may also affect youth neural structure, function, and connectivity involved in youth emotional functioning (Belsky & De Haan, 2011; Kerr et al., 2019); however, little work to-date has examined such associations among youth with externalizing behaviors, specifically.

Conceptual Neurodevelopmental Pathway Models

Existing models of parental factors and youth emotion-linked neurodevelopment include the dimensional model and the stress acceleration model (McLaughlin & Sheridan, 2016; Callaghan & Tottenham, 2016). The dimensional model presents two separate dimensions of adverse parental influences: deprivation and threat. According to the dimensional model, deprivation is associated with alterations in the frontoparietal network (e.g., reduced cortical thickness and volume), which has been linked to executive functioning and attentional processes. In contrast, threat is associated with alterations in the frontolimbic network (e.g., reduced volume, elevated amygdala activation to threat) associated with emotion regulation.

A recent systematic review on deprivation (e.g., institutional rearing, neglect) revealed accelerated cortical thinning in the frontoparietal network (Colich et al., 2020). Additionally, adolescents with parental history of major depression (which may be characterized by disengagement or withdrawal; England et al., 2009) demonstrated lower dorsolateral PFC (dlPFC) activation to fearful faces compared to controls (Mannie et al., 2011), consistent with deprivation-related alterations in frontoparietal functioning. With respect to threat, early threatening parenting behaviors (e.g., maternal hostility, intrusiveness, negative affect) have been associated with reduced youth hippocampal subregion volumes (Blankenship et al., 2019), accelerated cortical thinning in the ventromedial prefrontal cortex (vmPFC; Colich et al., 2020), and increased amygdala activity while viewing angry and fearful facial stimuli (Pozzi et al., 2020), consistent with alterations in frontolimbic development (McLaughlin et al., 2019). Further, infants exposed to prenatal maternal depression (which, in addition to withdrawal, may be characterized by hostility; England et al., 2009) showed reduced connectivity between the amygdala and prefrontal regions (Posner et al., 2016). Interestingly, youth with family history of alcoholism (which may be characterized by abuse or neglect; Semidei et al., 2001) demonstrated lower amygdala volume and lower frontoparietal connectivity (Cservenka, 2016), consistent with both deprivation- and threat-related alterations. Thus, of note, while specific parenting behaviors may be easily characterized as deprivation or threat contexts, parents with depression and/or parents who engage in substance use may exhibit behaviors that can be categorized as both deprivation and threat (i.e., withdrawal, hostility; England et al., 2009; Semidei et al., 2001), accounting for the brain alterations consistent with predictions accounted for by the dimensional model.

In contrast to the dimensional model, the stress-acceleration model posits that exposure to adversity, primarily parental deprivation, may lead to rapid development of frontolimbic regions. For example, early maternal deprivation has shown associations with accelerated youth frontolimbic maturation (Gee et al., 2013; Silvers et al., 2016), consistent with this model (Callaghan & Tottenham, 2016). Similarly, infants exposed to prenatal maternal depression showed evidence of higher amygdala-vmPFC connectivity (Qiu et al., 2015). In terms of threat, exposure to maternal hostility during the preschool years has shown associations with negative amygdala connectivity with frontal areas during facial emotion tasks (e.g., sad expressions) in late childhood, consistent with patterns seen in adulthood (Kopala-Sibley, 2020). Similarly, youth with family history of alcoholism showed decreased amygdala activation (Cservenka, 2016). Thus, in some cases, both deprivation- and threat-related contexts have been associated with accelerated frontolimbic development, consistent with the stress-acceleration model.

A prior review on childhood adversity and neurodevelopment found evidence consistent with the dimensional model (i.e., frontoparietal alterations with deprivation, frontolimbic alterations with threat), but inconsistent evidence (i.e., support from only half of the studies) for the stress-acceleration model (McLaughlin et al., 2019). While there was stronger evidence for accelerated development of neural structure, the findings on amygdala-mPFC connectivity were mixed, possibly due to variability in sample sizes and the nature of the specific tasks completed. Notably, a more recent meta-analysis and systematic review did not find consistent associations between early life adversity involving threat or deprivation and amygdala-PFC connectivity (Colich et al., 2020). Of importance, contextual parenting experiences are not deterministic (Thompson, 2016), prompting the need for more work to identify which youth are more prone to frontolimbic and frontoparietal disruption, so that interventions and prevention efforts may be more targeted. Thus, the current review will aim to identify patterns in brain structure and function that occur across studies and highlight any divergences which may be related to diversity of samples or tasks.

Youth Emotional Neurocircuitry and Externalizing Behaviors

Both structural and functional brain differences within emotion-based circuits, including those implicated in both the dimensional model and stress acceleration model, have been observed among youth exhibiting externalizing behaviors. For example, externalizing behaviors in typically developing youth, even at subclinical levels, have been associated with reduced amygdala and prefrontal cortex (PFC) volume (Caldwell et al., 2015; Montigny et al., 2013) and reduced hippocampal volume (Bos et al., 2018). In terms of functional activation, externalizing behavior in youth has been associated with hypoactivity in the medial PFC (mPFC) prior to risky decision-making and in the insula during facial emotion processing (e.g., happy expressions) among female youth (Crowley et al., 2017; Whittle et al., 2015).

With respect to functional connectivity, youth have shown lower white matter integrity in a connective tract between limbic and frontal regions (Andre et al., 2020a) and specifically, male youth with externalizing diagnoses have shown decreased connectivity between frontal and limbic regions (Shannon et al., 2009). Further, youth in middle childhood exhibiting higher levels of aggressive behavior have shown higher global functional connectivity within the amygdala and reduced frontolimbic connectivity at rest in comparison to controls (Sukhodolsky et al., 2021). Additionally, Ameis and colleagues (2014) found stronger coupling between the amygdala and vmPFC in adolescents exhibiting lower, but not higher, externalizing behaviors. In sum, the literature supports the notion that youth with externalizing behaviors exhibit atypical structural (i.e., reduced volumes and white matter) and functional (i.e., weaker frontolimbic connectivity) development in areas linked to emotional functioning and areas implicated in both the dimensional model and stress acceleration model. While these patterns appear to be better accounted for by the dimensional model, it will be important to examine the pathways to these youth externalizing behaviors from parental factors characterized by deprivation and threat, as this may account for heterogeneity in neurocircuitry among youth.

Current Review

Prior work suggests divergent neurodevelopmental pathways exist based upon adverse parental influences characterized by deprivation and/or threat; however, there have been varied findings due to inconsistencies in samples and tasks. Additionally, studies to date have not yet comprehensively examined how these neurodevelopmental pathways impact youth externalizing behavioral outcomes, despite work that has shown alterations in neurodevelopment (i.e., within the frontolimbic network) that may lead to youth psychopathology (Jones et al., 2017). The current review will build upon prior work by examining how parental factors characterized by deprivation and/or threat contexts are associated with youth neural structure and function (e.g. activation and connectivity) and consequently youth externalizing behaviors. We will present a conceptual model whereby youth neurocircuitry serves as a mediator between parental factors and externalizing behaviors (see Figure 1), which builds on the dimensional model and stress acceleration model (McLaughlin & Sheridan, 2016; Callaghan & Tottenham, 2016). Thus, this review will serve to improve the understanding of heterogeneity in youth neurocircuitry, as well as the etiology of externalizing behaviors, through the application and extension of preexisting models based on deprivation vs. threat contexts. An earlier review by Belsky and de Haan (2011) on parenting and youth neurodevelopment noted that this area of study was “not even yet in its infancy”; thus, the current review aims to synthesize the still limited, but emergent, work in this area. Based on prior studies, we expect to find evidence supporting both the dimensional model and the stress acceleration model. Further, as shown in the proposed model, we hypothesize that deprivation- and threat-related parenting factors will be associated with elevated externalizing behavior through altered neural structure and function (whether indicative of accelerated or delayed development).

• Download figure
• Open in new tab

Figure 1.

Conceptual Mediation Model of Neurodevelopment.

Methods

Search Procedures

For the current review, PRISMA reporting guidelines for a systematic review were used (Moher et al., 2009; see Figure 2). Two researchers (initials masked for review) completed two independent database searches and titles and abstracts were screened to determine eligibility (92.3% agreement). A third reviewer (initials masked for review) reconciled disagreements (two articles were confirmed for inclusion by consensus in this manner). The following search terms were utilized in PubMed and PsycInfo: (1) “((externalizing) AND parent*) AND ((fMRI) OR (functional MRI) OR (functional magnetic resonance imaging))”, (2) “((externalizing) AND parent*) AND ((EEG) OR (electroencephalogra*))”, (3) “((externalizing) AND parent*) AND emotion* AND neuro*”, and (4) “((externalizing) AND parent*) AND emotion* AND neural”.

• Download figure
• Open in new tab

Figure 2.

PRISMA Flowchart for identification and eligibility of articles. Template provided by (Moher et al., 2009).

Results

Out of the 291 total articles identified, which were published between 1987 and 2021 (excluding overlap across databases; 360 total articles with 69 duplicates), 26 articles were selected following the initial screening of titles and abstracts. Of the 265 articles that were deemed not eligible, 197 did not include all three constructs (parenting, youth neurocircuitry, and youth externalizing behavior), 43 met criteria for a medical/psychiatric exclusion (e.g., autism), 13 were genetic studies, nine were not independent empirical studies (e.g., meta-analysis, review), and three studies involved participants outside the target age range. After a full article review for inclusion, two articles were excluded due to the lack of a parent-specific factor (e.g., family history rather than parental history) and two articles were excluded due to lack of youth neuroimaging results. One additional article was found via reference tracing and was included due to meeting inclusion criteria for this review. Overall, 23 articles met inclusion criteria and were organized by three domains: parenting behavior (n=5), parental depression (n=6), and parental substance use (n=12). See Figure 2 for a detailed outline of inclusion for the reviewed studies in accordance with PRISMA guidelines and Table 1 for detailed study information (e.g., brief demographics, methods, results).

Discussion

Across healthy and high-risk (e.g., history of maltreatment) samples, parental factors were associated with youth behavioral outcomes in the expected directions (See Table 1 for study results). The model presented within this review (see Figure 1) expanded upon pre-existing models by distinguishing two pathways by which threat- and deprivation-related factors may result in elevated youth externalizing behaviors through altered functioning in frontolimbic and frontoparietal networks. As described previously, the dimensional model proposed deficient neural development, while the stress acceleration model proposed accelerated neural development due to parental factors.

Across nine studies, there was support for the deprivation-based dimensional model with disruption in frontoparietal regions (i.e., lower activation, reduced cortical pruning) due to poor monitoring and supervision, maternal depression, as well as parental substance use (Ashman et al., 2008; Barbosa et al., 2018; Carlson & Iacono, 2008; Dawson et al., 2003; Ehlers et al., 2001; Euser et al., 2013; Heitzeg et al., 2014; Holla et al., 2019; Kamarajan et al., 2015). When examining maternal stimulation (i.e., low deprivation; Holz et al., 2018), a mediation model was supported such that higher maternal stimulation led to lower ADHD symptoms through reduced caudate head activity. This suggests evidence for protective factors within the mediated pathway, although it is limited to one study. With respect to parental depression, Dawson and colleagues (2003) conducted the only study to report parental behaviors related to depression within the sample and noted higher levels of withdrawn behavior consistent with deprivation. While parents with depression may engage in hostile or withdrawn parenting or both (NRC and IM, 2009), parental neglect may be a more salient mediator than abuse in the association between parental depression and youth outcomes (Mustillo et al., 2011). The mediation model presented in this review was supported during toddlerhood by one study assessing for parental depression (Dawson et al., 2003), but was not supported during school age by a study assessing for poor monitoring and supervision (Barbosa et al., 2018) or a different study assessing for parental depression (Ashman et al., 2008). Studies assessing for parental substance use did not specifically test for mediation but noted significant associations between lower brain activation or reduced cortical pruning and externalizing behavior (Carlson & Iacono, 2008; Ehlers et al., 2001; Euser et al., 2013; Heitzeg et al., 2014; Holla et al., 2019); thus, these implied mediation pathways should be more closely examined in future studies.

The threat-based dimensional model was supported across nine reviewed studies, such that corporal punishment, negative parenting behaviors, maternal depressive symptoms, and parental history of substance use led to disruptions (i.e., smaller volumes, lower frontal and higher limbic activation) in frontolimbic regions (Andre et al., 2020b; Barbosa et al., 2018; Chaplin et al., 2019; Heitzeg et al., 2008; Heitzeg et al., 2010; Heitzeg et al., 2014; Qu et al., 2016; Venkatasubramanian et al., 2007; Wang et al., 2020). The proposed mediation model was not supported in studies assessing for corporal punishment, negative parenting, and parental substance use. However, it was supported in a study examining the association between parental depressive and adolescent externalizing symptoms through prolonged ventral striatum activation (Qu et al., 2016), consistent with the mediational threat pathway. Notably, this study recruited a community sample, suggesting that even mild to moderate levels of depressive symptoms may contribute to abnormal limbic activation and externalizing behavior. While this mediation model fell more in line with the threat-based pathway, data collection of depressive symptoms and diagnoses did not allow for parsing whether specific symptoms observed in the sample were more threat- or deprivation-based. Further, disrupted frontolimbic structure and function was associated with externalizing behaviors across studies examining a threat context, consistent with the proposed mediational threat pathway; however, full mediation was not assessed.

With respect to stress acceleration, six studies on negative parenting behaviors, maternal depressive symptoms, and parental substance were in line with this model (Chaplin et al., 2019; Heitzeg et al., 2008; Heitzeg et al., 2010; Hulvershorn et al., 2013; Kwon et al., 2021; Wang et al., 2020). In the study examining parenting behavior, a mediation model was not conducted, as there were no significant associations between brain function and externalizing behavior. While the studies examining parental depressive symptoms and parental substance use did not assess for mediation, two of the five studies demonstrated significant associations between frontolimbic activation/connectivity and externalizing behavior (Heitzeg et al., 2010; Wang et al., 2020).

Interestingly, some results were not accounted for within the models examined in this review. For example, disruption in frontoparietal regions, rather than in frontolimbic regions, was noted for studies examining threat-based parenting behaviors (i.e., corporal punishment, physical maltreatment; Barbosa et al., 2018, Shackman & Pollak, 2014). However, the proposed mediation model was not supported in these studies.

It was noted that neurodevelopmental patterns differed across studies examining the same or similar parental constructs. For example, in some cases biological sex may moderate these associations, as threat-related parenting behaviors were associated with altered frontoparietal activation in females, while threat-related parenting supported accelerated frontolimbic development among males. More longitudinal work is needed to assess whether this accelerated frontolimbic development in males is adaptive for youth, as early development in a given region may result in later structural underdevelopment (e.g., smaller volume; Deoni et al., 2016; Shaw et al., 2006). Similarly, there was sex-based variation observed in early childhood resting-state functional connectivity among youth whose mothers endorsed postnatal maternal depressive symptoms. Females exhibited altered frontolimbic functioning (i.e., difficulties with emotion processing and regulation), more consistent with the dimensional threat model, whereas males exhibited altered frontoparietal functioning (i.e., difficulties with attention to emotional stimuli), in line with the dimensional deprivation model. Thus, more studies examining biological sex as a moderator are needed to better understand how brain structure and function is altered in each sex.

Regarding findings on frontal brain volume, Demers and colleagues (2019) did not find strong evidence of accelerated brain development due to maltreatment history (i.e., threat) as expected based on prior work on the stress acceleration model (Callaghan & Tottenham, 2016). However, there was a marginal association identified between maltreatment history and higher frontal lobe volume for those who experienced only one type of maltreatment, as opposed to multiple forms of maltreatment. Thus, accelerated development of frontal areas may be dependent upon the number of and types of maltreatment exposure. It is important to note that incidents of both abuse and neglect, inclusive of but not restricted to a parent, were captured under maltreatment in this study; therefore, it is not clear whether this marginal effect was related to deprivation- or threat-related parental factors. Further, neuroimaging for this study was conducted in adulthood, at which point there may be developmentally appropriate reductions in frontal volume (Peters, 2006), limiting our knowledge on whether significant effects would have been observed earlier in adolescence.

In some studies, youth who had parents with substance use disorders demonstrated evidence of accelerated or typical frontolimbic development via higher frontal activation and lower limbic activation, consistent with the stress acceleration model. Prior work has shown that having a parent who engages in substance misuse can be considered a traumatic event (Parolin et al., 2016), potentially providing a similar maltreatment (i.e., threat) context as those youth demonstrating accelerated development of frontolimbic neurocircuitry. In contrast, two studies that included youth from middle childhood to young adulthood found smaller volumes in frontal regions and another study found lower frontal brain activation (i.e., delays in neurodevelopment), consistent with the dimensional model. In the latter study, the youth were preadolescents and thus may not yet have gone through normative PFC development. Given the wide age range, it is difficult to discern whether these findings are related to normative development prior to adolescence or suggest heterogeneity in structural brain outcomes of youth with parents who have history of substance use. Further, one study examined prenatal alcohol exposure, which may have different implications for brain development than a history of substance use. Notably, it is not possible to discern whether parents who used substances were more withdrawn or hostile in the reviewed studies. Kepple (2018) found that past year parental substance use disorder (SUD) diagnosis was associated with neglect, while past year substance use (i.e., not necessarily meeting criteria for SUD) was associated with abuse, suggesting differential pathways between substance use and parental deprivation or threat. Thus, heterogeneity in neural functioning may provide support for divergent pathways of neurodevelopment for youth of parents who have substance use difficulties. It is imperative for future work to further break down the assessment of parental psychopathology, examine parent-child associations longitudinally, and take pubertal factors into account in order to parse normative and non-normative development as well as resultant behavioral outcomes.

In sum, most studies demonstrated support for the dimensional model, while a more limited number provided support for the stress acceleration model, when considering both parental threat- and deprivation-contexts. Although parental behaviors and psychopathology consistently predicted youth externalizing behaviors in the expected directions, there was only limited overall support for the proposed mediational model, given that many of the reviewed studies did not explicitly test for mediation. Further, several findings fell outside of the proposed model (i.e., frontoparietal disruption in a threat context) and may suggest heterogeneity within parental factors.

Limitations

These findings may be limited as the existing literature primarily relied on community samples, which may have reduced the ability to identify significant brain-behavior associations due to low levels of negative parenting behaviors and youth symptoms. As such, examination of clinical samples will be important in future studies to identify the most at-risk youth. Additionally, conclusions were limited in the thorough examination of brain-behavior associations across parents and youth, as well as the examination of biological sex as a potential moderator, as some reviewed studies solely included either parents or children of one biological sex. Additionally, it was difficult to draw concrete conclusions about whether neural functioning was normative or delayed for a given sample, as not all reviewed studies considered hormonal development. Further current study limitations involved the reliance of most reviewed studies on a single assessment method for parental factors (e.g., observational vs. self-report, categorical diagnoses vs. continuous symptoms) and much variation across the task methodologies across studies (e.g., use of facial expressions versus emotional valence pictures) and neuroimaging (e.g., EEG, rs-fMRI, task-based fMRI). For example, reviewed studies may have produced divergent findings as a result of varied task demands or reporter biases rather than true differences related to parental deprivation or threat. Lastly, it was not possible to thoroughly assess changes in brain development and behavior within this study, as many of the reviewed studies were cross-sectional or were underpowered to detect a significant mediation effect due to small sample size.

Concluding Remarks

Overall, this review demonstrated that parental factors characterized by deprivation and threat are differentially associated with frontolimbic and frontoparietal structure and function in youth with and without externalizing behaviors. There was strong evidence to support the dimensional model, as well as some evidence to support the stress acceleration model with respect to the deprivation and threat context. Several studies examining parental depressive symptoms supported the newly proposed mediation model; the other reviewed studies either did not support this model or did not assess for mediation. More work is needed to fully conceptualize deprivation-versus threat-based factors among parents who experience depressive symptoms and have a history of substance use. Building from extant models of neurodevelopment, there was evidence of disruption in both frontolimbic and frontoparietal networks across deprivation- and threat-based factors, rather than disruption within just one network. More specifically, findings of this review extended beyond existing models to include frontoparietal alterations with respect to threat and frontolimbic alterations with respect to deprivation. Further, variations by biological sex and pubertal factors played a role in neural functioning. This review partially supported prior notions that youth exposed to threat-related and deprivation-related parental factors may demonstrate earlier frontolimbic maturation. In contrast to the stress acceleration model, some youth exposed to deprivation-related factors showed delayed maturation beyond developmentally normative periods. By calling attention to these differences, we can better predict risk and resilience trajectories of emotional development in youth. Additionally, this knowledge may better inform individualized targets (e.g., parenting, youth emotion regulation skills) in universal prevention and clinical intervention efforts. It will be especially important to involve parents who endorse their own maladaptive parenting behaviors and/or psychopathology in the treatment of youth. Further, studies should follow up on whether treatments that specifically target youth emotion regulation difficulties lead to changes in brain development (e.g., maturation of frontolimbic and frontoparietal pathways) and reductions in behavioral difficulties in youth. For example, dialectical behavior therapy, which specifically targets emotion dysregulation for both parents and children, has shown promise among adolescents and adults with externalizing disorders (Abootorabi Kashani et al., 2020; Bayat et al., 2020; Meyer et al., 2020). Overall, the current review provides a synthesis of the extant literature on parental factors and youth neural functioning and behavior, as well as a model from which to examine forthcoming results.

Competing Interests

The authors declare no competing interests.

Author Contributions

KEF conceived and designed the project. KEF, RPL and MMH conducted literature search procedures. KEF wrote the paper and all authors contributed to the revisions and approved the final version.