The moderating role of socioeconomic status on level of responsibility, executive functioning, and cortical thinning during adolescence

Abstract

Brain development is exquisitely sensitive to psychosocial experiences, with implications for neurodevelopmental trajectories, for better or worse. The premise of this investigation was that the level of responsibility in adolescence may relate to brain structure and higher-order cognitive functions. In a sample of 108 adolescents, we focused on cortical thickness (using FreeSurfer) as an indicator of neurodevelopment in regions previously implicated in executive functioning (EF) and examined performance on an EF task outside of the scanner, in the context of level of responsibility. We further investigated whether socioeconomic status (SES) and family stress moderated the relationship between responsibility and brain structure or EF. Findings revealed that greater responsibility was related to thinner left precuneus and right middle frontal cortex. In lower SES adolescents, greater responsibility predicted thinner left precuneus and right middle frontal cortex, which have been consistently implicated in EF. Higher SES adolescents did not show structural differences related to responsibility, however they did exhibit better EF performance. It may be that circumstances surrounding the need for greater responsibility in lower SES households are detrimental to neurodevelopment compared to higher SES households. Alternatively, responsibility may act as a protective factor that bolsters cortical thinning in regions related to EF.

Introduction

Adolescence represents a developmental period typified by strength and vulnerability; a dichotomy that neuroscientists have been attempting to understand from a neurodevelopmental perspective for decades. During this time, the brain is undergoing major topological remodeling and, as such, is exquisitely sensitive to a wide range of environmental inputs, for better or for worse. On the negative end of this spectrum, adverse experiences (e.g., child maltreatment, poverty, witnessing violence) are known to contribute to poor outcomes from adolescence into adulthood, including poor self-regulation, substance abuse, and other types of psychopathology (e.g., Evans & Kim, 2013; Jaffee, 2017; Lovallo, 2013; Rossow, Felix, Keating, & McCambridge, 2016; Wright, Fagan, & Pinchevsky, 2013). In contrast, more positive experiences such as supportive parenting/mentoring, safe neighborhoods, well-equipped schools and positive peer influences are predictive of healthy self-regulation and the absence of psychopathology throughout adolescence and into adulthood (e.g., Farley & Kim-Spoon, 2014; Hurd, Stoddard, & Zimmerman, 2013; Niehaus, Rudasill, & Rakes, 2012; van der Voort et al., 2014). A significant body of research has suggested that neurodevelopmental trajectories may be altered toward or away from positive outcomes via these contextual factors.

One psychosocial influence that may mediate or moderate the effects of adversity during adolescent development but has received little attention is the level of responsibility that is expected of, or actuated by, adolescents. Adolescent responsibility is conceptually defined herein as developmentally-appropriate tasks or activities that are expected or required of adolescents by caregivers in the household. Given the brain’s experience-dependence, a healthy degree of responsibility expected by caregivers may exert subtle, positive effects on brain development in regions associated with self-regulation and dimensions of executive functioning (EF) that are still under construction (e.g., Casey, Getz, & Galvan 2008). Alternatively, permissive parenting - with caregivers not having many expectations for their teen to contribute to household chores and other family needs - may place insufficient demands on neural systems to advance a healthy rate of neurodevelopment. In other words, the brain’s experience-dependence may translate to the ability of level of responsibility to exact subtle influences on the brain’s rate of development and subsequent functioning.

There has been little research that explicitly focuses on the relationship between neurocognitive development and level of responsibility, either expected or actuated, within the household among adolescents. Studies that have focused on other outcomes report both positive and negative effects of level of household responsibility on adolescent well-being. For example, some studies have found that greater expectations for performing housework are associated with happiness. In contrast, other studies report that a higher level of responsibility are related to feelings of burden and poorer grades (Telzer & Fuligni, 2009). Although not measured, other features of the household (e.g., parent-child relationship, emotional support, financial security) are likely play a crucial role. A few researchers have theorized that responsibility is largely a positive, autonomy-building experience for adolescents that may support development of self-regulation abilities and goal-directed behaviors (Raffaelli. Simpkins, Tran, & Larson, 2018; Riggio, Valenzuela, & Weiser, 2010; Wray-Lake & Syvertsen, 2011). Given that maturation of EFs supports development of self-regulation, goal-direction, and other abilities, it may be the case that level of responsibility helps to shape integrity of these functions throughout adolescence (Boelema et al., 2014).

A related body of work has examined “family obligation values” in Mexican adolescents, finding that they play a beneficial role in their well-being, autonomy building, and serve to mitigate risk for psychopathology, including internalizing symptoms and substance abuse (Telzer, Fuligni, Lieberman, & Galván, 2013; Telzer, Tsai, Gonzales, & Fuligni, 2015). Although, these findings may be dependent upon the family context in that increased family conflict and decreased resources may moderate the extent to which increased responsibility is beneficial (Telzer et al., 2013). Indeed, the parent-adolescent relationship is likely a major determinant of how adolescents respond to demands that are placed upon them in the home; e.g., greater parent-adolescent conflict has been found to moderate the relationship between housework and depressive symptoms (Lam, Green, & McHale, 2016).

Family economic strain also plays an important role in adolescents’ levels of responsibility. Compared to adolescents from higher SES families, those from lower SES families tend to be tasked with higher levels of responsibility due to greater need (Tsai, Telzer, & Fuligni, 2013). Moreover, adolescents from lower SES backgrounds with higher levels of household responsibility are more likely to use drugs (Telzer, Gonzales, & Fuligni, 2014). Indeed, in the context of family need (e.g., decline in parental physical health), adolescents with greater household responsibility exhibit more internalizing symptoms (Telzer et al., 2015). Thus, although responsibility may generally be a positive and formative experience for adolescents, it is largely dependent upon the home context in which adolescents reside. To date, no study has examined how home context and household responsibility interface to impact adolescent neurodevelopment. Time spent engaging in household chores increases significantly from childhood to mid-adolescence, making adolescence an ideal time to study this relationship (Lam et al., 2016).

In addition to the association between economic strain and the potential burden of adolescent responsibility, economic disadvantages have consistently been linked to perturbations in neural development (for review, Johnson, Riis, & Noble, 2016). Specifically, an array of SES indicators, such as neighborhood, family income, parental education, and the income-to-needs ratio have been reliably associated with aberrant structural brain development in economically disadvantaged compared to advantaged youths (Lawson, Duda, Avants, Wu, & Farah, 2013; Luby et al., 2013; Nobel, Houston, Kan, & Sowell, 2012; Noble et al., 2015; Ursache & Noble, 2016). Notably, in youths from low SES families, relatively small differences in income can confer considerable differences in brain structure development, particularly in areas supporting EF and learning (Hair et al., 2015; Noble et al., 2015). Income differences within high SES youths do not confer the same level of brain maturational deficits, suggesting that incremental income disparities in higher SES youths do not present the same risk for neurodevelopment as in lower SES youths. By and large, children from lower SES families tend to have smaller regional volumes and thinner cortical regions throughout the brain compared to those from higher SES families, suggesting a potential neurodevelopmental lag (Jednoróg et al., 2012). Regions impacted in lower SES children, including those that subserve development of EF skills, show less cortical surface area, including the inferior frontal gyrus, orbitofrontal gyrus, medial frontal gyrus, and precuneus (Noble et al., 2015). Other studies have found that SES may moderate patterns of cortical thinning such that in lower SES, age-related cortical thinning is initiated earlier and follows a curvilinear pattern, whereas in higher SES youth, cortical thinning is initiated later and follows a more linear pattern (Piccolo et al., 2016). Altogether, these findings suggest that examining SES as a moderator in the relationship between responsibility and cortical thinning may be critical to discerning the complex relationships among contextual factors in contributing to neurodevelopment.

The present investigation was designed to explore relationships between adolescent responsibility and the confluence of responsibility and contextual factors (i.e., socioeconomic status) on adolescent neurodevelopment (Zhou, Level, Treit, Evans, & Beaulieu, 2015). There are no previous studies to guide a priori expectations for which brain regions are likely to be recruited during responsible activities. In fact, responsibility is not just a task-oriented action; it is more of a parental technique and overall expectation for child behavior with potential to dynamically influence development. Thus, hypotheses were derived from conceptualizations of responsibility and the neural regions most likely to be affected by such ongoing expectations and experiences. We focused on cortical thickness as an indicator of maturation in neural regions that subserve EF and behavioral self-regulation. By and large, previous research indicates that many cortical regions reach peak thickness at approximately 10-12 years old, followed by a steady decline in cortical thickness due to pruning that stabilizes in the mid-to-late-twenties (Fjell et al., 2015; Giedd et al., 1999; Tamnes et al., 2017; Vidal-Pineiro et al., 2019). Brain regions examined here have generally been found to follow a similar trajectory; for example, longitudinal work has shown that the frontal and parietal cortices begin to undergo significant dendritic pruning and myelination around ages 11-12 (Giedd et al., 1999; Giedd 2004). Therefore, in our slightly older sample (11-13 years old), evidence of decreased cortical thickness with greater levels of responsibility may indicate more normative brain development. It was hypothesized that a higher level of responsibility would be associated with decreased cortical thickness in brain regions subserving EF (e.g., problem solving) and self-regulatory behaviors (e.g., superior, orbital and middle frontal cortices, precuneus, and anterior cingulate). In addition, we considered performance on a problem-solving task outside the scanner in the context of adolescents’ level of responsibility, with expectations that levels of performance and responsibility would be positively correlated.

Second, we considered the influence of contextual factors that may mitigate positive effects of responsibility on neurodevelopment. SES and family stress were included as proxies for the degree to which adolescents have a higher level of responsibility as a function of adversity, as opposed to a teaching or skill building tool applied by parents in higher income households. It was hypothesized that lower SES adolescents and those with greater family stress would have thicker regions in conjunction with greater responsibility, while adolescents from higher income households and who have less family stress would demonstrate a more normative pattern of thinning cortical regions in the context of greater responsibility. Furthermore, it was predicted that SES and family stress would moderate the relationship between responsibility and performance on a problem-solving task (i.e., Stockings of Cambridge). Based on prior evidence (e.g., Johnson, Riis, & Noble, 2016), economic and family strain were expected to impede development of EF; adverse contexts in which responsibility is imposed may mitigate the otherwise beneficial effects. Accordingly, adolescents from higher SES households and with less family stress were expected to benefit from higher levels of responsibility, as indicated by better performance on the EF task. This line of research has potential to eventually inform our understanding of how certain parenting expectations and demands may influence neurodevelopment in adolescence, and the extent to which environmental contexts may alter ultimate brain and behavioral outcomes.

Methods

Participants

Drug and alcohol naïve adolescents (N=147; age = 11 – 13 years old) were recruited to participate in the Adolescent Development Study (ADS), a prospective, longitudinal study of the neural, cognitive, and behavioral precursors and consequences of adolescent drug and alcohol use in the greater Washington, D.C metropolitan area. Participants were excluded on the basis of substance use, prior head injury, left-handedness, neurodevelopmental disorders (e.g., epilepsy, autism spectrum disorders, and motor disorders), as well as other contraindications for magnetic resonance (MR) imaging (e.g., MR incompatible metal foreign bodies). The RTI International and Georgetown University Institutional Review Boards approved the study. All caregivers and youth provided consent or assent, respectively, prior to participating in the study.

A range of neurocognitive, behavioral, contextual, brain imaging, and demographic measures were acquired during a substance-naïve baseline assessment (i.e., Wave 1), and at 18- and 36-month follow-up visits (i.e., Waves 2 and 3, respectively) to capture cognitive and affective processes/patterns hypothesized to be predictive of drug/alcohol use initiation and escalation (see Fishbein et al., 2016 for complete description of the full sample, recruitment technique, and methods). The data presented here are based on a subsample of adolescents from the Wave 1 assessment who had both responsibility questionnaire and usable MRI data (N = 108; see Table 1 for sample demographics). Participants ranged from ages 11-14 (M=12.65, SD= .72) with 57 females.

Table 1.

Sample Demographics

	Total (N)	Sex (F:M)	Age (M(SD)); Range in years	Race/Ethnicity (% W:L:B:Other)	Parental Education (M(SD); years)	Household Income (M(SD); Range)
Total Sample	108	57:51	12.65(.72); 11 – 14	57%:7%:30%:6%	16.45(2.86)	12.67(.74); 11 – 14
Low SES group	53	28:25	12.56(.78); 11 – 14	32%;7%;55%;6%**	14.13(2.02)**	12.56(.79); 11 – 14
High SES group	55	30:25	12.74(.65); 11 – 13	78%;7%;8%;7%	18.39(1.82)	12.78(.67); 11 – 14

Note: W = White/Caucasian; L = Latino/a; B = Black/African American. Household income range: 1 = < $5000 per year to 15 = $200,000 or more.

^**p<.01,

^*p<.05 group difference between low and high SES groups based on t-test or χ2 analyses.

Measures

Responsibility questionnaire.

After a thorough search, no existing instrument capturing specific aspects of level of responsibility we sought to measure were found; thus, we constructed a questionnaire to probe adolescents and their caregivers regarding adolescents’ level of responsibility, demands and expectations within the context of their family and home lives in the last 30 days. In particular, parents answered questions about their expectations of their adolescent to routinely complete certain tasks, while adolescents answered questions about their perceptions of their parent’s expectations as well as their actual responsible behaviors. Adolescents responded to 30 questions (15 items for perceived parental expectations, 15 for actual behaviors). Parents responded to 9 questions regarding expectations of adolescent behavior, all of which corresponded to the adolescent questions. Behaviors in the past 30 days were surveyed on a Likert scale of never, a few times per month, every week, every few days, and every day. Example behaviors included: setting the table, washing dishes, making one’s bed, cleaning the house, doing laundry. Survey items that were not normally distributed were excluded (e.g., shopping for household, cooking for family). Moreover, two items with a lack of affirmative responses were excluded; i.e., making a financial contribution to the household or to one’s education garnered < 7% affirmative responses. Following these data cleaning procedures, 7 adolescent-reported expectation items, 7 adolescent-reported actual behavior items, and 5 parent-reported expectation items remained. To ascertain both actual adolescent behavior and parental expectations of adolescents, these items were summed (i.e., 7 adolescent report items, 5 parent report items) (M = 26.90, SD = 7.14; range = 12 – 44). We excluded adolescent perceived expectations for two reasons: 1) to avoid biasing the survey toward adolescent responses (i.e., having two assessments of adolescent perceptions versus only one of parent report), which could distort the validity of the scale and 2) to gain both an adolescent and a parent perspective on adolescents’ responsible behaviors. Inter-item correlations between adolescent and parent report are reported in Table 2. Summed self-report actual and parent expected behaviors were moderately correlated (r = 0.51, p < 0.01). To examine whether variance in the 12 items from self- and parent-report was convergent, we also performed a Principal Component Analysis (PCA) with a Promax (oblique) rotation. The PCA revealed two components and indicated that the 12 items largely converged onto component 1, suggesting that combining the self- and parent-report measures is statistically appropriate. Internal reliability of the measure was acceptable (Cronbach’s α = .74), increasing confidence in the overall structure of the survey.

Table 2.

Correlations between adolescents’ self-reported actual responsibility and parent report of expected behaviors

Variables	Parent Expect – Setting table	Parent Expect – Washing dishes	Parent Expect – Making bed	Parent Expect – Cleaning house	Parent Expect – Laundry
Self Actual – Setting Table	0.46**	0.12	0.01	−0.02	0.07
Self Actual – Washing dishes		0.37**	0.05	0.12	0.07
Self Actual – Making bed			0.39**	−0.03	0.05
Self Actual – Cleaning house				0.31**	0.13
Self Actual – Doing laundry					0.49**
Self Actual – Taking care of pets*
Self Actual – Cleaning own room*

^*Notes. p < .05,

^**p < .01. Pearson correlations in bold are between items reflecting the same actual and expected behaviors. *Parents were not asked about adolescent behavior for this item.

Stockings of Cambridge.

The Stockings of Cambridge (SOC) task is a strategy and problem-solving task, which is an adaptation of the commonly used Tower of Hanoi task (Shallice, 1982). The SOC is administered on a tablet via CANTAB® (Cambridge Cognition, Cambridge, UK; http://www.cambridgecognition.com/cantab) and is typically completed within 10 minutes. In the task, participants were given two displays on the top and bottom of the screen, which both included 3 stockings containing 3 colored balls each. The top display contained a pattern that the participants were instructed to duplicate in the bottom display. Importantly, participants were instructed to duplicate the top display in as few moves as possible by moving the balls one at a time. Outcome measures include initial thinking time, level of difficult reached, and mean number of moves used. In light of the mean age of participants and potential ceiling effects at the easier level in non-impaired, older developmental samples (e.g., Gau & Shang, 2010; Matthews, Coghill, & Rhodes, 2008) we focused on performance outcomes (i.e., mean number of moves needed to solve the problem) for the most difficult level in our analyses. Specifically, we examined mean number of moves in a problem set that requires a minimum of 5 moves to complete, rather than the easier problem sets with that require 1 to 4 moves to complete.

Socioeconomic status and family stress.

Family socioeconomic status (SES) was calculated based on parental education and income using a method adapted from Manuck et al. (2010) (also see McQuaid, Darcey, Fishbein, & VanMeter, 2019). A standardized education z-score was calculated for each participant based on an average of maternal and paternal years of education. A standardized income z-score was calculated for each participant based on caregiver report of annual household income prior to taxes (i.e., from $5000 to ≥ $200,000). The final SES index was calculated based on the average of the standardized education and income variables for each participant. These values were then re-standardized to create a distribution with a 0-centered mean and a standard deviation of 1. In several analyses, a median split of SES was used to assess differences between high and low SES groups. However, within those subgroups SES was modeled as a discrete variable.

The Drug Use Screening Inventory-Revised (DUSI-R; Tarter, 1990) was used to assess household stress using the family systems subscale. Specifically, the DUSI-R includes 14 questions about family environment, including items about general household stress; e.g., “Did you feel in danger at home?”; “Did you have frequent arguments with your children, parents, or spouse which involved yelling and screaming?”; “Has a member of your family been arrested?”. This scale was used to account for general household stress as a potential correlate of SES and a contributor to variability in responsibility.

MRI data acquisition.

Participants were scanned using a 3T Siemens Tim Trio MRI machine (Erlangen, Germany) with a 12-channel head coil. Foam padding was placed around participants’ heads to minimize head motion. A high resolution T1-weighted anatomical scan was acquired using a magnetization prepared rapid acquisition gradient echo (MPRAGE) protocol with the following parameters: TR/TE = 1900 ms/TE = 2.52 ms, TI = 900 ms, flip angle = 9°, FOV = 250 mm², 176 slices with a thickness of 1.0 mm, effective resolution = 1mm³, scan time = 4 min, 18 s. Prior to scanning, participants underwent training to prepare for the scanning environment and task protocol.

MRI data quality control.

Prior to the FreeSurfer analysis, MPRAGEs were inspected for image quality by at least 3 trained raters. These structural images were scored on a scale of 1 (high quality) to 5 (low quality) for ghosting and ringing. Clipping and wrapping were scored with either a 1 (low quality) or 0 (high quality). A single quality control (QC) metric was calculated using the following computation (QC = 4*wrap + 10*clip + 2*ring + 1*ghost). Images with a QC metric >11 were deemed to be of poor overall quality and excluded from further analysis (N=23).

MRI data processing.

Cortical reconstruction and volumetric segmentation was performed using the Freesurfer image analysis suite (v6.0) (i.e., the recon-all pipeline; see http://surfer.nmr.mgh.harvard.edu/ for documentation). The technical details of these procedures are described in prior publications (e.g., Fischl et al., 2002; Reuter, Rosas, & Fischl, 2010, Reuter, Schmansky, Rosas, & Fischl, 2012). Briefly, image processing included motion correction and averaging (Reuter et al. 2010) of volumetric T1 weighted images, removal of non-brain tissue (Ségonne et al., 2004), automated Talairach transformation, segmentation of the subcortical white gray matter (Fischl et al., 2002) intensity normalization (Sled, Zijdenbos, & Evans, 1998), tessellation of the gray matter white matter boundary, automated topology correction (Ségonne, Pacheco, & Fischl, 2007), and surface deformation following intensity gradients (Fischl and Dale, 2000). Cortical thickness estimates of regions implicated in EF were derived from the Destrieux 2009 Atlas available in the FreeSurfer package (Destrieux Fischl, Dale, & Halgren, 2010). Regions selected included bilateral anterior cingulate, precuneus, middle frontal cortex, orbital frontal cortex, frontal pole, and superior frontal cortex, all of which have been identified as regions pertinent to EF (e.g., Collette, Hogge, Salmon, Van der Linden, 2006; Mace, Waters, Sawyer, Turrisi, Gansler, 2019).

Data Analysis

Linear regression analyses with the summed responsibility score were conducted to determine whether it predicted cortical thickness in regions associated with EF (i.e., bilateral orbital, middle, and superior frontal cortices, frontal pole, anterior cingulate, and precuneus). Importantly, because of the known age-related effects on cortical thickness (e.g., Tamnes et al., 2017), we included age as a first-level covariate in these analyses. Participant sex was not included as a covariate in any analyses, as there were no significant correlations with any of our study variables (i.e., cortical thickness, responsibility, family stress, SES, and SOC performance; all p > .05). We also conducted hierarchical regression analyses to distinguish a) the relative contributions of EF (SOC performance) and responsibility to cortical thickness and b) whether responsibility is predictive of cortical thickness above and beyond EF (i.e., SOC performance). To do so, SOC performance was entered first into the model, followed by responsibility.

In separate linear regressions, we also ascertained whether contextual factors (DUSI family stress and SES) played a role in level of responsibility. Further hierarchical regression analyses were conducted to understand the relative contributions of SES, family stress, and responsibility to cortical thickness. As in prior studies, we did not control for total brain volume or intra-cranial volume, as this violates various statistical assumptions (e.g., Wierenga et al., 2017, 2018), including that equal variances are assumed between groups.

Based on results of analyses showing an association between SES and responsibility, we examined whether cortical thickness of regions that were significantly related to responsibility differed between high and low SES groups (i.e., median split) via linear regressions, controlling for age. Our purpose was to examine whether SES status (high or low) moderated the relationship between responsibility and cortical thickness. We also conducted confirmatory hierarchical regression analyses in each SES group to examine whether responsibility, above and beyond variability in SES and family stress, predicted cortical thickness.

Finally, to explore the association between level of responsibility and EF as it relates to SES, we conducted additional linear regressions in high and low SES groups with performance in the Stockings of Cambridge task (i.e., mean moves for the most difficult problem set) predicting responsibility.

Results

Responsibility, Executive Functioning, and Cortical Thickness

The first set of linear regression analyses focused on thickness measures in regions that support executive control functions (i.e., bilateral orbital, middle, and superior frontal cortices, frontal pole, anterior cingulate, and precuneus) (see Figure 1 & Table 3). Consistent with our hypotheses, these analyses revealed that a higher level of responsibility in adolescents was associated with reduced cortical thickness in the left precuneus (r²=0.06, p = 0.04) and right rostral middle frontal cortex (r² = 0.06, p =0.04). Notably, age was not a significant predictor of cortical thickness in either region (left precuneus: r²= .02, p=.11; right middle frontal: r²= .02, p=.13). The hierarchical regression analyses focused on the relative contributions of responsibility and SOC performance (i.e., EF) on thickness measures revealed that EF did not significantly contribute to cortical thickness in the left precuneus (F_(1,102) = 0.63, p = .43) or the right middle frontal cortex (F_{(1, 101)} = 0.00, p = .98) (Supplementary Material Tables S1–2). Above and beyond EF, greater responsibility uniquely predicted thinner left precuneus when added to the model (F_(1,101) = 4.03, p = .047). The overall model, however, was not significant (F_{(2, 101)} = 2.34, p = .10). Likewise, above and beyond EF, greater responsibility uniquely predicted thinner right middle frontal cortex (F_{(1, 100)} = 3.86, p = .05). The overall model was not significant (F_{(2, 100)} = 1.93, p = 0.15). In the next set of analyses, we examined whether and to what extent contextual factors (i.e., SES and family stress) related to levels of responsibility and cortical thickness in these regions.

Figure 1.

Regression Results for Responsibility and Cortical Thickness. Fig 1a. Linear regression analyses indicated correspondence between left precuneus cortical thickness and responsibility; 1b. Likewise, linear regression analyses indicated correspondence between right middle frontal gyrus thickness and responsibility. Both 1a and 1b illustrate that with more responsibility, adolescents have thinner regions.

Table 3.

Regressions with Cortical Thickness Predicting Responsibility

Brain Region	B	SE B	β	t	p
L anterior cingulate	−1.76	2.97	−0.06	−0.59	0.56
R anterior cingulate	2.35	3.46	0.07	0.68	0.50
L precuneus	−11.37	5.25	−0.21	−2.17	0.04*
R precuneus	−8.37	4.81	−0.17	−1.74	0.09
L rostral middle frontal	−2.17	4.57	−0.05	−0.48	0.64
R rostral middle frontal	−9.47	4.46	−0.20	−2.12	0.04*
L medial orbital frontal	−6.70	4.39	−0.15	−1.53	0.13
R medial orbital frontal	−8.09	4.28	−0.18	−1.89	0.06
L frontal pole	−0.30	2.09	−0.01	−.14	0.89
R frontal pole	−3.93	2.04	−0.19	−1.93	0.06
L superior frontal	−3.97	5.19	−0.07	−0.73	0.47
R superior frontal	−3.34	5.05	−0.06	−0.66	0.51

^*Notes. p < 0.05,

^**p < 0.01. B: unstandardized beta, SE B: standard error for unstandardized beta, β: standardized beta. Significant p values are bolded.

Contextual Influences on the Association Between Responsibility and Cortical Thickness

We first evaluated whether potential sources of environmental stress—specifically SES and family stress (i.e., measured using the DUSI-R family subscale)—were associated with level of responsibility. Age was not a significant predictor of responsibility in separate linear regressions with SES (r² = 0.02, p = 0.08) and family stress (r² = 0.03, p = 0.09). Linear regression results suggested that both SES and family stress are predictive of responsibility (r² = 0.07, p = 0.01 & r² = 0.10, p = 0.01, respectively). In both cases, increased environmental stress was predictive of higher level of responsibility (see Figure 2).

Figure 2.

SES and DUSI Family Predicting Responsibility Levels. Fig 2a. Linear regression analyses revealed that adolescents’ family socioeconomic status (SES) is associated with adolescent responsibility such that the lower SES group engages in higher levels of responsibility. Similarly, 2b shows linear regression results for DUSI family stress and responsibility, indicating that with greater family stress, adolescents engage in higher levels of responsibility.

To disentangle how environmental stress and responsibility uniquely contribute to variance in cortical thickness in the previously identified brain regions, we ran hierarchical regression models for each region. As a direct test of our hypotheses, we entered SES into the first step of the model, followed by family stress, and then responsibility. This 3-step hierarchical regression allowed us to disentangle the relative contributions of each predictor to cortical thickness. Given the relatively high mean SES of this DC-based cohort and the lack of anticipated effects of SES for participants in higher economic brackets, in addition to constructing hierarchical regression models for the entire sample, we also considered the impact of responsibility for those in high vs. low SES cohorts (i.e., median split) separately.

Whole group.

In the left precuneus, SES did not contribute to the variance in thickness (F(_1,106) = 0.73, p = 0.39; Table S3). Similarly, adding family stress did not uniquely contribute to the variance in left precuneus (F_(1,105) = 0.18, p = 0.67), nor was the overall model significant (F_(1,105) = 0.45, p = 0.64). However, when responsibility was added to the final step of the model, it did account for a significant amount of additional variance in left precuneus thickness (F(_1,104) = 4.10, p = 0.04), but the overall model was not significant (F_(1,104) = 1.68, p = 0.18). That is, higher level of responsibility alone predicted thinner left precuneus.

The same pattern of results was found for the right rostral middle frontal cortex for the whole sample (Table S4). That is, SES and family stress did not contribute to variance in thickness (F_(1,105) = 0.00, p = 0.96 & F_(1,104) = 0.04, p = 0.84) and the overall model was not significant (F_(2,104) = 0.02, p = 0.98). When responsibility was added to the final step of the model, it accounted for additional variance in region thickness (F_(1,103) = 4.56, p = 0.03). Although, the overall model was not significant, F_(3,103) = 1.53, p = 0.21. Together, these findings demonstrate that for the whole sample, responsibility uniquely contributed to thickness of cortical regions.

High and low SES subgroups (see Table 4).

Table 4.

Regression Results for Responsibility Predicting Cortical Thickness in Low and High SES Groups

Low SES
Brain Region	B	SE B	β	t	p
L precuneus	−0.01	0.00	−0.39	−2.88	0.006**
R rostral middle frontal	−0.01	0.00	−0.42	−3.16	0.003**
High SES
Brain Region	B	SE B	β	t	p
L precuneus	0.00	0.00	0.00	0.03	0.98
R rostral middle frontal	0.00	0.00	0.04	0.29	0.77

^*Notes. p < 0.05,

^**p < 0.01. Significant p-values are bolded.

We considered whether there were linear associations between responsibility and cortical thickness (controlling for age) for each of the regions that were noted in the initial analysis (i.e., precuneus and rostral middle frontal cortex) in SES groups separately. In the low SES group, age was not a significant predictor of cortical thickness in either region (left precuneus: r²= .01, p=.64; right middle frontal: r²= .01, p=.59) Conversely, in the high SES group, age was a significant predictor of cortical thickness in the left precuneus (r²= .10, p=.01) but not the right middle frontal cortex (r²= .06, p=.06). Controlling for age, in the low SES group, level of responsibility was negatively associated with cortical thickness in the left precuneus (r² = 0.15, p = 0.01; Fig 3a), while in the high SES group, the same association was not found (r² = 0.00, p = 0.98) (Fig 3b). Likewise, in the low SES group, higher levels of responsibility significantly predicted thinner right rostral middle frontal cortex (r² = 0.11, p = 0.02; Fig 3c), while no such associations were found in the high SES group (r²= 0.00, p = 0.76; Fig 3d).

Figure 3.

Regression Results for Responsibility and Cortical Thickness in Low and High SES Groups. Linear regression results in Fig 3a illustrate that in lower SES adolescents, greater levels of responsibility correspond with thinner left precuneus. Conversely, 3b shows that in the higher SES adolescents, there is no association between responsibility and left precuneus thickness. Regression results in 3c show that lower SES adolescents’ greater levels of responsibility are significantly associated with thinner right middle frontal gyrus. Finally, 3d exhibits in the higher SES group, there is no association between level of responsibility and right middle frontal gyrus thickness.

To confirm these results, we repeated our hierarchical regression models to examine whether responsibility, above and beyond SES and family stress, predicted cortical thickness in each group. In the low SES group (Table S5), SES significantly predicted cortical thickness in the right middle frontal cortex (F_(1,46) = 10.85, p = 0.002). Adding family stress to the model did not account for any additional variance in right middle frontal thickness (F_(1,45) = 0.16, p = 0.69), though the model continued to be significant (F_(2,45) = 5.41, p = 0.01). In the final step, adding responsibility to the model did not contribute enough additional variance to result in a significant R² change (F_{(1, 44)} = 2.50, p = 0.12). The overall model remained significant, (F_(3,44) = 4.56, p = 0.01). For the left precuneus (Table S6), variability in SES significantly predicted cortical thickness (F_(1,47) = 6.15, p = 0.02). Adding family stress to the model did not add any additional variance (F_(1,46) = 0.39, p = 0.54) and the model was no longer significant (F_(2,46) = 3.23, p = 0.05). Finally, adding responsibility to the model contributed significant additional variance to the model (F_(1,45 = 5.24, p = 0.03), and the overall model was significant (F_(3,45) = 4.10, p = 0.01.

In the high SES group (Table S7), in the right middle frontal cortex, SES also significantly predicted cortical thickness (F_(1,55) = 7.11, p = 0.01). Family stress did not contribute any additional variance to the model (F_(1,54) = 0.42, p = 0.52), though the overall model remained significant (F_(2,54) = 3.72, p = 0.03). Adding responsibility in the final step did not add variance to the model (F_(1,53) = 0.00, p = 0.95), and the model was no longer significant (F_(3,53) = 2.43, p = 0.08. In the left precuneus (Table S8), SES did not predict thickness (F_(1,55) = 1.01, p = 0.32). In the next step, adding family stress to the model did not contribute additional variance (F_(1,54) = 0.00, p = 0.94) and the model was not significant (F_(1,54) = 0.50, p = 0.61). In the final step, adding responsibility to the model did not add additional variance (F_(1,53) = 0.00, p = 0.99) and the overall model was not significant (F_(3,53) = 0.33, p = 0.81).

Contextual influences on executive functioning.

To further elucidate whether responsibility in different economic contexts impacts cognitive processes in adolescents, we considered the degree of correspondence between level of responsibility and performance on an EF task (Stockings of Cambridge; SOC). Given the SES-group specific effects that were noted for cortical thickness, analytic models included separate regression analyses for low and high SES groups with responsibility predicting performance on participants’ mean number of moves in the most difficult problem set. Findings revealed that, in the low SES group, there was no correspondence between responsibility and performance on the SOC (Fig 4a). In contrast, in the high SES group, higher level of responsibility corresponded to better performance on the SOC (i.e., fewer overall moves), r² = 0.11, p = 0.01 (Fig 4b).

Figure 4.

Regression Results for Responsibility Predicting SOC Performance in Low and High SES Groups. Linear regression analyses revealed that socioeconomic status (SES) moderates the association between responsibility and performance on the Stockings of Cambridge (SOC) task. In 4a, the lower SES group does not show correspondence between responsibility and performance on the SOC. In contrast, 4b illustrates that in the high SES group, greater levels of responsibility relate to superior performance on the SOC task.

Discussion

To our knowledge, this preliminary study is the first to focus specifically on the impacts of actual and expected level of responsibility within the home on adolescent brain development. Our findings suggest that thinning of cortical regions crucial to EF may be related to higher levels of responsibility. Notably, we found that responsibility was predictive of cortical thinning in EF regions, above and beyond EF performance in a problem-solving task. Moreover, we found that the strength and direction of the relationship between responsibility and cortical thinning was related to SES, and occurred above and beyond age. These results are consistent with prior work demonstrating significant negative impacts of lower SES environments on children’s and adolescents’ brain structure (e.g., Johnson, Riis, & Noble, 2016). The present study introduces a new variable—level of responsibility—into that constellation of potential influencers; however, further explication is needed.

There are three core findings of particular interest. First, greater responsibility corresponded with thinner cortical regions implicated in EF (i.e., left precuneus and right middle frontal cortex; Burzynska et al., 2012; Munro et al., 2018). Second, environmental contextual factors—specifically SES and family stress—played a role in the association between responsibility and cortical thickness. Given that adolescents with lower SES and higher family stress exhibited greater levels of responsibility, we explored further and found that SES moderated the effect of responsibility on cortical thickness in regions critical to EF. That is, within the lower SES group, greater levels of responsibility corresponded with thinner regions (i.e., left precuneus and right middle frontal cortex). Conversely, the higher SES group of adolescents did not show this pattern of cortical thinning. And third, the higher SES group evinced a behavioral advantage, with greater responsibility relating to better EF performance (i.e., Stockings of Cambridge). In contrast, we did not find correspondence between EF and responsibility in the lower SES group. Of note, there was no association between this behavioral advantage in the higher SES group and cortical thickness.

Our initial finding that responsibility corresponds with cortical thickness in neural regions implicated in EF is consistent with the notion that responsibility may be beneficial to adolescents and may relate to healthy cortical thinning patterns. The significance of this finding is highlighted by the fact that chronological age did not contribute to the relationship; generally, we would expect age-related cortical thinning to occur in these regions during adolescence (Zhou et al., 2015). The implication is that adolescents engaging in greater levels of responsibility are experiencing thinning in brain regions implicated in EF; however, only under certain conditions. Specifically, the relationship appears to be largely driven by the socioeconomic stratification of our participants. One interpretation may be that stress conferred through low resource environments counteracts potentially positive behavioral effects of responsibility, as indicated by EF performance. More specifically, greater responsibility may promote cortical thinning and have the potential to lead to improved EF skills. However, in the context of stressors that are linked to responsibility, the actual performance on EF tasks shows no behavioral benefit. In other words, adolescents who are exposed to stressors and have greater responsibility may mature faster, with EF performance lagging behind in the short term. In contrast, greater responsibility in higher SES adolescents may not be “stressful” enough to drive brain development, perhaps because responsibilities are performed in a family context with fewer stressors and/or are not necessary to the same extent.

Specifically, lower SES adolescents in our sample engaged in higher levels of responsibility. Prior work has shown that more stressful family environments may result in reliance of caregivers on adolescents to contribute more to daily family functioning via responsible household behaviors, which may actually be burdensome to adolescents (Telzer & Fuligni, 2009). The current investigation builds upon previous studies by incorporating indices of brain structure to determine whether this practice has an effect on neurodevelopment. Although family stress and responsibility corresponded in our sample, it is unclear whether the association between higher levels of responsibility and brain architecture presents a risk or protective factor in lower SES adolescents. One interpretation is that greater responsibility confers additional stress for adolescents, and acts as a risk factor for premature thinning. In other words, premature cortical thinning and/or chronic stress-induced dendritic atrophy may be reflected in thinner cortical regions. Prior animal models have shown that chronic stress may induce localized dendritic spine loss, with reports of localized spine loss in medial prefrontal regions (Arnsten et al., 2015; Liston et al., 2006; Radley et al., 2005). Our findings of cortical thinning in middle frontal cortex in lower SES adolescents engaging in greater responsibility is consistent with dendritic spine loss. A parallel literature has introduced the notion of a “stress-acceleration” process, whereby early life stress accelerates aspects of neurodevelopment with potential evolutionary advantages, even if developmentally non-normative (Herzberg & Gunnar, 2020; Tyborowska et al., 2018). These studies might suggest that higher levels of responsibility for adolescents in economically-disadvantaged environments exacerbate the experience of stress, resulting in greater cortical thinning. As the present study did not measure the incidence or severity of early life stress, this explanation requires further inquiry.

Another interpretation is that responsibility may be a potentially protective factor by bolstering cortical thinning in regions related to EF in lower SES environments. In our data (Fig 3c and 3d), lower SES adolescents seem to have comparable levels of cortical thickness at greater levels of responsibility to the higher SES adolescents in the right middle frontal cortex, which may be suggestive of a “normalizing” effect of responsibility in the lower SES group. This possibility seems less plausible when considering prior work showing that cortical thinning may be initiated earlier in adolescents in lower SES environments, presumably due to higher levels of stress, as surmised above (Piccolo et al., 2016). It is also notable that with greater responsibility, the lower SES adolescents do not seem to exhibit the same performance advantages in the EF task as the higher SES adolescents do.

In contrast, the higher SES group appeared to benefit from engaging in more responsible behaviors in terms of their EF functioning. There is a vast literature documenting the ways in which individuals from enriched environments have enhanced EF skills compared to those from environments with fewer resources (Sarsour et al., 2011; Ursache, Noble, & Blair, 2015). Our findings add to this body of work by showing that responsibility is also associated with better performance on an EF task. It is possible that, for higher SES adolescents, engaging in greater levels of responsibility is supportive of optimal EF development and skills. Alternatively, this finding may suggest that adolescents with enhanced EF skills are more likely to elect to engage in responsible behaviors.

A number of limitations temper definitive conclusions. First, the current investigation was not longitudinal in nature; thus, it is not possible to infer causality regarding the impact of level of responsibility on patterns of cortical thinning. Future work on this topic would benefit from examining the extent to which the level of responsibility, or developmental changes in level of responsibility, are predictive of changes in cortical thinning in neural regions implicated in planning, decision-making, self-regulation, and even prosociality. Longitudinal studies could also determine whether and to what extent adolescents more likely to assume responsibility may have inherently healthier brain development due to other corollary characteristics (e.g., personality, executive functioning) or if becoming responsible drives brain development. Second, and relatedly, it will be important for future investigations to integrate multi-modal imaging methods to discern correspondence between structure (e.g., VBM, DTI) and function (e.g., functional connectivity, resting state, task-dependent fMRI) to parse out the relationships reported here.

Third, there is a lack of well-established measures of adolescent responsibility in the field. Although our measurement of responsibility incorporated both parent and adolescent reports, only household responsibilities were assessed. Future studies examining responsibility during adolescence could expand upon the current study by exploring different types of responsibility. For example, examining the potentially differential effects of types of responsibility in various settings (e.g., volunteer work, sports team participation, self-selected leadership opportunities) versus enforced or obligatory responsibility (e.g., parentally-driven responsibilities, homework, etc.) on brain development could have significant implications for future research on the impacts of experience on adolescent brain development and successful transitions to adulthood.

Finally, our sample was recruited from a relatively affluent area of the country, making the sample higher SES compared to the broader US demographic. To minimize the impact of this issue, our SES score was calculated to create a relative measure of SES for our specific sample. In addition, although the differences in our SES groups are minute, our findings are consistent with prior research demonstrating that even minor differences in SES, particularly in lower SES groups, can have pervasive impacts on adolescent brain structure development (Hair et al., 2015; Noble et al., 2015). Future work examining how SES moderates effects of responsibility on adolescent brain structure would benefit from recruiting a more economically diverse sample and incorporating measures of perceived SES. That is, measures of perceived SES have been shown to be predictive of an array of developmental outcomes, including structural brain development (Brito & Noble, 2014; Quon & McGarth, 2014).

Implications and Conclusions

The variability in level of social demands and expectations within the home may portend different neurodevelopmental trajectories, with some adolescents showing more protracted development while others exhibiting more normative maturation of neural circuits, particularly those that subserve dimensions of executive cognitive functioning, social competency skills and self-regulation of behavior. As suggested by the preliminary findings reported herein, family SES is one condition that may influence different neurodevelopmental outcomes in the context of responsibility expectations within the home. Given that the brain’s experience-dependence is not contingent upon any one type of experience, countless factors influence final outcomes, many of which we did not measure here. Thus, isolating the impacts of level of responsibility will require careful sampling and parsing of both study participants and conditions, including mediators and moderators of divergent trajectories. These results, however, are thought-provoking by raising the possibility that placing developmentally-appropriate expectations on children and youth may be advantageous for neurodevelopment and adaptive functioning in the long-term, but perhaps only in the presence of a supportive, nurturing and adequately resourced environment.

Heightened neuroplasticity during adolescence represents a prime developmental period for prevention efforts designed to mitigate environmental adversities. Although the negative effects of home environments on adolescent neurodevelopment tend to be well documented, less attention has been paid to the effects of positive, nurturing environments (Fishbein et al., 2019). In support of the ability to override, to some extent, negative environmental influences on functioning, there are a few promising intervention studies that are of particular relevance to the current investigation (Blair & Raver, 2016; Brody et al., 2017). For example, Brody and colleagues show encouraging effects on adolescents’ neurodevelopmental outcomes in economically disadvantaged families who received a supportive parenting intervention (Brody et al., 2017). Specifically, this work suggests that compared to controls, adolescents and their parents who received a supportive parenting intervention showed protective effects, such that poverty-related volume reductions were not indicated in hippocampus and amygdalae at age 25 (Brody et al., 2017). A growing body of literature from prevention science support the conjecture that positive psychosocial environmental input, particularly nurturing environments, can bolster normative brain development and adaptive behavior in adolescents and even buffer against maladaptive neurodevelopmental trajectories. Similarly, instilling developmentally-appropriate responsibility in adolescents may provide a unique opportunity for skill-building and teaching applied by caregivers; preventative interventions focusing on supportive parenting incorporating adolescent responsibility could provide a promising avenue for promoting adaptive functioning skills and general EF skills that are critical to adolescence and the transition to emerging adulthood.