Abstract
Executive functioning (EF) is key to students’ school and lifelong success and reflects both genetic predisposition and sensitivity to negative and positive experiences. Yet there is less available literature investigating the relationship between typical experiences within school environments and student EF development. This is unfortunate, as school environments are potentially more malleable than home- or community-based factors. Thus, the purpose of this article is to present a systematic review of the literature from 2000 to 2017 to understand how school-, classroom-, and dyadic-level (teacher–student and peer–student) experiences relate to student EF development. Across 20 studies, we found that classroom emotional support and teacher–student conflict were the most consistent predictors of student EF development, with emerging support for school-level and peer-level variables. We discuss findings in relation to school-based inhibitors and facilitators of student EF and provide implications for education research and practice.
Keywords
Executive functioning (EF) refers to neurocognitive processes that enable individuals to store and use information, control impulses, and switch gears to support goal-directed behavior (Best & Miller, 2010; Miyake et al., 2000). These top-down cognitive processes are essential for self-regulation, problem solving, and decision making—skills crucial for performance in school and life (Diamond, 2013; Rueda, Posner, & Rothbart, 2005; Zelazo & Müller, 2002, 2014). Extant research indicates that EF provides the foundation for children’s cognitive, social, and emotional development (Best, Miller, & Jones, 2009; Diamond, 2013; Zelazo & Müller, 2014), and it is a significant predictor of student achievement (Ahmed, Tang, Waters, & Davis-Kean, 2018). Conversely, EF deficits have been linked to maladaptive behaviors (Ellis, Weiss, & Lochman, 2009), mental health and neurological disorders (Banich et al., 2007; Kofler et al., 2011; Kyte, Goodyer, & Sahakian, 2005), and academic difficulties (Booth, Boyle, & Kelly, 2010; Gathercole & Pickering, 2000; Toll, Van der Ven, Kroesbergen, & Van Luit, 2011). As such, EF skills are vital for children’s short- and long-term success.
Individual EF reflects a genetic predisposition (Miyake & Friedman, 2012), but it is also sensitive to negative and positive life experiences (Zelazo, Blair, & Willoughby, 2016). Although EF is highly heritable, Friedman et al. (2008) caution against assuming that EF is fixed and cannot be influenced by environment. An established research base indicates that EF is malleable to environmental conditions (Diamond & Lee, 2011; Melby-Lervag & Hulme 2013). For instance, negative experiences that chronically activate the body’s stress response system have a profound and adverse impact on EF and the brain’s prefrontal cortex (Arnsten, 2009; Lupien, Maheu, Tu, Fiocco, & Schramek, 2007). Persistent or extreme stressors in a child’s home or community (e.g., poverty, violence exposure, abuse, neglect; Irigaray et al., 2013; Pechtel & Pizzagalli, 2011; Polak, Witteveen, Reitsma, & Olff, 2012) also adversely affect EF development. Conversely, resources (e.g., financial security; Hackman & Farah, 2009; Nesbitt, Baker-Ward, & Willoughby, 2013) and supportive relationships (e.g., positive parenting; Blair et al., 2011; Rhoades, Greenberg, Lanza, & Blair, 2011) can foster EF maturation and ameliorate the negative effects of stressors (Shonkoff, 2010).
Research on the impact of experiences outside of school on EF maturation is fairly well established (Irigaray et al., 2013; Pechtel & Pizzagalli, 2011), yet less research has examined how experiences in school shape EF development. Schools are a key leverage point by which society is able to affect student outcomes (e.g., Chetty, Friedman, & Rockoff, 2011). Thus, examining how school-based factors relate to EF development is crucial in fostering student success. In this article, we first highlight the role EF plays in student behavior and academic outcomes. Second, we briefly discuss what is currently known about effects of environmental experiences outside of school on EF development. Last, we present findings and implications from our systematic review of extant research on the association between environmental and interpersonal experiences in schools (i.e., at the school, classroom, dyadic, and individual teacher characteristic level) and student EF maturation.
Executive Function as a Foundation for Behavioral and Academic Outcomes
Despite variability in definitions, researchers generally describe three core EFs: (a) working memory, the theoretical cognitive structure responsible for the process of actively storing, maintaining, and manipulating information over brief periods of time (Miyake et al., 2000); (b) inhibitory control, the ability to suppress prepotent or dominant responses in order to display the subdominant (or desirable) response (Miyake et al., 2000); and (c) cognitive flexibility, the ability to adapt dynamically to changing task demands or contexts (Best & Miller, 2010; Miyake et al., 2000). These core EFs facilitate learning and social interactions. To illustrate, a student uses inhibitory control to wait to be called on (i.e., inhibit prepotent attentional or behavioral responses). Another student may use working memory to remember events in a story when answering reading comprehension questions. A student uses cognitive flexibility to switch tasks from reviewing multiplication facts to starting a math test (i.e., switch between mental sets). Due to the interrelated nature of EF, individuals may rely on multiple EFs when engaging in tasks. For instance, a student must use inhibitory control to disregard irrelevant information or distractions from peers while recalling from working memory the exact directions given by the teacher to complete a writing assignment.
Moreover, EF skills underlie self-regulation and other higher order skills (e.g., organization, planning, problem solving; Diamond, 2013; Miyake & Friedman, 2012). For example, a student in conflict with a peer must determine the best solution to resolve that conflict and then adjust in response to changing circumstances as that plan unfolds. Although the terms EF and self-regulation have been used interchangeably, Nigg (2017) argues that these two concepts are distinct from each other. Self-regulation has been primarily defined as a self-directive process where one’s own internal thoughts, feelings, and actions are managed, modulated, and cyclically adapted to the attainment of personal goals (Zimmerman & Schunk, 2011). Moreover, self-regulation involves continuously monitoring progress, checking outcomes, and changing them as necessary through self-reflection and self-evaluation. As a set of cognitive capacities, EF is required for successful self-regulation (Hofmann, Schmeichel, & Baddeley, 2012).
Thus, EF abilities predict student success in school and across their lifetime. For instance, longitudinal studies demonstrate that high EF development is associated with school readiness (Duncan et al., 2007; McClelland et al., 2007) and reading and math achievement (Ahmed et al., 2018; Best, Miller, & Naglieri, 2011; Blair & Razza, 2007; Bull, Espy, & Wiebe, 2008), as well as social adjustment, mental and physical health, and financial success (Moffitt et al., 2011). In contrast, students with poor EF are at risk of low academic achievement (Booth et al., 2010; Gathercole & Pickering, 2000; Toll et al., 2011), mental health problems (Zelazo & Müller, 2002, 2014), and school dropout (Fitzpatrick, Archambault, Janosz, & Pagani, 2015).
Some researchers further distinguish between cool and hot EF to highlight interactions between top-down processes (i.e., cognitive skills) and bottom-up influences (e.g., emotions and stress). For instance, Zelazo and Carlson (2012) applied the term cool EF to describe EF skills used in situations without strong emotional valence. These skills tend to be associated with cognitive problem solving and effortful control and are important to student academic achievement. Conversely, hot EF describes EF skills used in situations with strong emotional valence. These skills are related to affective decision making, delayed gratification, and down-regulated emotional responses, which are foundational to emotional self-regulation (Brock, Rimm-Kaufman, Nathanson, & Grimm, 2009; Zelazo & Carlson, 2012).
Executive Function Development and Environmental Experiences
Individual EF maturation follows a common developmental trajectory. The foundational components of EF begin to develop early in life, with pivotal maturation periods during preschool and middle school years (Best & Miller, 2010; Carlson & Wang, 2007). Although EF domains are interrelated, EF domain development is not uniform (Best & Miller, 2010). For example, inhibitory control is purported to develop first, with marked improvements during preschool years. Working memory shows gradual linear development from preschool to adolescence (Best & Miller, 2010). Last, cognitive flexibility develops gradually between ages 3 and 5 and continues to mature into adolescence (Huizinga, Dolan, & van der Molen, 2006).
Individual variability in EF can be influenced by stress and other life experiences, which can either facilitate or impede EF development (Hughes, 2011; Zelazo et al., 2016). Some stress is essential to functioning; however, uncontrollable, persistent, or extreme stressors can have a detrimental impact on EF maturation (e.g., Evans & Fuller-Rowell, 2013; Irigaray et al., 2013; Polak et al., 2012), particularly during the active EF maturation periods of childhood and adolescence (Hughes, 2011; Zelazo et al., 2016). For example, there is an irrefutable connection between child abuse, neglect, social deprivation, disaster, or violence exposure and individual EF maturation (DePrince, Weinzierl, & Combs, 2009; Irigaray et al., 2013; Nolin & Ethier, 2007; Pechtel & Pizzagalli, 2011). Even typical stressors in the home and community appear to have a negative effect. For instance, despite inconsistencies by race/ethnicity and gender, several researchers have found that poor parenting (e.g., harsh, controlling) is associated with low EF development in children (e.g., Bernier, Carlson, Deschenes, & Matte-Gagne, 2012; Cuevas et al., 2014; Family Life Project Key Investigators [FLPKI], 2013). Furthermore, researchers have consistently found that exposure to poverty and financial hardships is predictive of lower EF skills (Evans & Fuller-Rowell, 2013; Hackman & Farah, 2009), likely due to the systemic stressors in low-income communities.
Positive experiences, on the other hand, play a protective or facilitative role in children’s EF development. Quality home environments, characterized by access to learning materials, family companionship, and a warm emotional climate, promote children’s EF development (Clark et al., 2013; Rhoades et al., 2011). Positive parenting (e.g., emotionally warm, sensitive), in particular, has been found to predict higher child EF skills (Blair et al., 2011; Rhoades et al., 2011) and to moderate the adverse effects of stress on EF development (Doan & Evans, 2011). Specific aspects of parenting, such as scaffolding, play a unique and significant role in fostering children’s EF skills (Lengua et al., 2014). Thus, experiences in the home and community are crucial in shaping EF development.
Students spend a significant proportion of their day in school, yet less is known about school-based factors that contribute to student EF development. One recent meta-analysis by Vandenbroucke, Spilt, Verschueren, Piccinin, and Baeyens (2018) found that overall teacher–student interactions were related to overall EF, inhibition, and working memory but not cognitive flexibility. Although their findings highlight the importance of interpersonal relationships for EF development, the study had several limitations, which we address in the present review. First, the authors only included participants up to age 12. EF skills continue to develop through late adolescence; thus, it is essential to examine EF development as students enter middle and high school. Second, the authors included non-peer-reviewed sources. We believe it is important to include high-quality studies that have undergone a rigorous review process in order to draw conclusions, especially given that this area of research does not show evidence of publication biases (Vandenbroucke et al., 2018). Third, the authors did not include studies before 2009, although studies were published before that year (e.g., NICHD Early Child Care Research Network, 2005). Last, their review focused only on teacher–student interactions, ignoring other relevant school-related factors that may contribute to EF development, such as overall school quality, teacher characteristics, and peer relationships. Given the important role schools play in fostering students’ success, a more comprehensive review of how school-based factors relate to children’s EF development is warranted.
Purpose
Because schools are a primary setting with potential to play a significant role in student EF development, the purpose of the review is to examine relationships among school-, classroom-, and dyadic-level environmental and interpersonal experiences and students’ EF development, based on studies conducted from 2000 to 2017. We include studies on the role of school quality, classroom quality, teacher interactions, peer interactions, and teacher characteristics in facilitating or inhibiting EF maturation across the school years. We seek to identify gaps in the literature and offer implications for research and practice.
Method
We conducted a systematic search to identify studies examining relationships between school-based factors and students’ EF development. The following sections explain our inclusion and exclusion criteria, search procedures, and strategies for analyzing the extant studies.
Inclusion and Exclusion Criteria
We included studies written in English in peer-reviewed journals from 2000 to 2017. EF research in education was limited before 2000, as the seminal NICHD Early Child Care Research Network (2005) study was the catalyst for much research in this area; thus, we did not seek studies from before 2000. Because national policy contexts are unlikely to change the nature of relationships between school factors and children’s EF development, we did not limit our findings to the United States but rather included studies from around the globe. To be included, studies’ primary participants needed to be aged 3 to 18 years, both to ensure the inclusion of participants during critical EF developmental periods and to ensure findings were relevant to understanding school-based factors. Studies also needed to include a school-based environmental or interpersonal factor as a predictor or moderator, as well as a researcher-administered measure of EF (direct assessments of working memory, inhibitory control, and/or cognitive flexibility) as a dependent variable. Because of the variety of terms used to describe EF, we included studies measuring any of the core EF skills, as well as studies that measured a closely related construct using a tool that has been validated as a measure of EF or a subcomponent of EF. For example, we included articles in which authors reportedly assessed “executive control,” “effortful control,” and “self-regulation,” as there is significant overlap between these terms and EF (Liew, 2012; Zhou, Chen, & Main, 2012), and the measures used to assess these behaviors or skills are often the same as those used to measure EF (e.g., Stroop tasks).
We excluded studies that involved implementation of targeted interventions (Diamond & Lee, 2011) with a selected group or sample as part of their study procedures. Instead, we focused on studies that addressed existing schoolwide practices or classroom variables to determine their influence on EF. The reason we selected these studies is to examine how various aspects of the school environment are related to EF in natural settings, outside of the intervention of researchers. Greater insight is needed into students’ developmental context by focusing on schools as a naturally occurring setting.
We also excluded studies that solely used teacher or parent rating scales (e.g., Behavioral Rating Inventory of Executive Functions [BRIEF]; Gioia, Isquith, & Espy 2003; Gioia, Isquith, Retzlaff, & Espy, 2002), as these measures capture raters’ perceptions of the behavioral manifestations of EF and are not considered direct assessments of working memory, inhibitory control, and/or cognitive flexibility. Moreover, several studies have reported consistently low interrater agreement between parents and teachers’ behavioral EF ratings in samples of typically developing children and those with attention-deficit hyperactivity disorder (Dekker, Ziermans, Spruijt, & Swaab, 2017; Schneider, Ryan, & Mahone, 2019). In nonclinical samples using parent and teacher forms of the BRIEF, Dekker et al. (2017) reported low interrater agreement (.15) while Gioia, Isquith, Guy, and Kenworthy (2000) reported moderate agreement (.50). As for preschool ratings, such as the BRIEF-Preschool, agreement between parents and teachers is also weak (.19; Gioia, Espy, & Isquith, 2003) with reported differences across EF scales (e.g., Plan/Organize) and overall EF, where parents reported more executive dysfunction than teachers (Schneider et al., 2019). Due to these inconsistences along with potential rater bias, we focus our literature review on direct assessment measures.
Search and Selection Procedures
To identify studies measuring the relationship between EF and school-based experiences, we searched the following databases: ERIC, Academic Search Premier, Education Full Text, and PsycINFO. Search terms included all combinations of the following: (Field 1) executive function*, executive control, working memory, impulse control, inhibitory control, inhibition, cognitive flexibility, shifting, neurocog*, neuropsych*, self-regulation, stroop, flanker, or effortful control; (Field 2) school stress, peer stress, academ* stress, teacher student interact*, teacher quality, school quality, class quality, school climate, class* environment, teacher-student, teacher-child, peer relationship, toxic stress, chronic stress, cortisol, adrenaline, HPA Axis, bully, school violence, social relation, peer reject, or ostraciz*; (Field 3) child*, adolescent, or student*.
The initial search yielded 1,429 articles, excluding duplicates. We read titles and abstracts independently to determine if articles met inclusion criteria, resulting in 76 studies. Interrater reliability was 95%. Next, we read the methods section of all 76 articles in their entirety to determine whether each met our inclusion criteria (interrater reliability = 97%). We reread five articles for which there was disagreement, coming to agreement through consensus. We then hand searched all journals that had more than one included article for additional articles (Child Development, Developmental Psychology, Developmental Psychobiology, Developmental Science, and Early Childhood Research Quarterly), resulting in three articles (Cadima, Verschueren, Leal, & Guedes, 2016; Finch, Johnson, & Phillips, 2015; NICHD Early Child Care Research Network, 2005). Last, we reviewed reference lists of included articles and located three additional articles (Berry, 2012; Raver, McCoy, Lowenstein, & Pess, 2013; Weiland, Ulvestad, Sachs, & Yoshikawa, 2013). As shown in Table 1, we included a total of 20 articles in the review. Figure 1 illustrates the search parameters and final identification procedures.
Overview of study characteristics
Note. ANCOVA = analysis of covariance; ANOVA = analysis of variance; FLPKI = Family Life Project Key Investigators; HLM = hierarchical linear modeling; SEM = structural equation modeling.
Search parameters based on inclusion and exclusion criteria.
Analysis
We followed a systematic process to analyze each included study. First, we carefully read each study and created a table of the methods and major findings. Based on a socioecological systemic approach, in which individual outcomes are influenced by interactions between children and several nested contexts (e.g., families, schools, communities, and larger systems; Masten, 2013), and findings from previous family and community studies, we deductively coded each study based on a priori identified school-based themes (e.g., school-level environment, classroom-level environment, teacher interactions, peer interactions). One study did not fit into previously identified themes (i.e., teacher beliefs); therefore, we added it as a separate section.
As experts have previously noted (Jacob & Parkinson, 2015; McClelland & Cameron, 2012), it has been very difficult to identify trends in EF research due to widespread variability in (a) how researchers operationally define EF, (b) the EF domains they choose to measure (e.g., inhibitory control only vs. composite EF), and (c) the ways in which researchers measure EF. Therefore, after carefully investigating the tools each study used to measure constructs, we coded each based on the widely agreed on definition of EF (three distinct yet interrelated domains of inhibitory control, working memory, and cognitive flexibility; Miyake et al., 2000) instead of using terminology designated by the study’s researchers (e.g., self-regulation). Furthermore, because many direct measures of EF require a combination of inhibitory control, working memory, and cognitive flexibility, partly due to the interrelated nature of EF (Zelazo et al., 2016), we divided measures into simple and complex tasks to more accurately identify which EF domains the tool captured (see Garon, Bryson, & Smith, 2008). Simple tasks generally assess a particular domain or skill (e.g., inhibiting a prepotent response), whereas complex tasks incorporate multiple skills (e.g., maintaining an arbitrary rule and inhibiting a prepotent response, while producing an alternative response). We further stipulate that the Head–Toes–Knees–Shoulders (HTKS; Ponitz, McClelland, Matthews, & Morrison, 2009) task involves working memory, cognitive flexibility, and inhibitory control (McClelland et al., 2014) and, therefore, assesses overall EF. Additionally, the Tower of London (and Tower of Hanoi) assesses both overall EF and planning (Holmes et al., 2016). See Table 2. Previous researchers have set forth similar guidelines when investigating specific EF measures (Garon et al., 2008; Jacob & Parkinson, 2015). We then summarized each study’s findings and identified commonalities and differences.
Simple and complex EF and measures
Note. Adapted from Garon et al. (2008).
Tasks also found on the Preschool Self-Regulation Assessment (PSRA).
Results
In this section, we highlight the results of the 20 included studies. We first describe the study characteristics, define EF constructs and measures, and then report results based on school, classroom, dyadic, or individual teacher characteristics level variables.
Study Characteristics
The number of participants across studies was 14,771. Four out of 20 studies used the NICHD Study of Early Child Care and Youth Development data set (Berry, 2012; Berry, McCartney, et al., 2014; Holmes et al., 2016; NICHD Early Child Care Research Network, 2005) and two studies used the Family Life Project data set (Berry, Blair, et al., 2014; Burchinal et al., 2014). The number of participants across these studies that shared the same data set was 3,451 or 24.4% of the total sample. The average age of participants in the study was 5 years, ranging from 3 to 15 years (e.g., Holmes et al., 2016). The total percentage of boys in the studies was 49.9%. Only one study did not report the ratio of boys to girls (Day et al., 2015). Of the studies that reported race (N = 14), the overall sample was 44.2% Caucasian, 34.2% African American, 14.5% Hispanic, and 7.1% reported other races. Six studies did not report race (Cadima, Enrico, et al., 2016; Cadima, Verschueren, et al., 2016; de Wilde et al., 2016; Hawes et al., 2012; Leyva et al., 2015; Talwar et al., 2011), which made up 24.2% of the total participants across the studies. Out of the 20 studies selected, 16 had longitudinal designs and four were cross-sectional. Only one study used random assignment of participants into separate experimental condition groups (inclusion or exclusion condition; Hawes et al., 2012), while 16 studies used observational designs and three studies used quasi-experimental designs (Cadima, Enrico, et al., 2016; Raver et al., 2013; Talwar et al., 2011). In terms of the country where the studies were carried out, 14 studies were conducted in the United States, three in Europe (Cadima, Enrico, et al., 2016; Cadima, Verschueren, et al., 2016; de Wilde et al., 2016), and one each in Australia, Chile, and West Africa (Hawes et al., 2012; Leyva et al., 2015; Talwar et al., 2011).
Defining and Assessing EF Constructs
Across studies, there was variability in how researchers defined and measured EF. There was, however, consistency among 15 studies that defined EF as a set of higher order processes that consist of working memory, inhibitory control, and/or attention shifting or cognitive flexibility that underlie self-regulated action (e.g., Holmes et al., 2016).
Five research teams used the HTKS, where the experimenter instructs children to touch their head, but instead of following the command, the children must do the opposite and touch their toes. This requires recalling an arbitrary rule, withholding a natural response, and switching between stimuli, which measures overall EF. The HTKS has strong interrater reliability and internal consistency (α = .92–.94; McClelland et al., 2014).
Inhibitory control was the most frequently measured EF (N = 14). Three studies employed tasks assessing simple inhibitory control, which require participants to withhold an automatic or prepotent response with the delay of gratification tasks (Snack Delay, Gift Wrap) from the Preschool Self-Regulation Assessment (PSRA; Smith-Donald, Raver, Hayes, & Richardson, 2007). Complex inhibitory control was measured with two additional tasks (Balance Beam and Pencil Tap) from the PSRA; we define these as complex inhibitory control tasks because they require children to withhold a natural response while producing an alternative response based on an arbitrary rule. Consistency between assessor and coder responses has been established with the PSRA, with Cronbach’s alphas ranging from moderate to strong (α = .73–.99; average = .93; Raver et al., 2011) across all PSRA tasks. Other commonly used complex inhibitory control tasks included the Continuous Performance Test (Rosvold, Mirsky, Sarason, Bransome, & Beck, 1956) and Stroop-like tasks (Gerstadt, Hong, & Diamond, 1994). The Day–Night task, a common Stroop-like task for young children, has good reliability (α = .93–.99; McClelland et al., 2014).
Less than half of the studies (N = 9) incorporated a specific measure of working memory. Digit span tasks were among the most frequently used measures of simple working memory as they primarily assess recall of information presented in the same order as given. Variations of the digit span tasks, such as backwards digit span or recall, represent distinct or complex working memory skills related to executive control rather than automatic information retrieval (e.g., Hawes et al., 2012) and, thus, are described here as complex working memory tasks. These measures also pertain to nonverbal or visual span working memory tasks, such as Corsi Block Tapping (Corsi, 1972), where forward recall utilizes simple working memory, while an investigator-developed task, which required children to recall a sequence of dots on a grid in reverse order (de Wilde et al., 2016), utilizes complex working memory. Other working memory tasks, such as the Woodcock–Johnson Psychoeducation Battery–Revised (Woodcock, 1990), where children memorize either sentences or names, have strong test–retest reliability (.94–.92).
Cognitive flexibility, or attention switching, was least studied of the three EF skills (e.g., Leyva et al., 2015). Two studies measured cognitive flexibility using the Dimensional Change Card Sort (DCCS; Zelazo, 2006), and one used a flexible item selection task (Jacques & Zelazo, 2001). These tasks require participants to shift attention from one aspect of a stimulus to a new aspect of the stimulus under specific conditions. The DCCS has strong Cronbach’s alpha (α = .90–.93; McClelland et al., 2014).
School-Based Environmental and Interpersonal Factors
Researchers across studies investigated how a wide range of school-based factors shaped student EF maturation. Thus, in the following sections, we describe findings regarding (a) school-level factors, (b) classroom-level factors, (c) dyadic factors (interactions between students and teachers or between students and peers within schools), and (d) individual characteristics of school personnel. Additionally, in Table 3, for studies that reported a standardized regression coefficient or Cohen’s d, we report the magnitude of effect.
Overview of reported effect sizes
Note. EF = executive functioning; IC = inhibitory control; WM = working memory. Only studies that reported a standardized regression coefficient or Cohen’s d are reported and interpreted. We adapted Keith’s (2006) magnitude guidelines for school-related research to interpret β. Small = .01 to .09; Moderate = .10 to .24; Medium = .25 to .39; Large = .40 to 1. We followed Rubin and Babbie’s (2005) magnitude guidelines to interpret Cohen’s d. Small = .01 to .20; Moderate = .21 to .49; Medium = .50 to .79; Large = .80 to 1.
School Level
Two studies investigated relationships between school-level experiences and student EF maturation from preschool to early grade school. In their longitudinal study, Raver et al. (2013) examined the role of school quality (school-level poverty, school-level achievement, unsafe school climate, and low adult support) in supporting EF development among 391 students in Head Start (using Pencil Tap, Balance Beam), and then 4 years later, via teacher reports of student EF when students were in second or third grade. Although the sample was from a preexisting social–emotional intervention study when the children were 4 years old, we included this study in our review since the children (now about 8 years old) were not participating in any intervention in their current elementary school. The researchers examined the influence of unsafe schools and low adult support on student EF—not the effects of the intervention on EF. Using linear regression models, the authors found no relationship between school quality, controlling for Level 1 variables (e.g., family income), and later teacher-reported student EF; however, they found an interaction with the degree to which students felt unsafe in their school (measured using the Student Connection Survey; Osher, Kendziora, & Chinen, 2008). EF was significantly different from zero for both low and high early student EF difficulty. Specifically, for students with lower initial inhibitory control in Head Start, unsafe elementary schools were predictive of later teacher-reported EF deficits.
Similar to Raver et al.’s (2013), Talwar et al. (2011) examined a sample of 63 children (ages 3–6 years) in two private schools in one West African country and determined that overall school environment played a significant role in student EF. They included two groups of children who enrolled in two schools with different disciplinary practices (i.e., punitive vs. nonpunitive); these were the school’s disciplinary practices in the absence of intervention. They examined how schoolwide physical discipline practices (collected using school log books) were related to student performance on both cool and hot EF tasks. The researchers used simple inhibitory control tasks (Delay of Gratification, Gift Delay) and cognitive flexibility tasks (DCCS; Zelazo, 2006) to create a composite cool EF. They adapted each of these tasks to be emotionally salient for the students as measures of hot EF. The researchers found a significant school by grade interaction effect where, although the overall EF scores were similar in kindergarten, first-grade students in the school that did not implement punitive practices had significantly higher overall EF. Even when controlling for parental discipline, the researchers found that first-grade students in nonpunitive schools demonstrated higher EF scores than those students in schools that implemented punitive practices (e.g., corporal punishment).
Conclusions about school quality. Definitive conclusions are not possible, especially given very different school settings and method, but findings provide emerging evidence that aspects of school safety and climate play an important role in EF development, such that exposure to unsafe school settings may be detrimental to EF gains
Classroom Level
Thirteen studies examined how various dimensions of classroom quality related to students’ EF development during preschool through first grade. Classroom quality is most often operationalized along three domains: emotional support (e.g., sensitivity, modified lessons), classroom management (e.g., proactive discipline, routines), and instructional support (e.g., scaffolding and support; Hamre & Pianta, 2001). Nine of these studies used the Classroom Assessment Scoring System (CLASS; Pianta, La Paro, & Hamre, 2008), a well-validated instructional observation tool in which raters evaluate teacher–student interactions, which load onto three constructs (e.g., Weiland et al., 2013). In the present study, the constructs are defined as emotional support, instructional support, behavior support, or classroom organization. Because the CLASS was commonly used, we organized studies of classroom quality according to these dimensions: overall classroom quality, emotional support, instructional support, behavior support, and classroom organization. We included studies using other measures within the section that best describes the dimensions of classroom quality they examined. Within each section, we further distinguish between studies that predict student EF gains as a function of global quality (often measured using CLASS subscales) and those that predict student EF gains as a function of specific teacher practices at the classroom level (e.g., providing feedback).
Overall classroom quality
Four studies examined relationships between global classroom quality and student EF, with varying results. Both Cadima, Enrico, et al. (2016) and Hamre et al. (2014) found a relationship between overall CLASS scores and student EF. In their cross-sectional study, Cadima, Enrico, et al. examined how overall classroom quality (composite of emotional support, instructional support, and classroom organization) measured with the CLASS was associated with Portuguese preschoolers’ cool and hot EF. To measure EF, the authors used general EF tasks, which they identified as cool EF (HTKS, Toy Sort) or hot EF (Gift Wrap, Snack Delay, PSRA Attention/Impulsivity subtests) among 233 children from socioeconomically disadvantaged “at-risk” backgrounds (mean age: 63.6 months) and 252 children not “at-risk” (mean age: 56.5 months). Although, they explored at-risk and control groups of children based on socioeconomic status (i.e., household income, parent employment) and parent education level, they did not implement any intervention; control here refers to a natural difference between children the researchers deemed at risk and those they deemed not at risk. Using multiple group structural equation modeling, they found that, for both at-risk and not-at-risk children, overall classroom quality was a predictor of cool EF (β = .24) but not hot EF. Similarly, although Hamre et al. (2014) did not examine overall student EF, they did find positive effects for domain-specific cool EF skills in their sample of 1,407 preschool children. Using multilevel regression, the authors found that global ratings for responsive teaching (i.e., teachers’ overall classroom quality measured by the CLASS) predicted children’s gains in working memory but not inhibitory control.
Furthermore, Cadima, Enrico, et al. (2016) found two moderating effects. First, all children with higher family risk (e.g., family income level, parent employment) demonstrated stronger hot EF when they were in higher quality classrooms. Second, classroom quality moderated a relationship between temperamental negativity and cool EF; children from the nonrisk group with high temperamental negativity displayed lower cool EF when in lower quality classrooms.
In contrast, two studies (Cadima, Verschueren, et al., 2016; NICHD Early Child Care Research Network, 2005) that examined classroom quality, but did not include one of the three distinct classroom quality components defined in the CLASS (e.g., emotional support, instructional support, classroom organization), found no significant results. The NICHD examined how first-grade quality predicted complex inhibitory control (Continuous Performance Test), simple working memory (Memory for Sentences from Woodcock–Johnson Psychoeducational Battery–Revised; Woodcock, 1990), and overall EF and planning (Tower of Hanoi) among 727 children. When participants were in first grade, the authors used the Classroom Observation System for First Grade to measure teachers’ sensitivity and instructional conversations, which they aggregated into a single overall quality score. Using hierarchical regressions, they found no association between classroom quality and EF measures.
Similarly, Cadima, Verschueren, et al. (2016) found no relationship between classroom quality and later overall EF (HTKS) among 206 preschoolers (mean age: 59 months) in Portugal. In their sample, the emotional support and classroom organization subscales of CLASS were highly correlated and loaded as a single factor in a confirmatory factor analysis; therefore, the authors combined them into one emotional/organizational support factor. Using multilevel modeling, they found that the emotional support/classroom organization factor did not significantly predict EF gains, and results did not vary by initial EF skills for preschoolers at risk for poverty and social exclusion.
Conclusions about overall classroom quality. Both Cadima, Enrico, et al. (2016) and Hamre et al. (2014) provide initial evidence that the aggregated quality of emotional support, classroom organization, and instructional support that students receive at the classroom level may be important for their EF development. Particularly compelling was the finding that high classroom quality may serve as an important protective factor for students exposed to more risk factors. Contrary findings in the NICHD Early Child Care Research Network (2005) and Cadima, Verschueren, et al. (2016) may have been due to the way classroom quality was measured. For instance, Cadima, Verschueren, et al. combined emotional and organization support as one measure, without instructional support, and the NICHD used an aggregated score of the Classroom Observation System for First Grade, which did not include classroom organization. It could be that exclusion of some component of classroom quality limited their ability to detect associations between overall classroom quality and student EF maturation. Additional research is needed to further investigate the link between overall classroom-level quality and student EF.
Emotional support
Eight studies examined how emotional support related to student EF. None examined specific emotional support practices (e.g., naming emotions, expressing empathy) but rather captured overall teacher emotional support (e.g., sensitivity) provided at the classroom level. Because many studies examined emotional support, we report findings separately for early child care centers and preschool settings. No studies included older students.
Three research teams examined the association between the quality of classroom emotional support in early child care centers and student EF gains; overall, researchers found that the emotional support students received in early child care centers was related to their EF development. In the NICHD Early Child Care Research Network (2005) study, researchers began data collection at birth and continued through 6 years of age, with child care setting observations occurring four times between 6 and 36 months. The researchers used the Observational Ratings of the Caregiving Environment (ORCE) to evaluate child care quality along five dimensions—sensitivity to children’s nondistress signals, stimulation of children’s development, positive regard, detachment, and flatness of affect—to form one overall quality indicator. Based on these indicators, we classified their measure as assessing overall quality of teachers’ emotional support. Using hierarchical regressions, the researchers found a small direct relationship between the quality of emotional support students received during their early child care and their later first-grade simple working memory (
Similarly, Berry, Blair, et al. (2014), using data from the Family Project, examined how caregiver responsiveness and affection toward a child who spent more than 10 hours per week in early child care settings from 7 to 36 months related to composite EF skills (measured using six tasks; see Table 4) at 48 months. Analyzing data from a representative sample of 1,235 children from two high-poverty regions, the researchers found that higher levels of Caregiver Responsivity, as measured by the Home Observation for Measurement of the Environment (HOME) scale, was positively associated with student overall EF skills (β = .11, p < .03), controlling for family and child covariates. One SD improvement in caregiver responsivity corresponded to a modest but significant .11 SD improvement in EF skills. They found no evidence that this relationship varied by children’s cortisol levels.
Overview of associations between relevant dyadic-, classroom-, and school-level predictors and EFs
Note. * = Significant; NS = nonsignificant; CLASS = Classroom Assessment Scoring System; ECERS-R = Early Childhood Environment Rating Scale–Revised Edition; ELLCO = Early Language and Literacy Classroom Observation tool; IC = inhibitory control; HOME = Home Observation for Measurement of the Environment scale; STRS = Student–Teacher Relationship Scale; ORCE = Observational Ratings of Childcare Environment; PSRA = Preschool Self-Regulation Assessment; EF = executive functioning; DRD4 = dopamine receptor D4 gene; WM = working memory; + = significant relationship; ± = mixed finding; 0 = nonsignificant relationship. Burchinal et al. (2014) findings not included due to study limitations.
Four studies investigated the relationship between prekindergarten classroom emotional support quality and EF. Weiland et al. (2013) examined the associations between preschool classroom quality (measured with the CLASS on one day in spring) and the working memory and inhibitory control skills of 414 prekindergarten children (Age 4) attending 46 Boston public schools (83 classrooms). Although no linear relationship was found, using a quadratic model, they found significant associations between teachers’ emotional support and students’ scores on inhibitory control, as measured on a Pencil Tap task, with a stronger association at the higher end of the emotion support scale; emotional support more strongly predicted student performance when teachers provided stronger emotional support in the classroom.
To better understand how levels of classroom quality were uniquely related to student EF, Finch et al. (2015), Leyva et al. (2015), and Weiland et al. (2013) all investigated relationships between classrooms characterized by teacher emotional support and student EF using spline regression models. In their longitudinal study with 154 preschoolers, Finch et al. (2015) used ORCE to measure child care quality at 4 years of age and the Zoo Game to measure inhibitory control at 5 years of age. Setting spline knots at 2.5 and 3.25 on the ORCE positive caregiving quality scale, they found higher quality child care, characterized by caregiver sensitivity, at 4 years of age was positively related to inhibitory control when children were 5 years old (
Along similar lines, but on the other end of the spectrum, Leyva et al. (2015) found an adverse association between lower quality classrooms and student EF, also using the spline threshold approach. They identified a significant interaction between emotional support (measured by the CLASS) and student inhibitory control (Pencil Tap). Following 1,868 children (primarily from low-income backgrounds) in 91 classes in Chile from the beginning to the end of prekindergarten, they found that weaker emotional support in lower range classrooms was related to weaker complex inhibitory control, yet with a minimal effect size (d = −.03, p < .05). No association was found for emotional support and cognitive flexibility (DCCS; Leyva et al., 2015) or working memory (Weiland et al., 2013).
One study examined whether effects of emotional support varied depending on students’ initial EF skills. Choi et al. (2016) investigated how emotional support predicted inhibitory control gains (measured at the beginning and end of the year using the Pencil Tap task) among 169 preschool children (mean age: 56 months) in 51 Head Start classes in Oklahoma. The authors used the CLASS to rate teachers’ emotional support in the fall. They found no direct relationship between teachers’ emotional support and inhibitory control gains; however, the direct relationship became significant when they added interaction terms to the model. Emotional support was more strongly associated with gains among children with lower initial inhibitory control; the effect size was larger (d = .37) for students in the 25th percentile in inhibitory control skills versus those in the 75th percentile (d = .14). Furthermore, the authors found interaction effects for two of four CLASS emotional support dimensions: (a) negative climate and (b) teacher sensitivity.
Contrary to all other studies of emotional support, Burchinal et al. (2014) did not find significant results related to classroom emotional support and student EF. Drawing from the same data set as Berry, Blair, et al., (2014), they examined how prekindergarten quality, measured at 60 months of age with the Emotional Support scale of the CLASS, was associated with children’s working memory skills at 48 months, after controlling for child and family covariates measured at 6, 15, 24, and 36 months. 1 Using hierarchical linear regression, with a priori cut points, and then cubic b-spline regression models to generate cut points, they found no relationship between PK classroom quality at 60 months and student EF at 48 months. Because working memory data were collected several months prior to the classroom observation data, it is not possible to conclude that there was a lack of effect of classroom quality on student EF. Therefore, although Burchinal et al. (2014) examined the effects of other CLASS components (e.g., instructional support) on student EF, we did not report their findings in other sections of the review due to this very significant limitation of their study.
Conclusions about emotional support. Although researchers varied in the methods they used to assess emotional support, as well as the EF domains they chose to investigate, all of the studies found a significant relationship between the quality of the emotional support students received and EF skills. Studies obtained inconsistent results regarding whether the relationship was linear, quadratic, or spline. For instance, three of the studies obtained significant linear effects (Berry, Blair, et al., 2014; Berry, McCartney, et al., 2014; NICHD Early Child Care Research Network, 2015), while four (Choi et al., 2016; Finch et al., 2015; Leyva et al., 2015; Weiland et al., 2013) obtained interaction effects. Differences may be due to variability in measures used to assess classroom emotional support, as well as EF domains they chose to investigate (overall EF vs. domain-specific).
For example, all studies that investigated student inhibitory control (with the exception of the NICHD Early Child Care Research Network, 2005) found a significant relationship between teachers’ emotional support and preschool and first-grade students’ inhibitory control skills. Although the NICHD Early Child Care Research Network (2005) and Berry, McCartney, et al. (2014) used the same data set, they obtained contradictory findings related to inhibitory control. One distinction between the two studies is that Berry, McCartney, et al. (2014) added both Fosters Child’s Development and Intrusiveness scales to the ORCE quality composite, which may have provided sensitivity to an important aspect of child care quality not present in the NICHD Early Child Care Research Network (2005) study. Another distinction is that Berry, McCartney, et al. (2014) investigated the EF of younger children and included children’s genotype in the model, which may have additionally effected findings. Despite these differences, studies provide preliminary evidence that teachers’ emotional support at the classroom level appears to play an important role in development of the EF domain posited to mature first—inhibitory control (Best & Miller, 2010).
Studies that included other specific EF measures of working memory and cognitive flexibility were less conclusive. One reason for inconsistency in findings regarding different EF components may be related to how EF develops. Although EF as a whole becomes more sophisticated over time, each EF domain (i.e., inhibitory control, cognitive flexibility, working memory) is posited to mature at different rates. Specifically, inhibitory control develops more rapidly during preschool, while cognitive flexibility and working memory develop at a slower rate (Best et al., 2009; Best & Miller, 2010). Insignificant effects, therefore, may be because other EF domains, particularly cognitive flexibility, are still fairly underdeveloped in preschool. Moreover, due to the interrelated nature of EF domains (Zelazo et al., 2016), tasks that assess all three EF constructs or create composite scores may be more sensitive to overall EF development.
In addition, there is preliminary evidence that relationships between emotional support and students’ EF gains may be complex. Findings from three studies indicated that emotional support was more strongly associated with EF gains in classrooms that scored high/low on measures of classroom quality (Finch et al., 2015; Leyva et al., 2015; Weiland et al., 2013), while one study found that emotional support was more strongly associated with EF gains among children with weaker initial EF skills (Choi et al., 2016). Further research is needed to understand for whom and under what circumstances emotional support is most strongly associated with EF gains.
Instructional support
Four studies examined how global instructional support predicted preschoolers’ EF skills, and two studies examined how specific instructional practices predicted student EF gains. Leyva et al. (2015) found a significant linear association between instructional support (measured by the CLASS) and preschoolers’ inhibitory control gains (d = .06, p < .05), as measured by the Pencil Tap, such that all children benefited from classrooms with greater quality instructional support.
Contrary to these findings, neither Cadima, Verschueren, et al. (2016), Choi et al. (2016), and Hamre et al. (2014) nor Weiland et al. (2013) found a linear relationship between instructional quality and cognitive facilitation practices (e.g., engagement with children’s thinking by elaborating concepts, providing feedback) and preschool students’ performance on overall EF, working memory, and inhibitory control tasks. Additionally, Leyva et al. (2015) did not find an effect on cognitive flexibility.
Weiland et al. (2013), however, did identify a quadratic association between teachers’ instructional support (from the CLASS) and complex inhibitory control. As with emotional support, the association was higher at the higher end of the instructional support scale, indicating that instructional support more strongly predicted performance when teachers provided stronger instructional support. Moreover, using a spline regression model, they found instructional support negatively predicted inhibitory control (d = −.20; p < .01) in low-quality classes; this relationship was reversed in high-quality classes (d = .19; p < .01). Thus, there appears to be a threshold of overall classroom quality, below which more instructional support was associated with weaker gains, and above which more instructional support was associated with higher gains. In other words, when overall classroom quality is strong, instructional support facilitates EF gains; when the overall classroom quality is weak, instructional support is detrimental to EF.
Two research teams examined whether the effects of instructional support varied depending on students’ initial EF skills. Cadima, Verschueren, et al. (2016) found a significant interaction effect, such that instructional support was associated with stronger gains among preschool children who began the year with low EF skills. Additionally, they found that instructional quality differed for females (
Specific instructional foci. Two studies examined the extent to which content-focused instruction related to EF. Fuhs et al. (2013) examined how instructional foci were associated with EF gains over the school year, among 803 preschool students in 60 classes. They tested overall EF (Dimension Card Change Sort, Copy Design, Corsi Block Tapping, Peg Tapping, and HTKS) in fall and spring of preschool and observed teachers’ instruction on three occasions using the Teacher Observation in Preschool and the Child Observation in Preschool tools. The authors scored classroom observations for specific behavior management practices and the classroom’s focus on various academic areas. They found the degree to which the classroom was focused on instruction (
Weiland et al. (2013) used the Early Childhood Environment Rating Scale–Revised (ECERS-R) and the Early Language and Literacy Classroom Observation tool (ELLCO) to measure: (a) provisions for learning (i.e., “furnishings, room arrangement”; p. 203), (b) teaching interactions (e.g., encouraging children to communicate, using language to develop reasoning skills), and (c) literacy activities. Using a linear model, they found a small but significant relationship between teachers’ ELLCO literacy scores (support for literacy) and students’ scores on the inhibitory task (
Conclusions about instructional support. Overall, findings were mixed. Only one research team each found overall linear relationships (Leyva et al., 2015) and quadratic relationships (Weiland et al., 2013) between overall instructional support and student EF domains while two found moderating effects (Cadima, Verschueren, et al., 2016; Weiland et al., 2013) of classroom quality. Among studies examining how children’s initial EF skills moderated the effect of instructional support, one found a significant interaction (Cadima, Verschueren, et al., 2016), while another did not (Choi et al., 2016). In studies that examined more specific aspects of instruction than the CLASS, a shared finding was that a focus on literacy content predicted overall EF (Fuhs et al., 2013) and inhibitory control (Weiland et al., 2013). Because of variability across a small number of studies, additional research is needed to understand to what extent instructional support plays an important role in student EF maturation.
Behavior support
Two studies examined how teachers’ behavior support practices related to children’s EF gains; both focused on specific aspects of behavior support rather than global ratings of behavior support. Hamre et al. (2014) found that preschoolers whose teachers scored more highly on “positive management and routines” (a combined factor of the CLASS) experienced greater gains in their inhibitory control but not in their working memory. Fuhs et al. (2013) rated teachers’ instruction on three specific behavior management practices: (a) behavior approving statements, (b) behavior disapproving statements, and (c) listening to the child. They also coded these interactions for their emotional tone. Using multilevel modeling to predict student composite EF gains over the school year, they found that teachers’ behavior-approving statements significantly predicted EF gains (
Conclusions about behavior support. Findings are insufficient to draw conclusions, but it is encouraging that both found positive associations between well-established behavior management practices (e.g., routines, behavior-specific praise) and students’ inhibitory control and composite EF gains (Fuhs et al., 2013; Hamre et al., 2014). Given that other studies have found that overall classroom quality (e.g., Weiland et al., 2013) and students’ initial EF skills (e.g., Cadima, Verschueren, et al., 2016) moderate the relationship between teachers’ instruction and students’ EF gains, it is noteworthy that neither of these studies considered potential interactions.
Classroom organization
Three studies examined how classroom organization related to students’ EF skills. All focused on global classroom organization; none examined specific organizational practices. All included interaction terms to examine moderated relationships.
Two studies examined whether classroom quality moderated the association between classroom organization and EF gains. Although Leyva et al. (2015) found no direct or moderated relationships between classroom organization and inhibitory control and cognitive flexibility, Weiland et al. (2013) did find significant quadratic associations between teachers’ overall classroom organization and students’ inhibitory control. The association was stronger at the higher end of the classroom organization scale, indicating classroom organization more strongly predicted EF among teachers with stronger classroom organization. Using a spline regression model, they found classroom organization only predicted students’ performance gains on inhibitory control (
Only one study examined whether the effects of classroom organization varied depending on students’ initial EF skills. Choi et al. (2016) found no direct relationship between classroom organization and preschoolers’ inhibitory control gains. The interaction term, however, was significant, indicating classroom organization was more strongly associated with gains among children with lower initial inhibitory control (d = .36 for children in the 25th percentile but was insignificant for children in the 75th percentile). In a post hoc analysis, the authors examined which specific dimensions of classroom organization were associated with inhibitory control gains and found significant interaction effects for one of the three dimensions—productivity (e.g., clear routines, organized activities and directions) was more strongly related to gains among children with low initial inhibitory control skills.
Conclusions about classroom organization. There are too few studies with shared foci to draw conclusions, especially given that results were inconsistent. Further research is necessary to understand the relationship between classroom organization and gains in student EF.
Other aspects of classroom quality
One study examined other aspects of classroom quality. Day et al. (2015) examined how use of noninstructional time related to EF gains among 500 first-grade students in 51 classes in 18 elementary schools. They classified noninstructional time along two dimensions. First, they categorized noninstructional classroom activities as either child-managed (i.e., children working independently or with peers, with limited teacher interaction) or teacher-managed (i.e., the teacher leading activities and actively interacting with students). Second, they categorized noninstructional periods as either productive (i.e., used to support learning, including activity switching and organizational activities) or unproductive (i.e., a distraction from learning, including behavioral incidents, students waiting, and student off-task behavior). The authors collected 2-hour long videos of teachers’ classrooms on three occasions over the course of the year and coded each 15-second interval along these two dimensions (i.e., an interval could be coded as child-managed and productive, child-managed and unproductive, teacher-managed and productive, or teacher-managed and unproductive). Using multilevel, multivariate regression, Day et al. found that students who spent more time in child-managed productive noninstruction showed greater gains in EF skills in spring. They also found a significant interaction effect; among students with weaker fall EF scores, decreasing amounts of child-managed productive noninstructional time were associated with weaker spring EF skills compared with peers who began the year with stronger EF. In other words, decreases in child-managed productive noninstructional times were associated with exacerbated gaps between students with weaker initial EF skills and their peers with stronger initial EF skills. Additionally, a fall EF interaction effect was found related to teacher-managed unproductive time, such that students’ spring EF was greater among students whose teachers were progressively decreasing the proportion of teacher-managed unproductive time and lower among students whose teachers were progressively increasing the proportion of unproductive time.
Conclusions about other aspects of classroom quality. Researchers provided emerging evidence that the quality of noninstructional time may play an important role in first grade students’ EF abilities, particularly for students with lower initial EF. Additional research is needed to further investigate the role of noninstructional time in student EF development.
Dyadic Level
Five research teams examined the link between the quality of students’ relationships with other individuals in their schools, including their teachers and their peers, and EF maturation.
Individual teacher–student relationships
Three studies (Berry, 2012; Cadima, Verschueren, et al., 2016; de Wilde et al., 2016) investigated whether the quality of individual teacher–student relationships was related to student EF skills, with all three studies focusing on the role of teacher–student conflict. Examining data on 1,153 students enrolled in prekindergarten through fourth grade using autoregressive and cross-lagged models with multigroup (gender, race) analysis, Berry (2012) found that higher teacher-reported conflict on the Student–Teacher Relationship Scale (Pianta, 2001) in kindergarten was related to lower student inhibitory control on the Continuous Performance Test in first grade (β = .07). Yet this relationship in fourth grade was consistent for females only (β = .18). De Wilde et al. (2016) followed a group of 1,109 students in the Netherlands over a 2-year period from kindergarten through first grade, obtaining similar findings, although they used student self-report to assess perceptions of their relationships with their teachers (Young Children’s Appraisals of Teacher Support; Mantzicopoulos & Neuharth-Pritchett, 2003). Like Berry (2012), their cross-lagged models indicated that teacher–student conflict (e.g., “My teacher tells me that I am doing something wrong”) had a small negative relationship (β = −.05) with child working memory (visuospatial working memory task) development within (fall to spring) and between school years (β = −.08, kindergarten to first grade). Yet unlike Berry (2012), they found no differences based on gender.
In contrast, Cadima, Verschueren, et al. (2016) did not find a relationship between teacher–student conflict and student EF. Following a sample of 206, low-income Portuguese preschoolers from the beginning to the end of the school year, Cadima, Verschueren, et al. investigated the association between individual teacher–child relationships and student EF as measured by the HTKS. As in the Berry (2012) study, Cadima, Verschueren, et al. used teacher-reported conflict with each student in their class with the Student–Teacher Relationship Scale (Pianta, 2001). However, they found no significant association between perceived conflict and student EF. In essence, two out of three studies discovered a relationship between teacher–student conflict and lower student EF, with two investigating domain-specific EF skills.
Although teacher–student conflict was not associated with student EF in Cadima, Verschueren, et al.’s (2016) study, they did find a relationship between teacher–student closeness and student EF development over time. Teacher reports of teacher–student closeness (e.g., warmth, open communication) with individual students in the middle of the school year were associated with significantly stronger student EF gains (β = .18) from the beginning to the end of the preschool year. In contrast, de Wilde et al. (2016) did not find that teacher–student warmth was related to student working memory in their longitudinal study; Berry (2012) did not investigate warmth.
Two studies using crossed-lagged models found that the association between teacher–student relationships and student EF was reciprocal. After controlling for multiple family covariates (e.g., family income, maternal sensitivity), Berry (2012) found a reciprocal association between teacher–child conflict and changes in student inhibitory control over time. Berry found that not only did teacher–student conflict relate to lower student inhibitory control over time, but students with lower initial inhibitory control on school entry experienced more teacher–child conflict. This finding was consistent across elementary school, with lower levels of inhibitory control in first grade being predictive of teacher–student conflict in second grade; the standardized effect became more modest across time (β = .24 to .07). Similarly, de Wilde et al. (2016) found that students’ working memory negatively predicted teacher–student conflict within and between school years and positively predicted teacher–student warmth in first grade (β = .12) but not in spring of kindergarten. Although our focus is on how school-based factors contribute to EF development, it is worth noting that, across studies, EF was predictive of students’ problematic relationships with teachers, as this implies that cross-sectional associations may be capturing how EF contributes to teacher–student relationships, not merely the reverse.
Conclusions about teacher–student relationships. The most consistent finding was that teacher–student conflict was negatively associated with inhibitory control and working memory, particularly in lower grades, and was not associated with overall EF (Cadima, Verschueren, et al., 2016). Of the two studies investigating the associations between teacher–student closeness and EF (Cadima, Verschueren, et al., 2016; de Wilde et al., 2016), results were contradictory. Inconsistencies may be related to which EF domains researchers measured and the unique effect teacher–child conflict and closeness may have on overall and domain-specific EF skills. For instance, it could be that working memory is less sensitive to close teacher–student relationships, particularly during kindergarten and first grade, but more sensitive to the stress related to teacher–student conflict.
Although limited to two studies, these studies highlight that student EF skills and the quality of their interactions appear to be reciprocal, such that initial student EF skills relate to the quality of teacher–student relationships, which in turn relate to student EF skills. Additional research is recommended to investigate these relationships further.
Peer relationships
Three studies examined the role of peer relationships in students’ EF, with mixed results (de Wilde et al., 2016; Hawes et al., 2012; Holmes et al., 2016). Analyzing data from a national sample (n = 1,164) from the Study of Early Child Care and Youth Development, using SEM cross-lagged models, Holmes et al. (2016) found that high peer problems (i.e., peer rejection and peer victimization) were predictive of lower overall student EF (a composite of the Continuous Performance Test, Memory for Sentences task, and Parent Report), yet the association diminished as students aged. However, peer problems at 9/10 years old did not predict EF at 15 years old (β = .00, p = .92). They found no gender differences.
In contrast, Hawes et al. (2012) found a significant moderating effect of gender. They simulated an adapted computerized task in which 55 participants (aged 8–12 years) believed they were playing with fictional peers, programmed to either include or exclude (i.e., ostracize) participants on a number of experimental trials; participants were randomly assigned to either the inclusion or exclusion condition. Hawes et al. assessed simple working memory using forward digit span and recall tasks and assessed complex working memory using backward digit span and recall tasks (Working Memory Test Battery for Children; Pickering & Gathercole, 2001). Although they found no significant main effects of exclusionary status on working memory, they did find gender had a moderating effect. Girls who felt excluded performed more poorly than girls who felt included on both forward digit recall (d = .89, p = .022) and forward digit span (d = .86, p = .027), with large effect sizes noted. Additionally, they found no differences between excluded or included boys on simple or complex working memory tasks but did find that boys in the excluded condition performed better on backward digit span than girls in the excluded condition.
Contrary to Hawes et al. (2012) and Holmes et al. (2016), de Wilde et al. (2016) found no significant relationship between peer relationships and student EF. Focusing only on positive aspects of relationships, through nominations of likability and dyadic friendedness (i.e., whether two children agreed that they were best friends), they investigated how these predicted student working memory across 2 years of kindergarten and first grade. Their cross-lagged models did not reveal any significant relationships between peer social experiences and working memory.
Two research teams used crossed-lagged models to investigate the reciprocal relationship between peer relationships and EF, reporting varying findings. Holmes et al. (2016) found a significant reciprocal relationship between EF and peer problems. As previously highlighted, Holmes et al. found that students’ peer problems predicted their EF skills, with decreasing effects over time. Students’ EF skills also predicted the likelihood that students would experience peer problems. For instance, EF skills at age 4.5 years predicted lower peer problems at ages 6 years (β = −.36, p < .001) and 9/10 years (β = −.20, p < .001). Higher EF at age 9/10 years also predicted lower peer problems at age 15 years (β = −.09, p = .01). Similarly, de Wilde et al. (2016) found that individual student working memory positively predicted likeability in earlier time points (β = .08) but did not find a reciprocal relationship. Overall, individual student EF was predictive of whether students were more likely to experience peer problems across elementary through adolescence, although effect sizes diminished with age.
Conclusions about peer relationships. Results from the limited studies indicate that some dimensions of peer relationships may be predictive of EF. Specifically, peer problems and feeling ostracized may have an adverse relationship with students’ EF. It is difficult, however, to draw definitive conclusions given inconsistent measures across studies for both EF and aspects of peer relationships. For example, Holmes et al. (2016) and Hawes et al. (2012) focused on problems in peer relationships, whereas de Wilde et al. (2016) focused on positive aspects of peer relationships, such as peer likeability and friendship. Additionally, the gender-specific findings in the Hawes et al. (2012) study may be due to the differences in how peer problems were measured. Hawes et al. focused on exclusion, while Holmes et al. (2016) created a composite measure of peer problems that included peer rejection and peer victimization. It could well be that girls are more sensitive to exclusion/rejection, while both genders are equally affected by peer problems; yet additional research should investigate the distinct role types of peer conflict play across boys’ and girls’ EF development.
Although findings were limited to one study (Holmes et al., 2016), the link between peer relationships and student EF may be reciprocal. Not only may peer problems negatively predict student EF development, but EF skills may reduce the likelihood of experiencing peer problems. Additional research is needed to investigate these relationships further.
Individual Teacher Attributes Level
Only one study examined the individual attributes of school personnel and their relationship to students’ EF abilities. In their cross-sectional study, Hur et al. (2015) investigated the extent to which teachers’ child-centered beliefs were associated with student academic outcomes, with student EF as a mediating factor. Child-centered beliefs are the extent to which teachers believe students learn best when given choices and when teachers are sensitive and responsive to their students’ needs. The authors assessed 444 preschoolers’ overall EF with the HTKS task and 103 of their teachers’ beliefs with the Modernity Scale (adapted from the Parental Modernity Scale; Schaefer & Edgerton, 1985). Using path analysis, researchers found that teachers’ child-centered beliefs were positively associated with preschoolers’ overall EF (β = .09), which in turn predicted academic achievement, such as receptive vocabulary, phonological awareness, and mathematical skills (β = .26 –.35).
Conclusions about individual teacher attributes
This study provides preliminary evidence that teachers’ child-centered beliefs relate to student EF development. Findings highlight that teachers’ beliefs may be an additional variable shaping student EF development. Because the study is cross-sectional and did not measure whether beliefs were related to teachers’ classroom practices, it will be important for researchers to continue investigating the link between teachers’ beliefs and student EF.
Discussion
We conducted a systematic literature review to examine school-, classroom-, and dyadic-level experiences associated with students’ EF development. We first discuss results in terms of school-based environmental and interpersonal factors that served as potential inhibitors (stressors) and facilitators. We then highlight limitations and important directions for future research. Last, we discuss implications for practice to foster student EF development.
Potential School-Based Contributors to EF Development
Across studies, we found emerging evidence for factors at the school, classroom, and dyadic levels that likely serve as stressors that inhibit student EF, with the majority of effects in the small to moderate range (Table 3). First, aligned with research on the negative effects of stressors within community and home environments on students’ EF maturation, particularly in the form of physical safety (DePrince et al., 2009; Perkins & Graham-Bermann, 2012), our review provides preliminary evidence that unsafe or punitive school environments play an adverse role in student EF maturation (Raver et al., 2013; Talwar et al., 2011). For example, Talwar et al.’s (2011) findings affirm prior research indicating that frequent exposure to punitive discipline likely has a cumulative negative effect (Straus & Mouradian, 1998; Straus, Sugarman, & Giles-Sims, 1997). Second, one of the most consistent findings in our review was that low classroom emotional support in early child care and preschool settings likely has an adverse relationship with EF maturation (e.g., Leyva et al., 2015). These findings are consistent with a well-established research base indicating that parenting characterized by low emotional support and harsh interactions is often associated with lower child EF maturations (e.g., Bernier et al., 2012; Cuevas et al., 2014; FLPKI, 2013). Third, aligned with Pianta’s (1999) and Blair and Diamond’s (2008) position that conflictual teacher–student relationships serve as stressors that inhibit the development of self-regulation and associated EF, our review provides emerging evidence for the potential adverse link between teacher conflict and domain-specific child EF over time (e.g., Berry, 2012; de Wilde et al., 2016), as well as transactional relationship between dyadic relationships and EF (Berry, 2012; de Wilde et al., 2016). This finding is consistent with other research, which has found that increased child cortisol levels are associated with teacher and child conflict (Lisonbee, Mize, Payne, & Granger, 2008; Rappolt-Schlichtmann et al., 2009), while low student impulse control is often associated with teacher and child conflict (e.g., Rudasill & Rimm-Kaufman, 2009). Fourth, the initial findings (Berry, 2012; Cadima, Verschueren, et al., 2016; de Wilde et al., 2016) on the adverse effects of negative peer relationships align with the need-to-belong theory (Baumeister, DeWall, Ciarocco, & Twenge, 2005) and also with research linking social exclusion to lower teacher-/parent-reported student self-regulation (e.g., Stenseng, Belsky, Skalicka, & Wichstrøm, 2015). Last, some studies in our review found that the relationship between negative school-based experiences was most significant for students with initial low EF (Berry, 2012). These studies provide preliminary evidence that school-based stressors may exacerbate limited EF maturation for at-risk and vulnerable children, including those with mental health problems or disabilities, as well as those residing in stressful environments (FLPKI, 2013; Lengua et al., 2014; Rhoades et al., 2011) and coming from neglectful or abusive home and community environments (Pechtel & Pizzagalli, 2011).
We found promising evidence that classroom-level emotional support was a consistent facilitator of EF development, with the majority of effects within the moderate magnitude range (Table 3). In alignment with Blair and Diamond’s (2008) proposal that students’ self-regulation and EF are enhanced when teachers model self-regulation skills and create caring and supportive environments, our review found that young students’ EF skills, especially inhibitory control, developed more rapidly in early child care and preschool settings characterized by strong teacher emotional support (e.g., Berry, Blair, et al., 2014; Choi et al., 2016; Finch et al., 2015). Prior research indicates that high-quality classrooms that foster positive relationships, provide continuous feedback, and promote student learning are more likely to foster students’ cognitive, academic, and social–emotional needs (e.g., Howes et al., 2008; Mashburn et al., 2008); our findings indicate that they also likely foster students’ early EF skills. Additional research is needed to further investigate the role of specific instructional practices (e.g., scaffolding, behavior management), teacher warmth at the dyadic level, and teachers’ individual attributes (e.g., beliefs), as our review identified promising but inconclusive results for these factors. Overall, the positive effects of teacher quality on student EF maturation may be dependent on teachers’ ability to create classroom environments characterized by strong emotional support.
Last, researchers more often found significant relationships between school-, classroom-, and dyadic-level experiences with inhibitory control than with other components of EF (i.e., working memory, cognitive flexibility). This finding may indicate that inhibitory control is more malleable and sensitive to school-based environmental experiences than other EFs. Yet we recommend caution regarding this conclusion. Because the body of studies primarily focused on preschool and early elementary–age children, when inhibitory control is posited to develop rapidly (Best & Miller, 2010), findings may be only relevant for younger children; studies with older students may obtain different findings. Furthermore, the significant inhibitory control finding may be an artifact of the fact that more studies included inhibitory control measures; for example, only four studies included measures of cognitive flexibility, whereas 14 studies used measures of inhibitory control, limiting insight into the unique links between school-based experiences and individual EFs.
Critique and Future Research
The extant body of research on school-based contributors to EF development has several important strengths. All studies in the review included large samples and used well-validated measures of students’ EF skills. In addition, many studies tested models that accounted for complex (e.g., nonlinear, moderated) relationships between school-based factors and EF development. The studies reflect sophisticated conceptualizations of both EF development and schools as complex social systems. Although multiple studies drew from the same database, we do not believe results were inflated because each study investigated different variables of interest or age-groups. For instance, Berry (2012) investigated effects of dyadic-level variables, while Berry, McCartney, et al. (2014) examined classroom-level variables. These are noteworthy strengths that enhance the quality of the research and the conclusions that can be based on it.
There are, however, some important limitations to the extant research, which we encourage readers to consider when interpreting results. These limitations include (a) an almost exclusive focus on early childhood; (b) diversity of measures of EF, combined with a lack of conceptual clarity about the components of EF being measured; and (c) pervasive, substantial problems with the measurement of classroom quality. In the following sections, we critique these aspects of the extant research.
Participant Characteristics
Out of 20 studies, 17 included participants below the age of 6 years, and only one study included adolescent participants (Holmes et al., 2016). Adolescence, especially early adolescence, is a period of active EF maturation (Crone, 2009; Hughes, 2011; Selemon, 2013; Zelazo et al., 2016), high stress (American Psychological Association, 2014), stress reactivity (Dahl & Gunnar, 2009; Lupien, McEwen, Gunnar, & Heim, 2009; Spear, 2009), and increased risk for psychopathology and delinquency (Dahl & Gunnar, 2009; Lupien et al., 2009; Paus, Keshavan, & Giedd, 2008; Waldman & Lahey, 2013); as such, it is a heightened period of vulnerability for escalation of problematic relationships among perceived stress, ability to manage stress, EF (cool and hot), and behavior and academic performance. Yet given the predominant focus on younger children, findings for older children are inconclusive. Therefore, we encourage researchers to include older participants, particularly early adolescents. Middle school is a critical identification, prevention, and intervention period (Dahl & Gunnar, 2009; Lupien et al., 2009; Zelazo et al., 2016) and is thus particularly worthy of future study.
Furthermore, eight studies included predominantly White participants (e.g., Berry, 2012; Finch et al., 2015; Holmes et al., 2016), while only one study included mostly African American participants (Raver et al., 2013), and six studies did not report their participants’ race/ethnicity. Thus, it is difficult to determine if the findings from these studies could be generalized to different races/ethnicities, age-groups, and genders. As such, we recommend that researchers include participants from diverse backgrounds.
EF Measures
Researchers generally agreed on the core EF skills; however, there were differences in how they conceptualized and measured these constructs. Although we analyzed studies based on the EF measures they used, rather than the labels they assigned to each measure, discrepancies in how researchers described them can make it difficult to synthesize the literature. For instance, many studies described EF as components of behavioral or cognitive self-regulation (Berry, 2012; Choi et al., 2016; Day et al., 2015; Leyva et al., 2015), while others equated EF with cognitive self-regulation (Fuhs et al., 2013). Although there needs to be more consistency in EF terminology across research studies and disciplines, our review extends the theoretical and practical implications from Garon et al. (2008), Miyake et al. (2000), and Zelazo et al. (2016) by examining EF in children from a developmental–contextual perspective and emphasizing school-environmental mechanisms that influence this development. Future research should therefore explore how children develop simple EF to complex EF skills across early childhood and through adolescence. For example, how do schools and classroom practices promote EF development so children are able to transition from simple working memory to complex working memory? Moreover, examining cool and hot EF across childhood and adolescence can aid in understanding how specific classroom or targeted interventions (e.g., instructional support vs. emotional support) can develop or hinder specific EF domains, particularly ones that involve strong emotional valence.
Across studies, researchers used a variety of assessment tools to measure EF. Zelazo et al. (2016) contend that measuring pure EFs is challenging, as most direct laboratory tasks may require a combination of inhibitory control, working memory, and cognitive flexibility. Yet measuring only one aspect of EF, or only composite EF, limits our understanding of how predictors relate to students’ EF as they age, as each domain is posited to develop at varying rates. Moreover, different EF skills may become more important for student outcomes at different ages (Best & Miller, 2010). For instance, hot EF correlates significantly with social competence and, presumably, plays a much larger role in adolescent achievement than cool EF (Brock et al., 2009; Zelazo et al., 2016). Because adolescence is an important EF maturation period that is characterized by increasingly complex social interactions, heightened vulnerability to peers’ judgments (Lerner & Steinberg, 2004), and amplified reactivity to perceived stressful situations (Spear, 2009), secondary classrooms present more opportunities for students to engage in real-life decision making, which requires emotional restraint and self-control in tempting or stressful situations. The only study in the literature review that included older students did not measure hot EF.
We recommend that researchers not only use consistent EF (domain-specific and overall EF) measures, such as the NIH (National Institutes of Health) Toolbox (Slotkin et al., 2012), but also include hot EF measures, such as the IOWA Gambling Task (Bechara, Damasio, Damasio, & Anderson, 1994). These measures can provide greater insight into the unique role school environment and interpersonal relationships have on both the development of overall EF and each of the three domain-specific EFs over time and can increase the cohesion of findings across studies and age-groups. Although we did not include studies that used EF rating scales, we recommend including both EF performance-based measures and rating scales. Performance-based measures and rating scales do not consistently correlate (Isquith, Roth, & Gioia, 2013; Toplak, West, & Stanovich, 2013); yet rating scales may capture emotionally related EF and the application of EF skills within an everyday environment. As such, rating scales may provide insight into how well students use EF skills in the classroom.
Classroom Quality
Extant studies primarily relied on the CLASS. The CLASS is well validated, with rigorous procedures, but exclusive reliance on a single instrument may increase susceptibility to any biases in the CLASS. Using a range of instruments would increase the likelihood that all salient dimensions of classroom quality are accounted for in the literature.
Furthermore, only three studies collected classroom quality data on more than 1 day (e.g., Day et al., 2015); instead, most studies used the CLASS to rate four consecutive 20-minute windows of instruction. Collecting all classroom quality data in a single day raises three concerns. First, instructional quality varies widely over the year; because classroom quality data collected on a single day may not generalize to the instruction students typically received, researchers recommend rating instruction on a minimum of three separate occasions (e.g., Matsumura, Garnier, Slater, & Boston, 2008; Ramsdell, Raudenbush, Rowan, Staiger, & Winn, 2012). Second, having a single rater collect data for one teacher for four consecutive 20-minute windows may exacerbate both rater bias and recency bias. Even with rigorous certification and reliability procedures, raters tend to systematically over- or underestimate aspects of instruction (e.g., Ramsdell et al., 2012); as such, scholars recommend randomly assigning observations to raters, so teachers are not more likely to be rated by any one rater (Bell & Jones, 2018). Obtaining all four estimates of a teacher’s classroom quality from a single rater may increase rater bias. Furthermore, when raters score the same teacher multiple times in a single day, their rating of the second window is likely to be influenced by their rating of the first window, while their rating of the third window is likely to be influenced by their rating of the first and second, and so on; ratings of the four windows are likely to be dependent on one another. This may increase reliability, but it undermines accuracy of ratings for the second through fourth windows. Third, Day et al. (2015) found that changes in teachers’ classroom quality over the year were related to children’s EF gains. Collecting instructional quality data on a single day does not provide an opportunity to understand how changes in instructional quality over time relate to EF gains. All of these issues may decrease studies’ sensitivity to how classroom quality relates to EF development. Thus, future studies should evaluate instructional quality on more than 1 day, using more rigorous procedures to reduce biases associated with individual raters.
Finally, most studies examined broad components of overall classroom quality (e.g., emotional support, organization) and not specific instructional practices (e.g., praise, modeling). Specific practices are often easier to change, through professional development, than broad classroom climate (Kennedy, 2017); as such, identifying specific practices would have particular utility for helping school leaders and teachers improve their practice. Thus, we also recommend that researchers examine more specific instructional practices (e.g., self-monitoring, instructional scaffolding, formative feedback) that, theoretically, could be related to EF development; identifying specific practices would be especially useful for informing professional development.
Implications for Practice
EF is malleable and sensitive to both negative and positive experiences (Zelazo et al., 2016), including experiences in schools. As such, our findings have implications for practice. First, it will be important for school leaders, policymakers, and school personnel to understand the importance of EF in student behavioral and academic achievement, as well as the potential role school experiences play in student EF maturation across the school years. Thus, professional preparation and professional development should consider including content related to these areas to increase the likelihood that schools foster EF at the school, classroom, and dyadic levels.
Second, based on the findings of our review, school personnel should consider addressing potential stressors within schools related to school safety, teacher–student conflict, and peer problems, as well as foster EF skills through strong classroom-level emotional support. School leaders and policymakers should consider investing in strategies that improve school safety, while teachers should actively engage students in emotionally supportive interactions. Schoolwide and in-class universal social–emotional learning programs have demonstrated positive effects (Durlak, Weissberg, Dymnicki, Taylor, & Schellinger, 2011); although no studies have yet tested the effects of these programs on EF maturation, they are associated with positive outcomes (e.g., improved school safety) that are related to EF maturation. Additionally, school-based programs have shown promising results for improving EF abilities across grade levels (e.g., Tools of the Mind [Blair & Raver, 2014], Tools for Getting Along [Smith et al., 2014]); as such, these kinds of programs could be good investments for schools. Future research should continue to evaluate effectiveness of programs that address school-based stress and student EF.
Third, the information from our review has important implications for early identification of students at risk for poor EF. Studies found that students with weaker initial EF (e.g., Raver et al., 2013) were most affected by school-based factors (e.g., emotional support). Thus, schoolwide or universal screening of students for EF deficits, especially inhibitory control and stress in early childhood, may enable educational professionals to intervene early and provide targeted prevention and intervention programming for those at risk of poor EF development.
Conclusion
Executive functioning provides the foundation for students’ school and lifelong success. Our review provides compelling initial evidence that schools may play an important role in promoting student EF development. Schools have potential to be a key leverage point for fostering student EF development, if school leaders, professionals, and educators encourage students to build positive relationships with peers and teachers alike and consciously build learning environments that are sensitive to their needs at the school and classroom levels.
Footnotes
Notes
Authors
MICHELLE M. CUMMING, PhD, is an assistant professor of special education in the College of Arts, Sciences and Education at Florida International University, 11200 S.W. 8th St., ZEB 340A, Miami, FL 33199, USA; email:
ELIZABETH BETTINI, PhD, is an assistant professor of special education in the Wheelock College of Education & Human Development at Boston University, 2 Silber Way, Boston, MA 02215, USA; email:
ANDY V. PHAM, PhD, is an associate professor of school psychology in the College of Arts, Sciences and Education at Florida International University, 11200 S.W. 8th St., ZEB 360B, Miami, FL 33199, USA; email:
JEEYUN PARK, MS, is a doctoral student in curriculum and instruction in the College of Arts, Sciences and Education at Florida International University, 11200 S.W. 8th St., Miami, FL 33199, USA; email: