Abstract
Previous research has provided evidence for the robust relation between maltreatment and the development of externalizing behavior, including aggression. However, less empirical attention has been given to the specific role of neglect. The current study aimed to examine the role of working memory in the association between early neglect and aggression in toddlerhood. Longitudinal data were collected from 89 infants and their biological mothers when the infant was approximately 12, 26, and 38 months old. History of neglect was assessed at 12 months using official Child Protective Service records. Working memory and mental development were assessed at 26 months. Aggression was measured using maternal report at 38 months. Data were analyzed using structural equation modeling, and mediation was tested using 95% asymmetric confidence intervals. Results indicated that infants who experienced neglect exhibited poorer working memory abilities, specifically spatial working memory, and higher rates of aggression in toddlerhood. In addition, spatial working memory mediated the relation between neglect and aggression, suggesting that this may be one promising target for intervention.
It has been well established that child maltreatment is both a pervasive and significant public health concern. Currently, it is estimated that nearly one in four children experience some form of child maltreatment in their lifetime, with many experiencing more than one type and more than one incident (D. Barnett, Manly, & Cicchetti, 1993; Centers for Disease Control, 2014; Finkelhor, Turner, Shattuck, & Hamby, 2013). Neglect remains overwhelmingly the most common form of child maltreatment in the United States, with the youngest children (birth to 1 year) being the most vulnerable (Child Welfare Information Gateway, 2017). Neglect in infancy is not an isolated experience but, instead, has a cascading effect on multiple developmental domains. That is, neglect experienced at the hands of a caregiver is likely to fundamentally impact a child’s social, emotional, cognitive, and neurobiological development (Cowell, Cicchetti, Rogosch, & Toth, 2015; Lee & Hoaken, 2007). Despite its prevalence and deleterious effects, there has been a dearth of research on the specific effect of neglect on child development (Manly, Oshri, Lynch, Herzog, & Wortel, 2013). Many researchers have called for a more careful investigation of this maltreatment subtype (Hildyard & Wolfe, 2002).
One of the main deleterious effects of child neglect is that it predisposes children for the development of externalizing behavior, specifically childhood aggression (Kim & Cicchetti, 2010; Kotch et al., 2008; Vachon, Krueger, Rogosch, & Cicchetti, 2015). Emergence of maladaptive behavior is not surprising, as lack of parental care and nurturance threatens the health and development of children (Rutter & Sroufe, 2000; Sameroff, 2000). The impact of childhood aggression is widespread and linked to a number of adverse outcomes across various domains including health, education, employment, and interpersonal relationships (Fergusson, Horwood, & Ridder, 2005; Ladd & Burgess, 2001; Tremblay et al., 2004). These negative effects often endure through adolescence and adulthood (Odgers et al., 2008). Childhood aggression that develops into violence and criminality in adolescence and adulthood results in high monetary costs for society (Cohen & Piquero, 2009).
One objective of this study is to further examine the relation between neglect in infancy and aggression in toddlerhood. Although children who are abused are often also neglected, national data on child abuse and neglect have shown that a large percentage of children who have been neglected were not found to be maltreated in any other way by the child welfare agency (U.S. Department of Health & Human Services, 2016). Whereas previous research has often looked at the combined impact of abuse and neglect together, making it difficult to determine the effect of each subtype individually (Weiss, Dodge, Bates, & Pettit, 1992), the prevalence of neglect only cases further emphasizes the importance of examining the effect of neglect alone on child development. In addition, there is a dearth of research examining the effect of neglect on the development of aggression in early childhood. Research has documented that these aggressive tendencies may emerge as early as preschool and persist over time (Broidy et al., 2003; Moffitt, 1993). In order to fully understand aggression as a developmental process, it is imperative that we examine this construct as it begins to develop. Finally, understanding the mechanisms through which neglect affects aggression warrants further research given the possible intervention implications. The current study builds upon and extends prior research by examining one potential mechanism—working memory—using longitudinal data in a high-risk sample.
Neglect and Aggression
Consistent with a developmental psychopathology framework, neglect during infancy and early childhood undermines the mastery of developmental tasks and decreases the probability that children will encounter subsequent opportunities to help them build the skills necessary for adaptive development, therefore propelling them on a maladaptive pathway (Cicchetti & Toth, 1995; Jaffee & Maikovich-Fong, 2011). This may result in a number of deleterious outcomes that could persist into adulthood (Arnow, 2004), including maladaptation across the biological and psychological domains of development (Cicchetti & Toth, 2005). Unlike other forms of maltreatment, which are more likely to be incident-specific, neglect is often chronic and is not easily identified as a specific incident (Hildyard & Wolfe, 2002). Chronicity of neglect can be particularly damaging to child development and has been associated with an increased likelihood of childhood aggression (Éthier, Lemelin, & Lacharité, 2004).
Examining this link in cross-sectional research, neglect is associated with aggression in youth (Manly, Kim, Rogosch, & Cicchetti, 2001) and adolescents (Hussey, Chang, & Kotch, 2006). Specifically, physical neglect has been linked to aggression in early and late childhood (Knutson, DeGarmo, Koeppl, & Reid, 2005; Manly et al., 2001). In examining the effect of neglect in infancy and toddlerhood, similar to the current study, threats to a child’s physical (e.g., failure to provide food) and psychological (e.g., verbally aggressive discipline) safety and security have also been linked to early childhood aggression (English, Thompson, Graham, & Briggs, 2005). One longitudinal study found that chronic neglect in childhood, specifically failure to provide, predicted aggression in adolescence even after controlling for the effects of other maltreatment subtypes (Logan-Greene & Jones, 2015). Additionally, a nationally representative, longitudinal study has examined this relation in a younger sample of children aged 4, 6, and 8 (Kotch et al., 2008). Results suggested that early neglect (prior to Age 2) may be a more robust precursor of childhood aggression than later neglect, or physical abuse at any age. The current study extends this research by attempting to elucidate a potential mechanism, working memory, through which neglect predisposes children to the development of aggression.
Although these results provide evidence of a clear link between neglect and aggression, there is a dearth in the number of studies that examine this relation longitudinally. Longitudinal research is essential to examining the cascading impact of early neglect on later child development. Even more scarce are longitudinal studies examining this relation in young children. This is concerning considering that behavior problems in maltreated children, including aggression, have been documented as early as toddlerhood (Manly et al., 2013; Toth, Cicchetti, Macfie, Rogosch, & Maughan, 2000). If we know that the youngest children are most vulnerable to neglect and that aggressive behavior begins in early childhood, then the relation between these constructs and their ongoing development should be examined early on. Not only will this allow us to better understand the relation between these constructs as they develop in early childhood, but this information can then inform our ability to intervene at an early age in order to prevent the development of maladaptive behavior, such as childhood aggression.
Neglect and Working Memory
Executive function (EF) is an umbrella term for top-down mental processes that assist in self-regulation (Diamond, 2013) and together are “utilized in planning, problem solving, and goal-directed activity” (Blair & Razza, 2007, p. 648). In a large-scale study examining the effect of both neglect and abuse on specific components of executive functioning, Cowell, Cicchetti, Rogosch, and Toth (2015) found that children with a history of maltreatment performed worse on measures of inhibitory control and working memory compared to nonmaltreated children. In particular, children maltreated during infancy exhibited significantly poorer skills in these domains compared to those maltreated during later developmental periods or not maltreated. These results suggest that maltreatment occurring during infancy, a period of exponential brain growth, disrupts normative cognitive development.
Drawing from a neuropsychological perspective, researchers highlight ways in which early stress and deprivation can result in structural and functional impairment in the brain. Specifically, development of the prefrontal cortex (PFC), which serves as the brain’s “air traffic control system” by supporting a wide range of EFs, is influenced by the environment during the first few years of life (Center on the Developing Child at Harvard University, 2012; Schore, 1996). When a child’s early environment is laden with stress and the caregiver is nonresponsive to that stress, such as in a neglectful home, the child may develop neuropsychological impairments in this brain region that place him or her at risk of antisocial outcomes, including aggression (Lynam & Henry, 2001; Schore, 1996). Indeed, there is robust evidence for abnormalities in the PFC region of the brain in aggressive and violent individuals (Davidson, Putnam, & Larson, 2000; Raine, 2002). Neuroimaging studies of adults and adolescents with histories of neglect reveal smaller PFC volumes compared to nonneglected individuals (Edmiston et al., 2011; Frodl, Reinhold, Koutsouleris, Reiser, & Meisenzahl, 2010).
Impairment to the PFC has been associated with deficits in working memory (Hanson et al., 2012; Kringelbach & Rolls, 2004). Working memory is one component of EF that emerges early in life and develops rapidly in the preschool period, acting as a building block for other functions (Zelazo & Carlson, 2012). Individual differences in these skills can be assessed as early as toddlerhood (Bernier, Carlson, & Whipple, 2010). Working memory is defined as one’s ability to hold information in mind and manipulate it (Diamond, 2013). This construct includes verbal working memory and nonverbal (visual-spatial) working memory. In addition, a further fractionation of nonverbal working memory suggests separate stores for visual and spatial information. Numerous studies support visual (also referred to as object or nonspatial) and spatial working memory as distinct constructs, including brain imagining studies of adults that indicate activation of different regions for each (Awh & Jonides, 2001; Smith et al., 1995). Additionally, differences in prefrontal cortical volume have accounted for the association between cumulative life stress and spatial working memory in adolescence (Hanson et al., 2012).
There is a paucity of research examining the link between neglect and working memory specifically. Much of the research has been conducted on children that have experienced extreme forms of deprivation and neglect in institutionalized settings. One study examined the effect of neglect and deprivation on internationally adopted children (Pollak et al., 2010). The length of time institutionalized had a significant impact on children’s cognitive abilities at 8 and 9 years old. Specifically, spatial working memory abilities were impacted. This effect has been replicated in other samples of postinstitutionalized children (Bos, Fox, Zeanah, & Nelson, 2009; Merz, McCall, Wright, & Luna, 2013). Although informative, the generalizability of these results to children experiencing familial neglect is in question. Therefore, further examination of the impact of familial neglect on the development of working memory is needed.
Working Memory and Aggression
A number of studies have found a link between EF more broadly and aggression (Ellis, Weiss, & Lochman, 2009; Hancock, Tapscott, & Hoaken, 2010; Hughes & Ensor, 2011; Moffitt & Henry, 1989). However, much of this research has focused primarily on adolescents and adults and has not examined the subcomponents of EF (e.g. working memory). Few studies have examined the relation between EF and aggression in younger children. In a sample of preschoolers, investigators found that poorer performance on an EF battery was associated with more aggressive behavior, even after controlling for attention problems (R. Barnett et al., 2001). This EF battery included measures of set shifting, verbal fluency, and working memory. While these findings greatly contribute to an already scarce body of literature, weaknesses in any of these EF subcomponents may be responsible for performance failure. Therefore, it is difficult to draw conclusions about the specific effect of individual EF components on aggression in this study.
Working memory is one component of EF that has been linked to aggression, although there is a dearth of research examining this association, particularly in early childhood. Séguin and colleagues have conducted a number of studies examining the link between working memory and aggression in older youth (Séguin, Boulerice, Harden, Tremblay, & Pihl, 1999; Séguin, Nagin, Assaad, & Tremblay, 2004). The authors argue that working memory encompasses a set of “in-real-time” cognitive process that are essential to various phases of problem-solving, such as shifting perspectives on a problem, strategizing while anticipating consequences, and executing steps of a plan. Many of these necessary problem-solving abilities appear to be impaired in physically aggressive individuals (Séguin et al., 2004). Results from these studies reveal an association between working memory and aggression, even after controlling for Attention Deficit Hyperactivity Disorder, general memory, and IQ. Although these studies contribute to our understanding of the relation between working memory and aggression, the sample includes only males, begging further investigation into this relation in females. Additional research is also needed to further examine the longitudinal relationship between working memory and aggression in early childhood.
Based on this mounting evidence, we predict that working memory plays a role in the association between early neglect and aggression. The present study extends the literature by examining working memory as a mechanism through which infant neglect leads to aggression in toddlerhood. Based on the extant literature, the following hypotheses were posited: Infants from neglecting families will exhibit poorer spatial and nonspatial working memory performance and higher rates of aggression, compared to infants from nonneglecting families. Children who performed poorer on the working memory tasks will exhibit higher levels of aggression in toddlerhood. Spatial and nonspatial working memory performance will mediate the effect between neglect and aggression, illustrating that this relation is partially explained via working memory skills.
Method
Participants
The data for this study were drawn from a larger randomized clinical trial evaluating the efficacy of two preventive interventions for maltreated children designed to optimize mother–infant relationships and improve parenting (see Cicchetti, Rogosch, & Toth, 2006). Infants with and without histories of maltreatment and their biological mothers living in an urban city in the United States were recruited for the efficacy study. The institutional review board at the university approved all study procedures, and mothers provided informed consent for their and their child’s participation in the study. Mothers were paid in cash and children received a small toy for their participation in the study.
To eliminate the confounding intervention effect, only nonmaltreating families and maltreating families who were randomized to the community standard control condition were utilized for the current study. Families in the community standard condition did not receive enhanced services; however, they were offered services that were typically available to maltreating families in the community. For the purpose of this study, the authors were specifically interested in the role of neglect as it is the most prevalent maltreatment subtype in infancy (Child Welfare Information Gateway, 2017). Neglect was the most predominant form of maltreatment in our sample; therefore, the following analyses were limited to neglecting families. Infants with an individual history of physical (8%) and/or sexual abuse (0%) were excluded from the current study.
The neglect group consisted of 45 infants (18 boys, 27 girls) and their biological mothers. The nonneglect comparison group consisted of 44 infants (24 boys, 20 girls) and their biological mothers. At baseline, infants were on average 12.7 months of age (SD = .68). Mothers ranged from 18 to 41 years of age (M = 26.76; SD = 5.71). The majority of mothers were of minority race/ethnicity, 51% identified as African American, 24% identified as Caucasian, 12% identified as Latino or Hispanic, and 13% as other. In terms of marital status, 7% were married, and another 21% were living with a significant other as though married. Overall, 94% of families reported receiving monetary public assistance. At baseline, the two groups did not differ on any of these demographics.
Recruitment
This study used a Department of Human Services (DHS) recruitment liaison in order to recruit 12-month-old infants with histories of maltreatment and their mothers. As an employee of DHS, the liaison was able to access DHS Child Protective Service (CPS) and preventive service records to identify all infants known to have been maltreated and/or who were living in maltreating families with their biological mothers. Inclusion criteria for the neglect group analyzed in this study required at least one substantiated instance of neglect that occurred with the infant specifically or within the family with which they were living. Families were excluded if the child had a history of or was currently in foster care placement due to the potential role of changed environmental circumstances. In order to ensure that families did not feel coerced to participate in the study, we did not recruit families that were under current investigation by CPS. The DHS liaison contacted eligible families and explained the project to the mothers. Mothers who were interested in participating signed a name release giving the liaison permission to provide the family’s contact information to project staff. During the initial contact with project staff, the mothers provided informed consent for their and their infant’s participation. Mothers also signed a release allowing project staff to have access to the family’s DHS records regarding their CPS and preventive service involvement.
Once DHS records had been obtained, they were independently coded by trained research coders using the Maltreatment Classification System (MCS; D. Barnett et al., 1993). The MCS comprehensively classifies all forms of maltreatment that occur at the individual and family level. Other investigators have demonstrated that the MCS is reliable and valid way to comprehensively classify child maltreatment (Bolger, Patterson, & Kupersmidt, 1998; Dubowitz et al., 2005; Manly, 2005). Among those neglected in this sample, 62.2% of the infants were direct targets of neglect while the other 37.8% of the cases identified a sibling as the target. For 98% of neglecting families, mothers were named as a perpetrator of neglect, although additional perpetrators may have been present.
Because of the high prevalence of poverty in the neglect group, to recruit a demographically comparable sample, comparison families were recruited based on receipt of Temporary Assistance to Needy Families (TANF). The DHS liaison identified families with 12-month-old infants from the TANF rolls and screened them to ensure that there were no substantiated cases of maltreatment in the family’s record. The DHS liaison then contacted eligible families, explained the study, and for interested mothers, obtained a name release form. During the initial contact with project staff, the mothers provided informed consent for their and their infant’s participation. In addition, project staff obtained permission for all DHS records pertaining to the family to be reviewed. These records were reviewed at each time point to ensure that there was no existence of documented child maltreatment.
Participant Retention
All families were actively recruited for follow-up assessments at approximately 26 months (T2) and 38 months (T3) of age. Nevertheless, attrition did occur. The range of missing data on endogenous variables is 24.8–26.4%. Although there was attrition in both the neglecting and nonneglecting families, the rate of attrition through T3 was significantly higher among neglecting families (35%) than nonneglecting families (15%), χ2(1) = 5.736, p < .05. The final sample size was 89 (neglect families = 45). Families who were retained versus lost during follow-up visits were compared on baseline demographics. The groups did not differ on gender or age of child, nor maternal age, race/ethnicity, marital status, education, or SES.
Procedures
At baseline, research assistants administered a battery of demographic and research interviews and questionnaires to the mothers. When the toddler was approximately 26 months, two working memory tasks were conducted in the lab with the child: three boxes stationary and three boxes scrambled. These tasks were videotaped and reviewed by a research assistant to ensure adherence to task protocol and the child’s ability to engage in the task. Research assistants also assessed toddler’s mental development at this time point by administering the Bayley-II. Finally, 1 year later when the toddler was approximately 38 months, research assistants returned to the home to administer various demographic and research measures to the mother, including an assessment of toddler aggression.
Measures
Maltreatment Classification System (MCS)
The MCS (D. Barnett et al., 1993) was used to delineate diverse features of maltreatment that each infant experienced. The MCS utilizes DHS records of child maltreatment to comprehensively code all information available from the record in order to make independent determinations of maltreatment experiences. Specifically, the MCS codes all incidents that have been documented in the family’s DHS records and also designates subtype(s) of maltreatment including neglect, emotional maltreatment, physical abuse, and sexual abuse. The MCS defines neglect as the failure to provide for the child’s basic physical needs, including food, clothing, shelter, and medical attention. This subtype also includes lack of adequate supervision, moral-legal neglect, and educational neglect. For the current study, neglect was a binary variable, in which 0 was coded if there was no history of neglect and 1 was coded if there was a history of neglect. Coders included research assistants, doctoral students, and clinical psychologists trained in the MCS. Reliability was previously reported in a study of the overall sample (weighted ks ranging from .86 to .98; Manly et al., 2001).
Bayley Scales of Infant Development (Bayley-II)
Mental development (Mental Development Index [MDI]) was measured at age 2 using the Bayley-II in order to control for the potentially confounding effect of general mental capacity on working memory (Bayley, 1993; Engle, Tuholski, Laughlin, & Conway, 1999). The MDI of the Bayley-II broadly assesses cognitive and language development in children 1–42 months old. This is a standardized measure with index scores ranging from 40 to 160 with a mean of 100 and a standard deviation of 15. It has shown high test–retest reliability (r = .91) and excellent percentage of agreement for classification (96.50%) with children of a similar age. This assessment was administered when the toddler was approximately 26 months old—the same age at which the working memory measures were administered. Toddlers’ mental development did not differ based on neglect status, t(87) = .907, p > .05; neglect group: M = 82.80, SD = 15.39; comparison group: M = 85.50, SD = 12.52.
Three boxes stationary
This task followed the procedure outlined by Diamond, Prevor, Callender, and Druin (1997) and has proven to be a valid measure of spatial working memory (Diamond, Prevor, Callender, & Druin 1997). This task requires holding information in one’s mind, while inhibiting a proponent response to reach to a previously correct box. At the beginning of the task, three boxes were placed in front of the child, each with a different color and accompanying shape. Each box was baited with a reward of the child’s choosing while the child watched. The lids were then closed and the boxes were pushed toward the child, as they were encouraged to find which contained a reward. After the child retrieved the reward or an empty box was opened, the experimenter placed the boxes out of reach. A cardboard partition was then placed between the child and the boxes while a 5-s delay was imposed. Following the delay, the boxes were again presented to the child. The task continued until the child had found all three rewards, five consecutive errors occurred, or following 15 attempts. The goal of the task was to find all three rewards in the least number of reaches. The child’s performance was measured using an efficiency ratio (scale 0–1), in which the number of rewards retrieved was divided by the total number of reaches (Griffith, Pennington, Wehner, & Rogers, 1999). Previous research has found this task to exhibit high (>.90) intercoder reliability with children of a similar age (Diamond et al., 1997).
Three boxes scrambled
This task also followed the procedure outlined by Diamond et al. (1997) and has proven to be a valid measure of nonspatial working memory. This task is identical to the task described above, except instead of the boxes staying stationary after each reach, they were scrambled while being hidden behind the partition. In addition, the shapes and colors of the boxes were different from those used in the stationary task. In the stationary version of this task, the child may maintain a place keeper in mind to guide reaching, whereas the scrambled version requires the child to maintain more specific information regarding the visual appearance of each box (Griffith et al., 1999), therefore assessing object memory instead of spatial memory (Garon, Bryson, & Smith, 2008). Again, the task continued until the child had found all three rewards or five consecutive errors occurred. Three trials of up to 20 attempts were administered. Because there were three separate trials for this task, the child’s performance was measured using an average efficiency ratio (scale 0–1). In this case, the number of rewards retrieved was divided by the total number of reaches for each trial, and then, an average was taken of the three efficiency scores (Griffith et al., 1999). Previous research has found this task to exhibit high (>.90) intercoder reliability when conducted with children of a similar age (Diamond et al., 1997). The goal of the task remained to find all three rewards in the least number of reaches.
Child Behavior Checklist/2–3 (CBCL)
The CBCL was used to assess aggression by capturing maternal report of children’s behavior problems at 38 months old (Achenbach, 1992). The CBCL is a widely used and validated instrument to assess symptomatology by mothers (Achenbach, 2017). It contains 100 total items rated for frequency on a 3-point scale (0 = not true to 2 = very true or often true). For the purpose of this study, the Aggression subscale of this measure was examined, which includes items such as “easily frustrated” and “hits others.” This subscale had good internal consistency (∝ = .91). The T-score provided by Achenbach’s scoring system was used for the aforementioned scale. These scores have been standardized, permitting comparisons between genders.
Data Analytic Plan
In the data preparation phase, the following were addressed using SPSS (Version 24): (a) missing data (b) outliers, and (c) nonnormality of indicators. In addition, preliminary analyses were conducted in SPSS. These analyses included correlations among study variables as well as group differences on all endogenous variables and covariates. To address the main aims of this study, structural equation modeling (SEM) was conducted using Mplus (Version 7.1.4) (Muthén & Muthén, 1998–2015). Missing data for endogenous variables were handled using the Full Information Maximum Likelihood method and the Maximum Likelihood estimator was used. Global fit for the structural model was evaluated by examining the χ2 index, the root mean square error of approximation (RMSEA), the comparative fit index (CFI), and standardized root mean square residual (SRMR). Evidence of good model fit was determined using the following criteria: A nonsignificant χ2, RMSEA < .06, CFI > .95, and an SRMR < .08 (Hu & Bentler, 1999). Mediation was tested using 95% asymmetric confidence intervals (CIs) via RMediation (Tofighi & MacKinnon, 2011), which allows for a more accurate estimation in the event that the mediated effect is not normally distributed. CI that does not include the value zero was considered statistically significant. Finally, we calculated the proportion of total effect mediated.
Results
Preliminary Analyses
Tables 1 and 2 provide the zero-order correlations, means, standard deviations, and group differences among study variables. As illustrated in Table 1, poorer performance on the three boxes stationary task at 26 months was related to higher levels of aggression (r = −.33, p < .01) at 38 months old. However, performance on the three boxes scrambled task was not significantly associated with aggression (r = −.04, p > .05). A lower score on the MDI was related to poorer performance on the three boxes stationary task (r = .25, p < .05) and the three boxes scrambled task (r = .36, p < .01). A lower MDI score was also related to higher levels of aggression (r = −.26, p < .05). Finally, performance on the three boxes stationary task was associated to performance on the three boxes scrambled task (r = .25, p < .05). Table 2 illustrates the group differences between the neglect and comparison group on key study variables. Results indicated that children from neglecting families demonstrated higher levels of aggression, t(82) = −2.759, p < .01, and poorer performance on the three boxes stationary task, t(85) = 2.500, p < .05, compared to comparison children. There were no significant differences between groups on the three boxes scrambled task or the MDI.
Bivariate Correlations Among Study Variables.
Note. Higher scores of aggression indicate higher rates of aggressive behavior. Mental development and working memory were coded, such that higher scores indicate better performance on the task.
†p < .10. *p < .05. **p < .01. ***p < .001.
Means and Standard Deviations.
Note. Higher scores of aggression indicate higher rates of aggressive behavior. Three boxes tasks were scored on a scale of 0–1. Cognitive functioning tasks were coded, such that higher scores indicate better performance on these tasks.
Mediation Analyses
An SEM was conducted to examine whether working memory at 26 months mediated the relationship between neglect prior to baseline and aggression at 38 months. Neglect was entered as a dichotomous exogenous variable. Both spatial working memory (three boxes stationary) and nonspatial working memory (three boxes scrambled) were modeled as correlated mediators. Aggression was modeled as the outcome variable. In addition, mental development was modeled as a covariate, predicting performance on both working memory tasks. Gender was originally included as a covariate; however, it did not significantly affect the mediator (working memory) or the outcome variable (aggression), nor did it change the pattern of results when included in the model. Therefore, gender was excluded for the remainder of the analyses. The model was a good fit to the data χ2(1) = 2.480, p > .05, RMSEA = .129, CFI = .957, SRMR = .035. Although the RMSEA is slightly higher than the cutoff outlined by Hu and Bentler (1999), research has found that this cutoff may falsely indicate poor fit for models with small degrees of freedom, particularly those with a small sample sizes (Kenny, Kaniskan, & McCoach, 2015). Results of the SEM are illustrated in Figure 1.
Results of structural equation modeling. tp < .10, *p < .05, **p < .01, ***p < .001. History of neglect was coded 1 = neglect present and 0 = no history of maltreatment. Mental development and working memory were coded, such that higher scores indicate better performance on the task. Higher scores of aggression indicate higher rates of aggressive behavior.
The results indicated that neglect during infancy was a significant predictor of higher levels of aggression in toddlerhood, b = .215, SE = .103, p < .05. In addition, infants raised in a neglecting family demonstrated poorer spatial working memory, as exhibited by performance on the three boxes stationary task, b = −.241, SE = .098, p < .05. Those with poorer spatial working memory skills then exhibited higher levels of aggression 12 months later, b = −.286, SE = .108, p < .01. However, when examining nonspatial working memory, as evidenced by the three boxes scrambled task, children in neglecting families did not differ significantly from those in nonneglecting families, b = .056, SE = .100, p > .05, and performance on this task did not significantly predict aggression, b = .010, SE = .107, p > .05. These results suggest a differential effect of neglect on the development of these skills. 1 In addition, lower mental development scores predicted poorer spatial working memory skills, b = .229, SE = .098, p < .05, and nonspatial working memory skills, b = .365, SE = .093, p < .001. To test mediation, we used 95% asymmetric CIs via RMediation (Tofighi & MacKinnon, 2011). Results indicated a significant indirect effect of infancy neglect on toddler aggression via spatial working memory at 26 months of age, 95% CI [0.076, 2.434]. 2 These results are indicative of partial mediation. In order to further understand the magnitude of the mediated effect, we calculated the proportion of total effect mediated. The mediated effect was found to explain approximately 39% of the total effect of neglect on aggression.
Discussion
The current study examined the role of working memory in the association between early neglect and aggression in toddlerhood. These results suggest that the development of working memory, specifically spatial working memory, is impeded by early neglect. This impediment appears to be an underlying mechanism through which early neglect leads to higher rates of aggression in toddlerhood. Although these rates of aggression may not have yet reached a clinical level for many of the toddlers in our sample, it is important to remember that these children are young and even slight elevations may indicate an increased risk for development of a full blown psychiatric disorder later in life. In order to prevent this, early detection of elevations in aggression, coupled with known risk factors (e.g., neglect; working memory impairment), is imperative for intervention efforts.
Although many of our results were consistent with our hypotheses, we did not find the expected outcome for both measures of working memory. There are a few plausible explanations. First, although both three boxes tasks generally measure working memory, they assess two slightly different components (spatial vs. nonspatial working memory). As outlined in the literature reviewed above, nonverbal (visual-spatial) working memory can be further broken down into two components: spatial working memory and nonspatial (visual; object) working memory (Awh & Jonides, 2001). Research with adults suggests that these two types of working memory recruit from slightly different brain regions (Smith et al., 1995). In the Smith et al. (1995) study, a task of spatial working memory activated right-hemisphere regions of the brain, while a nonspatial working memory task activated left-hemisphere regions. Of particular interest, the prefrontal region of the brain was activated for the spatial but not the nonspatial working memory task. This is the same region of the brain that has been found to be impacted by early neglect (Edmiston et al., 2011) and to be associated with antisocial and aggressive behavior later in life (Raine, 2002). Based on previous research, working memory abilities may be differentially impacted by neglect, as suggested by these results. Further research is needed to differentiate these specific working memory abilities in relation to neglect and aggression.
Alternatively, significant results might have only been found for the three boxes stationary task if it was more developmentally appropriate for this population. Three boxes stationary and three boxes scrambled are nearly identical tasks; however, in the scrambled version, toddlers cannot rely on spatial location and need to additionally track where the opened box has been moved to; therefore, increasing the cognitive load of the task. Although both of these tasks have been validated for use in toddlers as young as 15 months, it appears that many of the toddlers in our sample struggled significantly with the scrambled version. Diamond et al. (1997) published results on the three boxes scrambled task for a sample of normally developing toddlers. When comparing performance, the average efficiency score was .86 for toddlers (Age 27 months) in Diamond’s sample, and .60 for our sample, with 90% of toddlers scoring below the .86 score. Therefore, it appears that a majority of the toddlers in our study were performing significantly worse on this task compared to a sample of normally developing toddlers. However, it should be noted that Diamond’s sample was relatively small (n = 25). In addition, the correlation between the two working memory tasks in our sample was weaker than expected, based on previous research (Jenkins & Berthier, 2014).
One reason for this could be the demographic composition of our sample, which is composed of families characterized by poverty, racial/ethnic diversity, and trauma exposure. Research has found a strong association between poverty and poverty-related stressors and lower EF due to higher allostatic load (Blair, Raver, Granger, Mills-Koonce, & Hibel, 2011; Willoughby, Kupersmidt, Voegler-Lee, & Bryant, 2011). There are SES disparities specifically in working memory, suggesting that children raised in low SES families show a relative weakness in this area, compared to those raised in high SES families (Farah et al., 2006). As specified above, the toddlers in our overall sample exhibited below average mental development. In fact, more than half of our participants fell more than 1 SD below the mean score for the general population. Therefore, it is possible that their delayed mental development made the more difficult task (three boxes scrambled) developmentally challenging.
The current study uniquely contributes to the literature by examining the specific role of neglect on the development of early aggression. The prominence of neglect, particularly in infancy, necessitates the examination of its impact on the development of maladaptive behavior. Although we examine neglect as a single construct in this study due to available data and statistical constraints, it is important to recognize that it is in fact a multifaceted construct. There are a multitude of experiences that are classified as neglect, including physical or supervisory neglect, psychological neglect, medical neglect, and educational neglect. Future research could examine whether specific subtypes of neglect are more strongly associated with aggression. This study begins to fill a gap in the literature by examining working memory as one potential neuropsychological explanation for the relation between neglect and aggressive behavior in early childhood. Additional research is necessary to further understand the role of working memory. Specifically, future research should look to replicate these findings with a larger battery of working memory tasks. A final strength of the study is the diversity of the sample which allows the results to generalize to similarly diverse, low-income populations.
Despite the strengths of the study and its contributions to the literature, some methodological limitations should be addressed. First, aggression was only measured as an outcome at the last time point. For this reason, midpoint levels of aggression could not be controlled for. Ideally, aggression would have also been measured at 26 months and included in the SEM in order to provide stronger evidence for the inferred direction of effect. Also, the reliance on maternal report of toddler aggression is a potential limitation. Corroborating maternal report with observational data may provide a more reliable depiction of the child’s aggressive behavior.
The conclusions drawn from this study highlight many future directions for research to proceed. One possibility would be to follow these children over a longer period of time; specifically, past the normative decline in externalizing behavior. Previous research on longitudinal profiles of externalizing behavior, including aggression, has found that children typically exhibit higher levels of externalizing behavior around Age 2, followed by a normative decline by Age 4 or 5. However, those who continue to exhibit above average levels of externalizing behavior at this age are considered part of a “chronic-clinical profile” and are at risk of other negative outcomes, such as violent delinquency in adolescence (Broidy et al., 2003). Therefore, a study that followed children past this normative decline could differentiate those that had short-term versus potentially long-term aggressive tendencies. When childhood aggression develops into criminality in adolescence and adulthood, there is high monetary costs for society; successful intervention in early childhood can save society millions of dollars per youth (Cohen & Piquero, 2009) and reduce the pain and suffering that these behaviors cause. A clearer understanding of the role of working memory in the relation between neglect and aggression can inform future intervention and prevention efforts. Future research could examine the effect of an intervention designed to enhance working memory on the development of aggression in the context of neglect.
The results of the current study suggest that working memory, specifically spatial working memory, is one underlying mechanism through which early neglect may lead to later aggression. Although a novel contribution to an already scarce literature, further research is needed to provide more robust evidence and a deeper understanding for the relations examined in this study. Despite the unfortunate prevalence of neglect in early childhood, a better understanding of how neglect affects early cognitive development can aid our efforts in preventing long-term maladaptive outcomes.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by Spunk Fund, Inc., National Institute of Mental Health (MH54643), and Administration of Children, Youth, and Families.