Parental Education, Household Income, Race, and Children’s Working Memory: Complexity of the Effects

Background. Considerable research has linked social determinants of health (SDoHs) such as race, parental education, and household income to school performance, and these effects may be in part due to working memory. However, a growing literature shows that these effects may be complex: while the effects of parental education may be diminished for Blacks than Whites, household income may explain such effects. Purpose. Considering race as sociological rather than a biological construct (race as a proxy of racism) and built on Minorities’ Diminished Returns (MDRs), this study explored complexities of the effects of SDoHs on children’s working memory. Methods. We borrowed data from the Adolescent Brain Cognitive Development (ABCD) study. The total sample was 10,418, 9- and 10-year-old children. The independent variables were race, parental education, and household income. The primary outcome was working memory measured by the NIH Toolbox Card Sorting Test. Age, sex, ethnicity, and parental marital status were the covariates. To analyze the data, we used mixed-effect regression models. Results. High parental education and household income were associated with higher and Black race was associated with lower working memory. The association between high parental education but not household income was less pronounced for Black than White children. This differential effect of parental education on working memory was explained by household income. Conclusions. For American children, parental education generates unequal working memory, depending on race. This means parental education loses some of its expected effects for Black families. It also suggests that while White children with highly educated parents have the highest working memory, Black children report lower working memory, regardless of their parental education. This inequality is mainly because of differential income in highly educated White and Black families. This finding has significant public policy and economic implications and suggests we need to do far more than equalizing education to eliminate racial inequalities in children’s cognitive outcomes. While there is a need for multilevel policies that reduce the effect of racism and social stratification for middle-class Black families, equalizing income may have more returns than equalizing education.

At least some of the effects of high SES on school performance [81] can be attributed to the role of family SES on structure and function of the brain [33] and SES effect on memory [82][83][84]. Many brain structures, such as the amygdala [33,60,85], hippocampus [86], and PFC, carry the effects of SES on cognitive and behavioral outcomes. The amygdala is more involved in emotion regulation [87][88][89][90][91], while the hippocampus [86] and PFC [92][93][94] are more involved in cognitive tasks, executive function, and memory.

Aims
To investigate the complexities of social determinants of children's brain development in the US, we explored racial variation in the effects of two family SES indicators, namely parental education and household income, on working memory among 9-and 10-year-old children. We expected racial differences in the magnitude of the association between parental education on working memory, in line with the observed MDRs [36,37,45]. More specifically, we expected the weaker effects of parental education on working memory for Black than White children. This expectation is in line with the other research on a wide range of phenotypes and behaviors [36,37,45].

Design and Settings
This secondary analysis was a cross-sectional analysis of the baseline data of the ABCD study [95][96][97][98][99]. ABCD is a national, state-of-the-art brain imaging study of childhood brain development [95,100].

Participants and Sampling
The ABCD study sample was recruited from 21 cities across states. ABCD sampling was primarily through school systems. For sampling in the ABCD study, school selection was informed by race, ethnicity, sex, SES, and urbanicity [101]. Inclusion criteria were having data on our variables. Participants could be included regardless of race or ethnicity (n = 10,418). As this is a general population study of children, participants have been enrolled regardless of their psychopathologies. That means participants were not included or excluded from the sample based on the presence of psychopathology.

Primary Outcome
The primary outcome was working memory, measured by NIH Toolbox, the Dimensional Change Card Sort [102]. This measure [103] has shown high reliability and validity [9,104]. The NIH Toolbox card-sorting test is a part of the NIH Toolbox Cognition Battery (NIHTB-CB). This measure evaluates the executive function. The NIHTB-CB is designed for use in epidemiologic studies and clinical trials for ages 3 to 85. Some studies have documented very acceptable psychometric properties of the NIHTB-CB and card sort test. These are computer-based instruments assessing executive function: the Dimensional Change Card Sort, which measures cognitive flexibility, and a flanker task, which measures inhibitory control and selective attention. These measures show convergent and discriminant validity and correlate with SES. These measures also show excellent sensitivity to age-related changes during adulthood, excellent test-retest reliability. As a result, the Dimensional Change Card Sort can be used effectively in epidemiologic and clinical studies. Our outcome was a continuous variable in this study, with a higher score indicating higher cognitive flexibility [103,105].

Independent Variable
Parental Educational Attainment. Participants reported their years of schooling. This variable was operationalized as a five-level nominal variable: less than a high school diploma, high school diploma, some college, bachelor degree, graduate studies.

Moderator
Race. Race was reported by parents, and operationalized as a nominal variable: Black, Asian, Other/Mixed, and White (reference group).

Confounders
Ethnicity. Parents were asked if they were of Latino ethnic background. This variable was coded as Latino = 1 and non-Latino = 0.
Age. Age was a dichotomous variable coded 1 or 0 for 10 years and 9 years of age. Parents reported the age of the children.
Sex. Sex was 1 for males and 0 for females. Parental marital status. Parental marital status was 1 for married and 0 for any other condition (reference).

Data Analysis
We used SPSS for data analysis. Frequencies (n and %) and mean [standard deviations (SDs)] were reported for descriptive purposes. To estimate bivariate associations between the study variables, we used the Chi-square and Analysis of Variance (ANOVA) test in the pooled sample. To perform our multivariable analyses, we performed mixed-effects regressions. First, we tested the assumptions. We excluded collinearity between the study variables. We also tested the distribution of our outcome and error terms and quantiles ( Figure 1). We ran six models. All models were performed in the pooled sample. Model 1 to Model 3 did not have interaction effects. Model 1 had education but not income. Model 2 had income but not education. Model 3 had both education and income. Model 4 had interactive effects of education and race but not income. Model 5 had interactive effects of income and race. Model 6 had interactive effects of race and education and also controlled for income. Box 1 lists our model formulas. Unstandardized regression coefficient (b), standard error (SE), and p-values were reported for each model. A p-value of equal or less than 0.05 was significant.

Ethical Aspect
Our analysis was exempt from a full review. However, the ABCD study protocol was approved by the University of California, San Diego (UCSD) Institutional Review Board (IRB) [100].

Regression Results
Tables 2 and 3 report the results of six pooled sample mixed-effects regression models. All models are significant. Effect sizes as shown in Table 2. Model 1, which only included the main effect of race and parental education and covariates, showed that high parental education is associated with higher working memory. Model 2 showed that high income is associated with higher working memory. Model 3 showed that parental education and household income have both associations with working memory. Model 4 showed that parental education and race interact, meaning that parental education's boosting effect on working memory was less pronounced for Black than White children. Model 5 did not show an interaction between race and household income. Model 6 showed that household income explains why parental education and race interact with our outcome. (Figures 2-5).

Discussion
This study had three primary findings. Although higher parental education and household income predicted higher working memory (1st finding), parental education and household income showed weaker effects on Black children than White children's working memory (2nd finding). Third, income differentials (lower income levels of Black families with the same parental education and family structure) explain why parental education shows a weaker effect on working memory for Black than White children (3rd finding).
Our first results can be compared with the literature on the protective effects of high SES on children's cognitive outcomes [106]. A large body of literature has also documented poor educational outcomes in low SES than high SES children [107][108][109][110]. High SES is associated with better school performance [15] and working memory [16][17][18]. Literature has shown the effects of family SES indicators such as poverty and household income on the brain [111] and behavior [33,[59][60][61]112]. The SES-health link may be because SES is a proxy of stress, adversities, trauma. Thus, many brain 50

Discussion
This study had three primary findings. Although higher parental education and household income predicted higher working memory (1st finding), parental education and household income showed weaker effects on Black children than White children's working memory (2nd finding). Third, income differentials (lower income levels of Black families with the same parental education and family structure) explain why parental education shows a weaker effect on working memory for Black than White children (3rd finding).
Our first results can be compared with the literature on the protective effects of high SES on children's cognitive outcomes [106]. A large body of literature has also documented poor educational outcomes in low SES than high SES children [107][108][109][110]. High SES is associated with better school performance [15] and working memory [16][17][18]. Literature has shown the effects of family SES indicators such as poverty and household income on the brain [111] and behavior [33,[59][60][61]112]. The SES-health link may be because SES is a proxy of stress, adversities, trauma. Thus, many brain structures such as PFC, hippocampus, and amygdala [85,[113][114][115][116][117][118][119][120][121][122] correlate with SES. The prefrontal cortex [33], hippocampus [86], and amygdala [123] have been shown to be under the influence of trauma. SES impacts reduced connectivity in neural networks involved in memory and emotion regulation [124]. In a recent study by Brody et al., data of 119 African American youths living in the rural South were used. The study measured poverty status and supportive parenting at ages 11-13 and 16-18. The study conducted brain imaging at age 25. This study applied resting-state fMRI to study two brain networks' functional connectivity: (1) central-executive and (2) emotion-regulation. The authors found that more years spent in poverty was associated with lower levels of connectivity in both neural networks; however, this was more robust among young adults who received low levels of supportive parenting. The study did not show an effect of income on connectivity in the presence of high levels of positive and supportive parenting [124]. In a study that analyzed data from a prospective longitudinal study of emotion development showed that lower income-to-needs ratio at preschool age was associated with reduced connectivity between hippocampus and amygdala and several regions at school age, including the cortex, lingual gyrus, posterior cingulate, and putamen. This study included preschoolers 3-5 years of age selected from the St. Louis area. Participants were followed for up to 12 years. Individuals underwent annual behavioral assessments. Participants also underwent neuroimaging at school age to measure brain resting-state functional connectivity with the left and right hippocampus and amygdala. This study showed that a lower income-to-needs ratio predicted a greater connectivity between the left hippocampus and the right superior frontal cortex and between the right amygdala and the right lingual gyrus. As this study showed, brain functional connectivity mediated the relationship between SES and depression [125]. Thus, low SES predicts reduced connectivity between the amygdala and hippocampus with brain regions, including the lingual gyrus, superior frontal cortex, posterior cingulate, as well as putamen [125]. Social adversities have cumulative (additive) effects on brain structures and functions that govern emotion regulation [85] and memory [126]; however, these effects may differ across demographic groups [60]. In part, these are due to parents' health [127] and behaviors [128].
While our first finding documented a link between SES and working memory, our second suggested that this effect differed across demographic groups. Most of the literature on the effects of race, parental education, and income on memory have focused on additive rather than multiplicative effects of SES indicators and race. Our second finding can also be seen as a reflection of the MDRs. Many studies have shown more significant effects of SES on outcomes for White than Black American children [49,62,129]. For example, family SES has shown larger effects on ADHD [79], anxiety [71], aggression [45], tobacco dependence [45], school bonding [130], school performance [73,131], obesity [69], and health [68] for White than Black American children.
As a result of our second finding, parental education shows a more salient role in shaping White's impulsivity than Black American children [67]. As a result of this pattern, higher than expected risk of poor self-rated health, obesity, poor mental health, chronic disease, impulsivity, aggression, smoking, and low school performance are observed in high SES Black American children [69,77,79]. These patterns are also called MDRs and seem robust as they hold across SES indicators, outcomes, population groups, birth cohorts, age groups, and settings [36,37]. The findings observed in this analysis, however, did not support MDRs.
MDRs, differential effects of parental education across racial groups, maybe due to racial discrimination in high SES Black families. Racial and ethnic discrimination affect the amygdala's structure and function [150][151][152][153][154][155]. In a study in the US, 74 adults (43% women; 72% African American; 23% Hispanic; 32% homosexual/bisexual) reported their discrimination experience. The study also measured spontaneous amygdala activity and functional connectivity between the amygdala and other brain regions during resting-state functional magnetic resonance imaging (fMRI). In this study, greater experience of discrimination was associated with an increased level of spontaneous amygdala activity. Similarly, an increase in discrimination was associated with stronger functional connectivity between the amygdala and several neural regions such as the anterior insula, putamen, caudate, anterior cingulate, medial frontal gyrus. The most robust effect of discrimination was seen for the connectivity between the amygdala and thalamus [156]. As high SES, particularly high subjective SES, is a proxy of high not low discrimination [50][51][52][53][55][56][57][58]137], high SES Black American children still report lower than expected brain and behavior outcomes because of the effect of discrimination.
Differential effects of family SES indicators for Black and White families contribute to the transgenerational transmission of inequalities [45,[67][68][69][70]. Differential effects of SES mean the same level of SES may generate unequal outcomes for the next generation, which results in the reproduction of inequalities across generations. However, most of the previous studies on MDRs have relied on self-reported outcomes. Thus, the evidence lacked biological studies that test the differential effects of SES on children's brain imaging. This paper documented complex, non-linear, multiplicative effects of SES, and race on working memory.
The observed MDRs suggest that Black American children suffer from three jeopardies. The first risk is that they live in low-SES families. The second risk is that they have worse outcomes (working memory in this study). The third jeopardy is their SES shows a weaker impact on their brain development. The weakened effect of SES for Black children suggests that it is very difficult to improve the health outcomes and close the Black-White gaps. Policymakers should not expect drastic effects as a result of their interventions. These diminishing returns are likely to be due to unique stressors in Black people's lives across all SES levels.
It should be emphasized that we see race as a social factor (as a proxy of social status, treatment by society, access to the opportunity structure, interpersonal discrimination, environmental injustices, societal obstacles, and historical injustice) on how people are treated by society. As our results suggested, race alters the implications of family SES for working memory not because Blacks are inferior or different than Whites, but because society has historically oppressed them and continues to discriminate against them. All these injustices take a toll in terms of health and development.
Our paper was on complex and multiplicative effects of social determinants on 9-and 10-year-old American children. Working memory is closely associated with executive functioning and is mainly performed in higher cortical areas, especially the prefrontal cortex (PFC) [157][158][159]. It is recognized that the PFC is one of the last regions of the brain to mature [157]. During preadolescence, the increase in gray matter volume is observed, especially in the PFC, among other frontal lobe regions [157]. Furthermore, it is recognized that females' brains develop about, on average two years earlier than male brains, so they are more likely to have a late developing male brain than females [157].

Limitations
All studies have some methodological and conceptual limitations. This study, which was a secondary analysis of existing data, is not an exception to this rule. Our first limitation was a cross-sectional analysis. As a result, we can only conclude associations, not causal effects. Our second limitation was a lack of inclusion of many confounders such as psychopathologies, learning disabilities, or physical health. Third, all our SES measures were reported by parents. Some measurement bias should be expected in the measurement of SES in this study. Also, the sample was not balanced regarding race and SES. Racial groups were also not comparable in their SES. Finally, brain development is behind in male than female children. This study, however, did not explore race by sex differences in social determinants of working memory. Built on intersectionality, future research may explore how groups based on the intersection of race, sex, SES, and place differ in the effects of SDoH on memory function.

Conclusions
Among American children, high parental education and household income correlate with better working memory. However, the effect of parental education is unequal across racial groups, with the marginal return of parental education being smaller for Black families than White people. Income, however, generates a similar outcome for Black and White families. Finally, parental education generates less outcome than the differential income of highly educated Black and White families. These findings have policy solutions for achieving equality. First, the solution to racial gaps lies beyond closing the SES inequalities. We should address barriers that interfere with the SES from generating equal outcomes for Blacks. Equalizing income may be a more effective way of equalizing outcomes than equalizing education, because more processes can interfere with the return of education than income. More research is needed on how we can equalize Black and White families for the effect of SES on brain development. Social determinants' influence on children's brain development is complicated and multiplicative rather than simple and additive. Moderated mediation and mediated moderation models are more realistic than simple additive models. Funding: Author Funding: Assari received support from the following NIH grants: 2U54MD007598, U54 TR001627; CA201415-02, 5S21MD000103, R25 MD007610, 4P60MD006923, and 54MD008149. ABCD Funding: The ABCD Study is supported by the National Institutes of Health and additional federal partners under award numbers U01DA041022, U01DA041028, U01DA041048, U01DA041089, U01DA041106, U01DA041117, U01DA041120, U01DA041134, U01DA041148, U01DA041156, U01DA041174, U24DA041123, U24DA041147, U01DA041093, and U01DA041025. A full list of supporters is available at https://abcdstudy.org/federal-partners.html. A listing of participating sites and a complete listing of the study investigators can be found at https://abcdstudy.org/ Consortium_Members.pdf. ABCD consortium investigators designed and implemented the study and/or provided data but did not necessarily participate in analysis or writing of this report. This manuscript reflects the views of the authors and may not reflect the opinions or views of the NIH or ABCD consortium investigators. The ABCD data repository grows and changes over time. The current paper used the Curated Annual Release 2.0, also defined in NDA Study 634 (doi:10.15154/1503209).