Development of hot and cool executive functions in middle childhood: Three-year growth curves of decision making and working memory updating

https://doi.org/10.1016/j.jecp.2018.04.002Get rights and content

Highlights

  • Working-memory and decision-making abilities increased during middle childhood.

  • Initial levels of working-memory were predicted by socioeconomic status, verbal ability, and processing speed.

  • Boys had higher initial levels of decision-making ability than girls.

  • Rate of change was not predicted by gender, socioeconomic status, verbal ability, or processing speed.

Abstract

Although middle childhood is an important period for the development of hot and cool executive functions (EFs), longitudinal studies investigating trajectories of childhood EF development are still limited and little is known about predictors for individual developmental trajectories. The current study examined the development of two typical facets of cool and hot EFs over a 3-year period during middle childhood, comparing a younger cohort (6- and 7-year-olds at the first wave [T1]; n = 621) and an older cohort (8- and 9-year-olds at T1; n = 975) of children. “Cool” working memory updating (WM) was assessed using a backward digit span task, and “hot” decision making (DM) was assessed using a child variant of the Iowa Gambling Task. Linear latent growth curve analyses revealed evidence for developmental growth as well as interindividual variance in the initial level and rate of change in both EF facets. Initial level of WM was positively associated with age (both between and within cohorts), socioeconomic status, verbal ability, and processing speed, whereas initial levels of DM were, in addition to a (potentially age-related) cohort effect, exclusively predicted by gender, with boys outperforming girls. None of the variables predicted the rate of change, that is, the developmental trajectories. However, younger children, as compared with older children, had slightly steeper WM growth curves over time, hinting at a leveling off in the development of WM during middle childhood. In sum, these data add important evidence to the understanding of hot and cool EF development during middle childhood.

Introduction

Executive functions (EFs), a set of cognitive skills necessary for goal-directed behavior and self-control, are correlates and predictors of a large range of social and cognitive developmental outcomes over the lifespan (Best et al., 2009, Hughes and Ensor, 2011, Moffitt et al., 2011). The most prominent and most extensively studied developmental period for EFs is preschool age (Carlson, Zelazo, & Faja, 2013). However, EF performance shows medium to large age-related increases during middle childhood (between 5 and 11 years; Romine & Reynolds, 2005). Age-related increases then become smaller from 11 to 14 years, but there is consensus that EF development continues throughout adolescence (Hughes, 2011). Different facets of EFs, such as inhibitory control, flexibility, and working memory updating (WM), seem to follow distinct developmental trajectories (Best and Miller, 2010, Hughes, 2011). Furthermore, it has been suggested that “hot” EFs (which are needed in situations that elicit emotion, motivation, or a conflict between immediate reward and long-term goals) show a more decelerated development during childhood than “cool” EFs (which support goal-directed behavior under relatively decontextualized, nonemotional conditions; Peterson and Welsh, 2014, Prencipe et al., 2011). Given the importance of EFs for children’s cognitive and social functioning, it is crucial to understand how EFs develop and to identify predictors for the course of individual EF development during middle childhood, which is a period of increased self-regulatory demands (Otero & Barker, 2014). In the current study, we used growth curve modeling to tackle these questions and to examine developmental trajectories of one hot EF facet and one cool EF facet across a 3-year period, comparing younger and older age cohorts (ages at the first wave [T1] = 6–7 and 8–9 years, respectively).

Cool EFs have often been operationalized following a tripartite structure using measures of inhibitory control, flexibility, and WM (Miyake et al., 2000). Whereas in preschoolers a unitary EF factor was frequently found to be the best fitting model, differentiation of EF facets seems to increase during childhood. In school-aged children, at least two distinct cool EF components can be found, namely flexibility (or set shifting) and WM (Carlson et al., 2013). Unlike simple working memory tasks that merely require the maintenance of information (e.g., forward digit span), executive working memory tasks (e.g., backward digit span) call for maintenance and manipulation of information (Best & Miller, 2010). Childhood development of WM has been studied mainly using cross-sectional designs (Carlson et al., 2013). For example, in a cross-sectional study with four age groups from 7-year-olds to young adults, WM capacity increased until the age of 15 years, followed by stabilization, whereas an inhibitory component of the task showed increased capacity until young adulthood (Carriedo, Corral, Montoro, Herrero, & Rucián, 2016). In the same vein, Best and Miller (2010) concluded in their review of EF development that most studies including large age ranges suggest linear WM development from preschool throughout adolescence.

However, there is also evidence for nonlinear WM development during middle childhood. Using a cross-sectional approach in 6- to 13-year-olds, Brocki and Bohlin (2004) found small improvements for a combined factor of WM and verbal fluency across the examined age range, with developmental spurts occurring around the ages of 8 and 12 years. Discontinuities in WM development, albeit with a different time course, also appeared in a 2-year longitudinal study using the backward color recall task, an analogue of the backward digit span (Röthlisberger, Neuenschwander, Cimeli, & Roebers, 2013). Here, large WM improvements were found in the younger cohort (starting in prekindergarten) and a slight leveling off was found in the older cohort (starting in kindergarten). To sum up, the cool EF facet WM develops throughout childhood and adolescence. Some evidence (Brocki and Bohlin, 2004, Röthlisberger et al., 2013) suggests developmental spurts at certain age periods during middle childhood, but because the findings are somewhat inconsistent, we aimed at thoroughly elucidating WM development during middle childhood by using a 3-year longitudinal design with three measurement waves and by comparison of two age cohorts.

Typical tasks to measure hot EFs include delay of gratification, delay discounting, and gambling tasks that assess decision making (DM; Peterson & Welsh, 2014). One of these tasks, the Iowa Gambling Task (IGT; Bechara, Damasio, Damasio, & Anderson, 1994), has been designed to simulate real-life DM under uncertainty. In the IGT, participants are told to collect as many points as possible by choosing cards from several decks. Every time they choose a card, participants are informed about the amount of points they gained and lost. Unknown to the participants, the card decks differ in the probability and magnitude of gains and losses: Whereas the “advantageous” decks yield small gains but even smaller losses, resulting in a long-term net gain, the “disadvantageous” decks yield larger gains (immediate reward) but also unpredictable larger losses (negative consequences), resulting in a long-term net loss. In the IGT, people differ in their preference for immediate reward and their sensitivity to negative future consequences (Franken & Muris, 2005).

In a developmental perspective, young children’s decisions in the IGT and its child-appropriate variants are based primarily on short-term outcomes. That is, young children prefer the immediate reward, resulting in relatively frequent disadvantageous choices (Crone, Bunge, Latenstein, & van der Molen, 2005), but this tendency decreases throughout childhood until late adolescence (e.g., Crone and van der Molen, 2004, Hooper et al., 2004). A cross-sectional study revealed age-related improvements in DM between 7 and 15 years, but only when frequency of losses was low (10% of trials) in a two-choice version of the IGT (Crone et al., 2005). The youngest age group (7–9 years) showed a larger preference for the immediate reward and a weaker sensitivity to long-term consequences than did both older age groups (10–12 and 13–15 years), with no difference in performance between the latter two groups. This indicates stabilization of DM during the period from middle childhood to early adolescence. In contrast, in another cross-sectional study, early adolescents (aged 10–13 years) were myopic to future negative consequences, whereas younger children (aged 8–9 years) and older adolescents (aged 14–17 years) performed better in the classic version of the IGT (Smith, Xiao, & Bechara, 2012). It should be noted that these conflicting findings could be explained by differences in the task difficulty between the studies. Moreover, cross-sectional studies can be confounded by cohort effects (Hughes, Ensor, Wilson, & Graham, 2010). Taken together, middle childhood is an important developmental period for the hot EF facet DM, but because prior studies have obtained mixed results and have been limited to cross-sectional comparisons, the current study aimed at contributing to a better understanding of the developmental trajectories of DM during middle childhood by using a longitudinal design, two age cohorts, and appropriate statistical procedures.

Regarding the developmental trajectories of EF development, it has been found that individual differences in EFs are relatively stable from middle childhood to adolescence (Harms, Zayas, Meltzoff, & Carlson, 2014). However, there are children who progress more or less as compared with their peers of the same age. Therefore, the current study also sought to illuminate possible interindividual variance and to detect variables that might influence the magnitude of individual development of WM and DM. Among other variables, gender, socioeconomic status (SES), verbal ability, and processing speed have been found to predict, or to be associated with, EFs in children and adolescents (Carlson et al., 2013). In the IGT, boys quite consistently outperform girls (Crone and van der Molen, 2007, Overman, 2004), whereas findings regarding gender differences in WM are mixed (e.g., Boelema et al., 2014, Bull et al., 2008, Gur et al., 2012). Both WM and DM are positively related to children’s SES (Cauffman et al., 2010, Hackman et al., 2015, Mata et al., 2013). Verbal ability and processing speed positively correlate with WM (Fry and Hale, 2000, Hughes, 1998), but associations with DM seem to be weak (Toplak, Sorge, Benoit, West, & Stanovich, 2010).

So far, only few studies have used latent growth curve analysis to investigate predictors for EF development in children or adolescents, and these studies have primarily focused on cool EFs. For instance, verbal ability and SES have been found to predict individual differences in the initial levels of a latent EF factor, which included measures of WM, planning, and inhibitory control, in 4-year-old children (Hughes et al., 2010). However, only verbal ability predicted the rate of change from 4 to 6 years of age, such that children with relatively lower verbal abilities showed steeper growth in EF ability. In an age period from 3 to 5 years, parallel latent growth curve modeling indicated positive effects of the rate of change in verbal ability on the rate of change in a latent factor of EF, which included measures of WM, inhibitory control, and flexibility (Kuhn, Willoughby, Vernon-Feagans, Blair, & Family Life Project Key Investigators, 2016). Similar to Hughes et al. (2010), other studies also found that family SES had a positive effect on the initial levels of WM, but SES did not predict the rates of change through the following years: from preschool (about 4.5 years) to the elementary school years (Hackman et al., 2015) and from early adolescence (mean age = 11 years) to late adolescence (mean age = 19 years) (Boelema et al., 2014). Boelema et al. (2014) also found that girls had better initial WM levels than boys at 11 years of age but that boys had steeper growth curves from than girls at 11 to 19 years. Taken together, growth curve modeling has helped to identify some variables that influence both the initial level and developmental trajectories of WM and cool EFs in general. However, there is a lack of longitudinal analyses during middle childhood and of studies focusing on hot EF facets such as DM.

In the current study, therefore, we examined the development of WM, as a typical facet of cool EFs, and DM, as a typical facet of hot EFs, over a 3-year period during middle childhood, comparing a younger cohort (6–7 years at T1) and an older cohort (8–9 years at T1) of children. Based on data from three waves, we used latent growth curve analysis that allows not only examining average developmental trajectories but also detecting and predicting interindividual differences in the initial levels and rates of change (Duncan & Duncan, 2009).

Although the prior studies briefly summarized above led to mixed findings, we expected higher initial levels of performance for WM and DM in the older cohort than in the younger cohort. We also expected developmental improvements of WM and DM across the 3-year period in both cohorts. Because of prior inconsistent findings regarding the developmental trajectories in different age groups, we analyzed whether the two cohorts would differ in their rates of change of WM and DM. However, because the two cohorts overlap in their age across the 3-year period, comparisons of growth curves should be interpreted tentatively and in conjunction with descriptive data on a mean level.

In a second step, we aimed at getting a better picture of the variables that predict EF development during middle childhood. We used a parallel process latent growth model of WM and DM, and in addition to age (within the cohorts) we included SES, verbal ability, processing speed, and gender as predictors for both the initial levels and rates of change. We expected SES to predict initial levels of WM and DM in both cohorts (e.g., Cauffman et al., 2010, Hughes et al., 2010). Furthermore, we expected verbal ability and processing speed to predict initial levels of WM but not of DM (Fry and Hale, 2000, Hughes, 1998, Toplak et al., 2010). Moreover, we expected gender to predict initial levels of both DM and WM in both cohorts (Boelema et al., 2014, Crone and van der Molen, 2007). Regarding the rate of change, we did not expect a predictive effect of SES on WM change (Boelema et al., 2014, Hackman et al., 2015, Hughes et al., 2010). Findings regarding the predictive effect of verbal ability on the rate of change in WM were mixed (Hughes et al., 2010, Kuhn et al., 2016), and we did not find prior reports on the effects of SES or verbal ability on DM trajectories, so we examined these effects exploratively. Likewise, we did not find prior reports on the effects of processing speed on the rate of change of WM or DM, so we investigated these potential effects in an explorative fashion.

Concerning a potential moderating effect of age, Carlson et al. (2013) noted that EFs differentiate increasingly with age from other cognitive skills, so there is some reason to expect that the predictive effects of processing speed and verbal ability on WM would be smaller in the older cohort as compared with the younger cohort. We did not expect the other predictive effects to differ between cohorts.

Finally, we examined the interrelation of both EF facets at the initial level and at the rate of change. Based on the evidence reviewed by Toplak et al. (2010), we expected no relations, or just small relations, between the two facets of EFs, and we used our data for a further test of that prediction.

Section snippets

Participants and procedure

The current study was part of a larger, three-wave longitudinal project on intrapersonal risk factors during childhood and adolescence. Prior publications have used data from the first wave (e.g., Groppe & Elsner, 2014) and second wave (e.g., Groppe & Elsner, 2015). However, these publications did not focus on EF development and did not include data from the third wave of the current study.

Participants were recruited from 33 elementary schools in Brandenburg, Germany. Although schools were

Results

Table 1 presents means, standard deviations, and ranges of the continuous variables for both cohorts. Fig. 1 displays mean WM and DM scores for both cohorts by mean age. It can be seen that, although cohorts did not overlap in their age at T1, there was an age overlap across the 3-year period, with a similar mean age in the younger cohort at T2 as in the older cohort at T1 and in the younger cohort at T3 as in the older cohort at T2. Descriptively, Fig. 1 illustrates continuous growth in both

Discussion

Previous research has indicated that middle childhood is an important period for EF development, but longitudinal research exploring developmental trajectories of cool and hot facets of EFs is still scarce. Therefore, we examined growth curves of WM, as a typical facet of cool EFs, and DM, as a typical facet of hot EFs, during middle childhood. WM was measured using a backward digit span task (Petermann & Petermann, 2008), and DM was measured using a simplified 60-trial version of the Hungry

Acknowledgments

This work was supported by the German Research Foundation (DFG) under Grant GRK 1668/1. The study is based on the PIER study, which is a longitudinal project at the University of Potsdam funded by the DFG. Other articles have emerged from the project using data from the first two measurement waves (e.g., Groppe and Elsner, 2014, Groppe and Elsner, 2015). We thank the schools and children for participating in the study and thank all team members for contributing to the large project. We also

References (52)

  • M.E. Toplak et al.

    Decision-making and cognitive abilities: A review of associations between Iowa Gambling Task performance, executive functions, and intelligence

    Clinical Psychology Review

    (2010)
  • J. Audusseau et al.

    Working memory in children predicts performance on a gambling task

    Journal of Genetic Psychology

    (2015)
  • P.B. Baltes et al.

    Life-span developmental psychology

    Annual Review of Psychology

    (1980)
  • A. Bechara et al.

    Dissociation of working memory from decision making within the human prefrontal cortex

    Journal of Neuroscience

    (1998)
  • J.R. Best et al.

    A developmental perspective on executive function

    Child Development

    (2010)
  • S.-J. Blakemore et al.

    Decision-making in the adolescent brain

    Nature Neuroscience

    (2012)
  • S.R. Boelema et al.

    Executive functioning shows differential maturation from early to late adolescence: Longitudinal findings from a TRAILS study

    Neuropsychology

    (2014)
  • K.C. Brocki et al.

    Executive functions in children aged 6 to 13: A dimensional and developmental study

    Developmental Neuropsychology

    (2004)
  • R. Bull et al.

    Short-term memory, working memory, and executive functioning in preschoolers: Longitudinal predictors of mathematical achievement at age 7 years

    Developmental Neuropsychology

    (2008)
  • S.M. Carlson et al.

    Executive function

  • N. Carriedo et al.

    Development of the updating executive function: From 7-year-olds to young adults

    Developmental Psychology

    (2016)
  • E. Cauffman et al.

    Age differences in affective decision making as indexed by performance on the Iowa Gambling Task

    Developmental Psychology

    (2010)
  • E. Crone et al.

    Characterization of children’s decision making: Sensitivity to punishment frequency, not task complexity

    Child Neuropsychology

    (2005)
  • E. Crone et al.

    Developmental changes in real life decision making: Performance on a gambling task previously shown to depend on the ventromedial prefrontal cortex

    Developmental Neuropsychology

    (2004)
  • E. Crone et al.

    Development of decision making in school-aged children and adolescents: Evidence from heart rate and skin conductance analysis

    Child Development

    (2007)
  • I.N. Defoe et al.

    A meta-analysis on age differences in risky decision making: Adolescents versus children and adults

    Psychological Bulletin

    (2015)
  • Cited by (0)

    View full text