2.1 Data sources and search strategy

This scoping review centered on the research question, “Is variability in normative range parenting associated with neural function and structure in human offspring?” All review processes were conducted in accordance with the PRISMA extension for scoping reviews (PRISMA-ScR; Tricco et al., 2018). The initial search was conducted on February 5, 2019 in three electronic databases: MEDLINE/PubMed, Scopus, and PsycInfo. The search was tailored to the specific requirements of each database. The search query consisted of three primary sets of terms: (a) parenting: parent*, caregiv*, maternal, paternal, family, familial; (b) neuroimaging: fMRI, MRI, dMRI, DTI; and (c) psychological relevance: psycholog*, psychiatr*, psychopathology. The search query also consisted of terms preceded by the Boolean operator “NOT” in order to exclude large numbers of studies deemed irrelevant and/or outside the scope of the present study. A complete list of search terms and syntax is provided in the supplemental materials (Table S1). The search query included only peer-reviewed empirical articles, published in English, in samples of human subjects. Citations from all three databases were imported into bibliographic manager Zotero (version 5.0.60), wherein duplicate citations were manually removed and merged.

2.2 Data charting and screening

Data selection, screening, full-text review, and synthesis were conducted by a single reviewer (MJF) who was not blind to record author or journal name throughout the reviewing process. The first level of screening involved the examination of articles for relevance and inclusion criteria by title and abstract only. Articles for which no abstracts were available were included for further screening in the full-text review. During the full-text review, each article was read in its entirety to (a) further screen for relevance and inclusion criteria, and (b) extract relevant data for subsequent qualitative analysis and synthesis. Reasons for exclusion during all stages of data charting and screening are detailed in Table 1.

2.3 Data characterization and synthesis

Citations were imported into Microsoft Excel (version 16.22) for citation management, data organization, and qualitative and descriptive statistical analyses. Extraction of data pertinent to the research question was conducted throughout the full-text review and was recorded in Microsoft Excel.

3.1 Search and selection

The original search yielded 1,411 potentially relevant records. 814 records remained after the removal of duplicates. Of these 814 records, 40 met inclusion criteria after screening by title and abstract. A large number of citations were excluded upon the title and abstract screening as several terms used in the search algorithm also corresponded to other study parameters; a complete list of topics deemed irrelevant and/or beyond the scope of the present review can be found in Table 1. These 40 records were then procured for full-text review. After data characterization, 23 articles were included in the final analysis. The flow of records from identification through final inclusion is depicted in Figure 1.

3.2 Characteristics of included records

A summary of general characteristics, parenting assessment methodologies, and neuroimaging assessment methodologies of the 23 included records is reported in Tables S2–S4, respectively. The 23 included records reported on 19 independent data sets in total. The timeline of data collection for each independent data set is illustrated in Figure 2. The number of assessments, age of child at assessment, and age range of sample at assessment varied widely. Eleven included records also reported on child behavioral or clinical outcome; however, because not all records included such outcome data, these non-neuroimaging outcomes are not discussed in the present review. Further details of each individual study can be found in Tables 2–4.

3.3 Functional activity (Table 2)

Ten included records examined parenting in relation to task-related functional activity as indexed by blood oxygen level-dependent (BOLD) signal change (Coan et al., 2013; Farber et al., 2018; Gollier-Briant et al., 2016; Kim-Spoon et al., 2017; Lauharatanahirun et al., 2018; McCormick et al., 2016; Qu et al., 2016; Romund et al., 2016; Taylor et al., 2006; Telzer et al., 2013). Although varying in exact task design, these studies primarily investigated three constructs: (a) behavioral control/inhibition, (b) goal-directed behavior, and (c) threat responsivity. Inconsistent associations were observed between parenting and brain activity associated with responding to threat. For example, increased parental warmth or similar constructs were associated with relatively decreased (Romund et al., 2016) and relatively increased (Farber et al., 2018) amygdala activity. This could stem from differences in task, age range, and precise parenting construct assessed (i.e., maternal warmth and support vs. family affective responsiveness). Regardless, these studies suggest altered activity within a corticolimbic circuit including the amygdala and prefrontal cortex as well as the insula, which supports interoception. Studies exploring parenting in relation to goal-directed behavior reported that increased familial obligation and positive parenting are associated with decreased activity in the ventral striatum (Qu et al., 2016; Telzer et al., 2013), which plays a critical role in associative learning and the translation of motivation into action, while increased parental knowledge and decreased household chaos are related to increased insula activity (Lauharatanahirun et al., 2018). Although findings related to the ventral striatum and insula are less mixed than those related to the amygdala, it is worth noting that these studies varied in task and age range as well. Last, increased prefrontal activity associated with cognitive control and response inhibition was consistently associated with greater control-related parenting constructs (e.g., parental control, family obligation, household stability; Kim-Spoon et al., 2017; McCormick et al., 2016; Telzer et al., 2013). Collectively, these studies of task-based activation reveal mixed findings between normative range parenting and threat responsivity, relatively consistent findings with goal-directed behavior, and consistent findings with behavioral control/inhibition.

3.4 Functional connectivity (Table 3)

Only one included record examined how functional connectivity between brain regions was modulated by an experimental task (Gee et al., 2014), and three included records examined parenting in relation to intrinsic functional connectivity (Brauer et al., 2016; Degeilh et al., 2018; Graham et al., 2015), which assesses the strength of correlated activity between brain regions when participants are “at rest” or otherwise not engaged by an experimental task. Although small in number, these studies converge to suggest positive associations between maternal touch frequency and within-network patterns of intrinsic connectivity (Brauer et al., 2016) and negative associations between maternal caregiving quality and parental conflict and between-network patterns of intrinsic connectivity (Degeilh et al., 2018; Graham et al., 2015). It is possible that such associations are indicative of stronger communication between regions within networks supporting specific processes (e.g., cognitive control) but less efficient modulation in the ability of different networks to selectively engage or disengage as needed (e.g., salience detection and cognitive control). As with studies of functional activation, such possibilities are speculative given the vast range in study methodology. Moreover, the varied age ranges at the time of scanning further limit the ability to draw clear conclusions from these findings.

3.5 Brain structure (Table 4)

Eight included records explored parenting in relation to volumetric estimates of brain structure (Ganella et al., 2015; Kok et al., 2018; Luby et al., 2012; Matsudaira et al., 2016; Sethna et al., 2017; Whittle et al., 2009, 2014, 2016), and four included records investigated parenting in relation to cortical thickness (Avants et al., 2015; Kok et al., 2018; Whittle et al., 2014, p. 201; Whittle et al., 2016). No studies utilizing DTI to assess white matter structural integrity met full inclusion criteria. With regard to volumetric measures, some studies reported positive associations between higher parental care and brain volume, while others reported negative or non-significant associations. As with the functional data, it is difficult to determine the extent to which these inconsistencies are due to the broad age range across records (3 months–19 years) and/or different methods of indexing parenting and brain volumes. The associations between parenting and cortical thickness appear similarly mixed. Some studies reported a positive association between higher parental care and cortical thickness, while others reported that care is associated with greater thinning of the cortex over time. Again, varied methodologies across a small number of studies obscure the nature of these inconsistencies.

4.1 Limitations of existing research

What is striking from this review is not only how few studies have explored associations between the brain and normative range parenting, but also how little methodological consistency exists across these studies. With regard to parenting, studies varied in: (a) number of assessments, (b) age at assessment(s), (c) parent(s) assessed, (d) method of assessment, and (e) construct assessed. With regard to the brain, studies varied in: (a) number of assessments, (b) age at assessment(s), (c) type of neuroimaging acquired, and (d) approach to data analysis. Studies also varied in sample size and covariates. Methodological diversity is not necessarily a limitation—indeed, parenting is incredibly complex and should be studied in all of its multifaceted complexity. However, it is worth noting that such methodological diversity within an already limited body of existing literature both makes attempting replication difficult and helps explain the absence of consistent findings at this time. As Marek et al. (2020) noted in their recent preprint advocating for very large samples to identify true brain-behavior associations, it will remain difficult to make clinically relevant progress unless we can combine across studies, which necessitates a finite set of measures. We next consider some specific limitations before outlining a potential path forward.

The first and perhaps most pressing barrier to a thorough investigation of this research question is feasibility, which we address in detail below in making recommendations for future research. Briefly, an ideal study design by which to assess the impact of normative range parenting on brain development would start with a population-representative birth cohort that is followed longitudinally over the lifespan with multiple data collection timepoints including neuroimaging data; structured clinical interviews of parent(s) and offspring; observed parent–child interactions; in-home interviews; child- and parent-reported parenting data; and genetic data from all family members. Suffice it to say that such a data set is not and will not be easy to come by. We certainly do not fault any authors of the included records for their research designs, which already are resource-intensive.

Second, publication bias may limit the available literature. It is theoretically possible that brain development is only impacted by caregiving at its most extremes, leading to the publication of only those studies examining the developing brain in relation to more severe variations in parenting such as trauma, abuse, or neglect. This would imply that the developing brain is robust to the influence of caregiving except in its most extreme forms, a possibility that seems unlikely given a vast literature on broader environmental influences on brain development (Belsky & Haan, 2011; Belsky & Pluess, 2009; Cicchetti, 2015; Tottenham, 2018) and the notable influence of normative range parenting on socioemotional behavior and psychopathology (Ainsworth, 1969; Baumrind, 1967; Bowlby, 1958; Darling, 1999; Maccoby & Martin, 1983; Power, 2013). We appreciate the substantial barriers to publication, particularly in the case of null findings. Nevertheless, recent trends in the field toward open science and preprint publications make the present an apt time to revisit previous projects and make them publicly available in the service of advancing our collective understanding of this to date relatively unexplored question. To this end, we hope that researchers having measures of parenting and neurodevelopment will consider revisiting their data and publishing even null findings.

Relatedly, neuroimaging standards of practice are constantly in motion. From a technological and methodological standpoint, the field has advanced even in the time since many of the included records were published. While neuroimaging studies of 10–30 subjects were once accepted as the norm, issues of low statistical power and high false-positive rates have resulted in concerted efforts to increase sample sizes and attempt replication whenever possible (Lieberman & Cunningham, 2009; Vul et al., 2009; Yarkoni, 2009). Better acquisition protocols (e.g., multiband) and preprocessing strategies (e.g., motion scrubbing) than those employed by many of the records included in this review are now available. Consistent findings across the available literature, albeit comprised of a small number of studies, may be lacking because of such methodological limitations.

Last, while we aimed to be comprehensive in our review of the literature, practical considerations required that we disregard a number of variables that could inform the research question of interest in meaningful ways. For example, we did not address the potential influence of genetics in the present review despite known genetic influences on both brain and behavior (Plomin et al., 2008; Rakic, 1988). Similarly, we did not explore the potentially bidirectional nature of parenting and the possibility that child temperament and parent temperament interact to influence parent behavior, which then influences child brain development. We also did not examine interactions between parenting and socioeconomic status, which is known to impact parenting (Roubinov & Boyce, 2017). Additionally, we did not explicitly investigate differences by race and ethnicity, which will be especially important for future work to explore given the known differences in parenting among families of varied racial and cultural backgrounds (Berger et al., 2005; Julian et al., 1994; Robinson & Harris, 2013). Genetic predisposition, child temperament, socioeconomic status, culture, race, and ethnicity, and family structure may moderate the relationship between parenting and the brain. Given the limited number of studies otherwise identified, however, we think it is premature to integrate such variables in examining links between normative parenting and the brain.

Finally, it is important to note that all stages of data screening and analysis were conducted by a single reviewer (MJF), which naturally leaves results vulnerable to bias. The sole reviewer's methods were rigorously scrutinized by both co-authors as well as other collaborators (see acknowledgments). Moreover, following the PRISMA guidelines for a scoping review further limits potential bias by providing discrete search parameters. Nevertheless, these findings reported herein would benefit from independent evaluation—ideally with the inclusion of new articles having been published since this review was conducted.

4.2 Implications

Despite these limitations, the relationship between normative range parenting and brain development remains a critically important area of study. It has long been demonstrated that the first two decades of life are vital in determining how the brain will develop to support behavior. Moreover, the extent of neural plasticity and associated capacity for change during these early years is unique to the lifespan; the brain is more susceptible to environmental factors—for better or for worse—during the first two decades of life than any other time period. Importantly, brain development is hierarchical and thus even small perturbations early in life can have broad effects later through a cascade of interrelated processes (Casey et al., 2019; Thelen, 2005; Tottenham, 2018). Through a better understanding of how normative range parenting is associated with changes in brain structure and function, we can leverage this knowledge to promote positive changes during periods in which the developing brain is most open to environmental influence. That is, strengthening parenting in the first two decades of life holds the potential to create lifelong neural scaffolding for resilience.

The relationship between parenting and brain development has clinical implications. Mental illness across the lifespan is embedded within a broader family context. The salience of family context may be most readily apparent when treating child and adolescent patients, but the same is true of adult patients as well. A preschool-aged girl with anxiety requires active work with parents to extinguish learned fears and safety concerns. A high school-aged young man with anhedonic depression requires parent support to aid in successful behavioral activation outside of therapy. A middle-aged person with volatile romantic relationships requires exploration and understanding of how their upbringing contributed to their conceptualization of partnership. Regardless of a patient's age, clinical interventions nearly always involve the examination of patterns learned in early life within the home environment and among caregivers. Thus, deepening our understanding of how various parenting styles and behaviors impact biobehavioral outcomes is essential to optimizing mental health treatments.

The American Psychological Association (APA) last provided guidelines for parenting practices in 2012, encouraging parents to attend to and praise good behavior; ignore minor misbehavior and prevent misbehavior when possible; educate themselves on normative developmental milestones; implement consequences that are brief, calm, and immediate; teach children how to cope with demands; and take time for themselves (Novotney, 2012). While most of these guidelines remain true, they can be further refined and advanced by identifying trajectories of brain development associated with variability in normative range parenting that provide a foundation for understanding developmental biomarkers of risk and resilience, which can then be translated to work with patients and parents of patients across clinical contexts.

Such a foundation has implications at the policy level as well. An ongoing cultural narrative surrounds the notion of “helicopter parenting,” in which parents pay extremely close attention to their child's experiences and problems, often intervening before their child has an opportunity to handle the challenge independently. There is an increasing debate over the extent to which parental protection is beneficial to child development, and the point at which involvement veers into over-involvement. Indeed, some evidence suggests that overprotection can have negative consequences such as increased rates of depression and anxiety and decreased life satisfaction and general psychological well-being for children in certain circumstances (LeMoyne & Buchanan, 2011; Schiffrin et al., 2014). In a somewhat reflexive response to this style of parenting, others advocate for “free range parenting” (i.e., encouraging children to function independently and with limited parental supervision, in accordance with their age of development and with a reasonable acceptance of realistic personal risks) and “rewilding” (i.e., encouraging children to explore nature independently; Flynn, 2018; Outside Magazine & Online Editors, 2018; Prichep, 2018). Of course, most parents will fall in-between these extremes and express caregiving more consistent with the above-mentioned APA guidelines. Nevertheless, this debate continues and even has been raised within law- and policy-making. In 2018, Utah became the first U.S. state to protect free-range parenting by passing a state law changing the definition of neglect (Flynn, 2018). In this and similar cases, law- and policy-makers are challenged with decisions requiring an appropriate determination of whether parents were too harsh or appropriately protective; too permissive or effectively supporting autonomy. There is a similar ongoing debate regarding the effects of family leave policy on child well-being (Bartel et al., 2018; Burtle & Bezruchka, 2016; Ferrante, n.d.). Such debates beg the question of whether the effects of parenting may be U-shaped in reality, such that extremely high or extremely low parental care or control has a negative impact, while striking a balance yields optimal outcomes. The scientific community can leverage evidence-based research to inform and advocate for policies that will yield optimal developmental trajectories, and such advocacy will benefit from increasing our understanding of how normative parenting affects the brain.

4.3 Recommendations for future research

In light of these existing gaps in the literature and the clear need for improved understanding of the research question, we propose the following recommendations for the continued investigation of normative range parenting and brain development. First, we recommend a conservative approach to neuroimaging that emphasizes phenotypes readily collected in most studies and populations, such as high-resolution measures of brain structure (e.g., gray matter volume, cortical thickness, surface area, and fractional anisotropy) and intrinsic functional connectivity. Striving for such data harmonization will allow for attempts to replicate observed associations, which is critical for any clinically meaningful applications. Similarly, neuroimaging protocols should be administered to large samples providing adequate power to identify likely small effects and protect against false positives. Successfully translating research into the development of clinically meaningful biomarkers of risk or resilience is further dependent on establishing and employing measures with good test-retest reliability (Enkavi et al., 2019; Hedge et al., 2018; Nord et al., 2017; Plichta et al., 2012).

An equally conservative and thorough approach should be taken in measuring normative range parenting. In related research, a recent meta-analysis (Baldwin et al., 2019) found poor agreement between prospective and retrospective measures of childhood maltreatment. Similar discrepancies may be characteristic of retrospective versus prospective reports of normative parenting as well. Acquiring observational or in-home data continues to be the gold standard for gathering objective data of any kind, and parenting is no exception. However, it is reasonable to expect that, for most researchers, acquiring observational data on parent-child dyads will not be feasible. Moreover, as the expected sample size for neuroimaging studies grows, obtaining observational parenting data on a full sample of participants becomes increasingly difficult. In the absence of more objective parenting data, we recommend that researchers collect questionnaire data from as many family members as well as others familiar with the family dynamics (e.g., close neighbors, grandparents) as possible. Researchers can then combine these multi-informant reports to identify shared variance that may provide better insight into parenting than child- or parent-report alone. As many studies have questionnaire data available as rated by only one observer (i.e., child-report or parent-report but not both), researchers should take great care in interpreting and framing observed associations with brain development. Studies examining child-report as opposed to parent-report, or retrospective versus prospective measures, may be addressing different research questions.

Last, a return to theoretical frameworks of parenting may be useful in building common ground from which researchers can investigate normative range parenting and the developing brain. Fifty years have passed since Baumrind (1967) originally proposed the three parenting style model that was later expanded upon by Maccoby and Martin (1983). Despite many decades of research, these remain the only parenting styles with a strong empirical basis (Power, 2013). Much of the challenge in synthesizing results across the records included in this review stemmed from the wide variety of parental constructs assessed. We think an increased focus on the specific dimensions of parental care and control will facilitate the convergence of research findings with meaningful implications for development and well-being. The Parental Bonding Instrument (PBI; Parker et al., 1979) is one example of a validated child-reported measure designed to assess care and control, but we encourage researchers to utilize tools appropriate to their specific research questions and data sets.