Abstract

This cross-sectional study examined the development of set-shifting ability from childhood into early adulthood. Six hundred and forty-nine participants (aged 8–30) were assessed on the verbal fluency, design fluency, trail making, color-word interference, and card sorting subtests of the Delis–Kaplan Executive Function System (D-KEFS). Multiple regression analyses revealed modest effects of age and gender on set-shifting tasks, after controlling for IQ and component skills. The current study provides evidence for generally increased performance of set-shifting abilities through adolescence. Women overall had statistically better performance than men on all executive functioning tasks. There were significant age by gender interactions suggesting differential age-related improvements between men and women. On color-word interference and verbal fluency switching tasks, men tended to show larger improvements than women, whereas on a design fluency switching task, women showed larger improvements than men.

Introduction

Executive functions have received a great deal of attention in the past decade. These are often described as goal-directed behaviors, requiring planning, flexibility, impulse control, and self-monitoring (Barkley, 1997, 2001; Welsh, Pennington, & Grossier, 1991). These sets of skills and behaviors are often associated with functioning in the prefrontal areas of the brain (Stuss, 1992), and development of executive functions is believed to mirror the concurrent maturation of the various sections of the frontal lobes. Studies indicate that the regions of the frontal lobes, like other brain areas, begin developing prenatally; however, maturation of the regions of the frontal lobes occurs later relative to other brain regions (Casey, Giedd, & Thomas, 2000), thus delaying the development of executive functions (Stuss, 1992). As cognitive capacity increases throughout childhood, there are pronounced changes in frontal lobe structure (Blakemore & Choudhury, 2006; Durston et al., 2001). Frontal gray matter increases until puberty (usually between 10 and 12 years of age) and then decreases during adolescence (Blakemore & Choudhury, 2006; Giedd et al., 1999; Stuss, 1992), presumably allowing for greater specification of cognitive functions (Casey et al., 2000; Stuss, 1992). White matter increases throughout childhood, presumably due to greater myelination, which corresponds with faster and more efficient processing (Blakemore & Choudhury, 2006).

Presumably as a result of significant growth in the prefrontal areas prior to puberty, the years between childhood and adolescence are heightened periods for the development of executive functions (Brocki & Bohlin, 2004; Welsh et al., 1991; Luciana et al., 2005). Given the complexities involved in frontal lobe development that continue through adolescence (i.e., changes in gray and white matter, declines in synaptic density), the corresponding development of executive functions is hypothesized to progress gradually, with variability in emergence and acquisition of domains (e.g., working memory, set-shifting, etc.) within executive functions (Denckla, 1994; Luciana et al., 2005). Though many researchers have pointed toward the link between frontal lobe development and increases in executive function abilities (Brocki & Bohlin, 2004; Casey et al., 2000; Welsh et al., 1991), it is only recently that the development of specific domains within executive functions, such as set-shifting and working memory, has been studied. In addition, most studies that examine executive function development have neglected to address development into early adulthood (Anderson, Anderson, Northam, Jacobs, & Catroppa, 2001; Conklin, Luciana, Hooper, & Yarger, 2007; Huizinga & van der Molen, 2007; Rueda et al., 2004; Welsh et al., 1991).

Set-shifting is an example of a complex executive ability, involving the ability to alternate between different response sets (Wecker, Kramer, Hallmam, & Delis, 2005). It is considered a domain of executive functions because of the self-directive action involved. Furthermore, set-shifting does not appear to significantly overlap with other cognitive constructs (Kramer et al., 2007). Huizinga and van der Molen (2007) looked at set-shifting and set maintenance abilities on the Wisconsin card sort and found that adult-level performance was achieved at 11 and 15 years of age, respectively. Though this study provides preliminary evidence for the development of shifting abilities in adolescence, it has been recommended that because of the multifaceted nature of set-shifting paradigms, it is also necessary to assess performance on component tasks, or tasks that measure basic-level ability presumed to be invoked during more complex cognitive functions (Kramer et al., 2007). Furthermore, when measuring more complex cognitive tasks, it is important to isolate basic-level skills (e.g., sequencing letters) that contribute to complex skills (e.g., alternating sequences between sequences of numbers and letters; Anderson et al., 2001).

Though executive functions are often related to ability on basic cognitive tasks, they do not entirely overlap with general intellectual ability, thus warranting separate measurement and study of executive functions with respect to IQ (Delis et al., 2007; Welsh et al., 1991). Delis and colleagues (2007) found that subsets of youth had significantly higher IQ compared with executive function abilities, whereas others had significantly higher executive function abilities compared with IQ scores. Thus, it is important to parcel out the contribution of general intellectual and basic-level abilities from higher level executive functions.

Gender differences in executive functions may be linked to differences in frontal lobe development. Although at full adult development, men have approximately 10% greater overall brain volume than women, there are significant differences in developmental trajectories due to gender during adolescence (Caviness, Kennedy, Richelme, Rademacher, & Filipek, 1996; Giedd et al., 1999). Although boys and girls both show linear increases in overall white matter through adolescence, on average girls tend to reach maximum gray matter density in the frontal lobes earlier than boys (11 years of age for girls compared with 12.1 years of age for boys; Giedd et al., 1999). This earlier relative gray matter peak corresponds to a subsequent earlier relative reduction in frontal lobe volume in women during adolescence. Given these findings, it is also possible that women may also show earlier development of executive functions relative to men.

Giedd and colleagues (1999) have suggested that structural brain differences between men and women during adolescence may be related to pubertal onset, with earlier relative pubertal onset and gray matter synaptogenesis in women. Gender-related differences in brain structure may also be the result of different locations of androgen and estrogen receptors in the brain (Caviness et al., 1996; Durston et al., 2001). There is evidence suggesting that men have significant age-related increases in white matter volume in the left inferior frontal gyrus (Blakemore & Choudhury, 2006) and the amygdala (Durston et al., 2001) as a result of the onset of greater steroid levels in puberty, a pattern that is not evident with women. Thus, though women may initially exhibit better performance on executive function tasks, men may show better executive function improvements after the onset of puberty.

The aim of this study was to determine age- and gender-related differences on set-shifting tasks from late childhood into early adulthood, during a period of time when more complex executive function performances are presumed to emerge. The current study addresses limitations in the previous research by studying a specific domain of executive functions (set-shifting), accounting for measures of more basic-level abilities, and including late childhood through early adult participants. Executive function performance was measured on tasks that have been found to be sensitive to frontal lobe injury and required either cognitive switching (set-shifting) or some other type of mental flexibility (Baldo et al., 2001; Kramer et al., 2007). To assess the amount of variance uniquely attributable to set-shifting, IQ and basic-level abilities (i.e., the respective component tasks involved in each switching task) were first factored out. We predicted that performance would increase with age until adulthood on each set-shifting task. Although it was hypothesized that there would be a differentially better early performance for women versus men, it was predicted that men would show larger improvement with increasing age than women (i.e., steeper learning curve), thereby reducing the gender difference across ages.

Materials and Methods

Participants

A sample of 649 participants (302 men and 347 women), between the ages of 8 and 30, was drawn from the Delis–Kaplan Executive Function System (D-KEFS) standardization study (Delis, Kaplan, & Kramer, 2001). The standardization study was recruited to be representative of the U.S. population according to the 2000 U.S. Census; gender, race/ethnicity, years of education, and geographic region were stratified by age group. The exclusion criterion for the standardization study was endorsement of at least one medical or psychiatric symptom that might affect performance on cognitive tests. Further information regarding the standardization study can be found in the D-KEFS Examiner's Manual (Delis et al., 2001).

Participants in the current study were further broken down by age group: 8–9 (n = 90), 10–11 (n = 92), 12–13 (n = 94), 14–15 (n = 92), 16–17 (n = 78), 18–24 (n = 103), and 25–30 (n = 100). The mean age was 16.11 years (SD = 6.36 years) and 10.2% were African American, 79% Caucasian, 8% Hispanic, and 2.8% other. This subset of the standardization sample for this age range was restricted by the requirement that concurrent intellectual data were available.

Materials and Procedure

Examiners with experience in psychometric testing, certification, and licensing administered participants the “Wechsler Abbreviated Intelligence Scale” (WASI; Wechsler, 1999) and the D-KEFS (Delis et al., 2001). The WASI was designed to provide an abbreviated measure of intellectual functioning (full scale IQ) and correlates with other full measures of intelligence (e.g., the WISC-III and the WAIS-III; Wechsler, 1999). The WASI consists of four subtests (block design, vocabulary, similarities, and matrix reasoning) and takes approximately 45 min to administer. The D-KEFS was designed as a comprehensive measurement of executive functioning abilities. The following five D-KEFS tests were selected for use in this study: Trail making, design fluency, verbal fluency, color-word interference, and card sorting. These tests were selected because they were intended to measure set-shifting ability and/or mental flexibility (Delis et al., 2001). Four of these tests incorporated switching conditions: Trail making, design fluency, verbal fluency, and color-word interference. The standard versions of all tests were given. The five selected D-KEFS tests take approximately 45 min to administer.

The trail making test is a motor task consisting of a visual cancellation task and a series of connect-the-circle tasks (Delis et al., 2001). The switching condition of the trail making test requires alternating connections between number and letter sequences. Thus, the two component tasks of the trail making test included number sequencing and letter sequencing. These tasks require the participant to make serial connections between numbers and then letters, respectively.

The verbal fluency test is a verbal task consisting of three conditions: Letter fluency, category fluency, and category switching. The category switching task involves alternating between naming fruits and furniture in a 60 s time span. Thus, the corresponding component task selected for this study was category fluency (animal naming), which consists of naming animals in a 60 s time span.

The design fluency test was designed as a nonverbal analogue to the verbal fluency test and consists of three conditions: Filled dots, empty dots, and switching. The switching condition necessitates making designs by alternating between filled and unfilled dots; therefore, both filled and empty dots were considered component tasks for the switching condition. These component tasks require the participant to draw designs using only “filled dots” or “unfilled dots.”

The color-word interference test was designed to be similar in structure to the Stroop test (Stroop, 1935) and consists of four conditions: color naming, word reading, inhibition, and switching. The switching task of the color-word interference test requires the participant to switch between saying a dissonant ink color and saying a printed word. The component tasks for the switching condition of the color-word interference test in this study were considered to be word reading and inhibition. The word reading task consists of reading words. In the inhibition task, the participant reads the dissonant ink color and does not name the word.

The card sorting test is a passive set-shifting task that assesses both verbal and visual recognition; it consists of two conditions: Free sorting and sort recognition.

The card sorting test requires the participant to describe each of eight sorting rules of two sets of six cards that can be sorted into two groups, with three cards in each group. The sort recognition description score was included in the analyses, as it is a primary measure of the card sorting test intended to measure flexibility in thinking. Further descriptions of these measures can be found in the D-KEFS manual (Delis et al., 2001).

Analyses

The purpose of this study was to determine the effects of age and gender on set-shifting performance. A series of multi-staged, forced entry multiple regression analyses were performed for raw scores of each dependent variable (performance on switching task or card sorting task). In each model, the first step consisted of full scale IQ. The second step included performance on the related component task(s) as previously outlined (e.g., for verbal fluency switching, performance on category fluency or “animal naming” was considered a component task). Inclusion of component tasks in the model prior to age and gender was intended to partial out the variance attributable to basic-level ability. Component tasks were included in the analyses for all tasks, with the exception of the card sorting test, which did not have a relevant component task. In the third step, gender and age as a continuous variable were entered. Thus, age- and gender-related performance in set-shifting ability could be considered independent of intellectual and basic-level abilities. Finally, the fourth step consisted of the interaction between age and gender. All continuous variables in the model were centered prior to analysis, and a simple slopes analysis was conducted for all significant interactions according to recommendations by Holmbeck (1997). For each model, collinearity statistics (variance inflation factor [VIF] and tolerance) were also examined to ensure that analyses were not significantly affected by collinearity between the independent variables in the models. VIF and tolerance statistics indicated that there was not multi-collinearity between the variables in each model.

Results

Results for the five regression analyses are presented in Table 1. In the first step of the regression analysis regarding the trail making test, IQ significantly predicted switching performance, F(1, 642) = 66.69, p < .001, accounting for 9% of the variance. The component tasks for the trail making test in the second step, number sequencing and letter sequencing, accounted for an additional 42% of the variance in switching ability, ΔF(2, 640) = 275.35, p < .001. In the third step, age and gender accounted for 2% additional variance, ΔF(2, 638) = 10.27, p < .001. Age was a significant predictor of switching performance (β = −.13, p < .001), with better performance (i.e., faster times) with increasing age. Gender was also a significant predictor (β = −.06, p < .05), with women performing better overall than men. Fig. 1 depicts age-related performance on trail making test by gender and by task (number sequencing, letter sequencing, switching). The age by gender interaction in the fourth step was not a significant predictor of trail making switching, p > .05.

Fig. 1.

Trail making test performance.

Fig. 1.

Trail making test performance.

Table 1.

Standardized regression coefficients and R2 values for regression analyses (n = 645)

 Trail making
 
Design fluency
 
Verbal fluency
 
Color word interference
 
Card sorting
 
 β R2 β R2 β R2 β R2 β R2 
Step 1: IQ −.31*** .09 .29*** .09 .27*** .08 −.15*** .02 .49*** .24 
Step 2: IQ −.16*** .51 .17*** .34 .08* .30 −.03 .62 .49*** .24 
 Component task .26*** .13** .51*** .20*** — 
 (Component task) .49*** .42*** — .64*** — 
Step 3: IQ −.19*** .53 .20*** .38 .17*** .39 −.04 .63 .51*** .32 
 Component task .22*** .09 .34*** .18*** — 
 (Component task) .44*** .36*** — .58*** — 
 Age −.13*** .24*** .33*** −.13*** .27*** 
 Gender −.06* .07* .14*** .01 .10**      
Step 4: IQ −.19*** .53 .20*** .39 .17*** .40 −.04 .64 .51*** .32 
 Component task .22*** .09 .34*** .18*** — 
 (Component task) .44*** .36*** — .58*** — 
 Age −.11** .30*** .27*** −.08* .27*** 
 Gender −.06* .07* .14*** .01 .10** 
 Age × Gender −.03 −.09* .10* −.07* .01 
 Trail making
 
Design fluency
 
Verbal fluency
 
Color word interference
 
Card sorting
 
 β R2 β R2 β R2 β R2 β R2 
Step 1: IQ −.31*** .09 .29*** .09 .27*** .08 −.15*** .02 .49*** .24 
Step 2: IQ −.16*** .51 .17*** .34 .08* .30 −.03 .62 .49*** .24 
 Component task .26*** .13** .51*** .20*** — 
 (Component task) .49*** .42*** — .64*** — 
Step 3: IQ −.19*** .53 .20*** .38 .17*** .39 −.04 .63 .51*** .32 
 Component task .22*** .09 .34*** .18*** — 
 (Component task) .44*** .36*** — .58*** — 
 Age −.13*** .24*** .33*** −.13*** .27*** 
 Gender −.06* .07* .14*** .01 .10**      
Step 4: IQ −.19*** .53 .20*** .39 .17*** .40 −.04 .64 .51*** .32 
 Component task .22*** .09 .34*** .18*** — 
 (Component task) .44*** .36*** — .58*** — 
 Age −.11** .30*** .27*** −.08* .27*** 
 Gender −.06* .07* .14*** .01 .10** 
 Age × Gender −.03 −.09* .10* −.07* .01 

Notes: Trail making test component tasks: Number sequencing, letter sequencing, respectively; design fluency test component tasks: Filled dots, unfilled dots, respectively; verbal fluency test component task: Animal naming; color word interference test component tasks: Word reading, inhibition, respectively.

*p < .05.

**p < .01.

***p < .001.

With regard to the regression analysis for design fluency test, IQ significantly predicted switching performance in the first step, F(1, 647) = 60.25, p < .001, accounting for 9% of the variance. The second step, which included the two components task for the design fluency test, connecting filled and unfilled dots, accounted for an additional 21% of the variance in switching ability, ΔF(2, 645) = 234.97, p < .001. In the third step, age and gender accounted for an additional 5% of the variance, ΔF(2, 643) = 26.39, p < .001. Age was a significant predictor of switching performance (β = .24, p < .001), with better performance with increasing age. Gender was also a significant predictor (β = .07, p < .05), with women performing better overall than men. The age by gender interaction in the fourth step was also significant, and contributed an additional 1% of the variance in switching performance, ΔF(1, 642) = 4.66, p < .05. Fig. 2 depicts age-related performance on design fluency switching by gender and task (filled and unfilled dots). As evidenced in Fig. 2, during the 12–13 and 14–15-year age groups, girls showed an increased switching performance, whereas men in the 18–24 age group showed an increased performance.

Fig. 2.

Design fluency test performance.

Fig. 2.

Design fluency test performance.

In the first step of the regression analysis from the verbal fluency test, IQ contributed 8% of the variance in switching ability, ΔF(1, 644) = 52.43, p < .001. In the second step, the component task, animal naming, significantly predicted switching performance, ΔF(1, 643) = 202.13, p < .001, accounting for an additional 22% of the variance. Age and gender were entered in the third step and accounted for an additional 9% of the variance in switching performance. Age was a significant predictor of switching (β = .33, p < .001), with better performance with increasing age. Gender was significant (β = .14, p < .001), with women performing better overall than men. The age by gender interaction, entered in the fourth and final step, was also significant, ΔF(1, 640) = 4.94, p < .05, and accounted for an additional 1% of the variance. Fig. 3 depicts age-related performance on verbal fluency switching by gender and task (category fluency). In the 14–15-year-old age group, boys showed a steeper rate of improvement (β = .40) than do girls (β = .26).

Fig. 3.

Verbal fluency test performance.

Fig. 3.

Verbal fluency test performance.

In the first step of the regression involving the color-word interference test, IQ significantly predicted switching performance F(1, 645) = 15.56, p < .001, accounting for 2% of the variance. The second step, which included the component tasks, word reading and inhibition, accounted for an additional 60% of the variance in switching ability, ΔF(2, 643) = 518.47, p < .001. In the third step, age and gender accounted for an additional 1% of the variance, ΔF(2, 641) = 9.18, p < .001. Gender was not a significant predictor of switching performance (β = −.01, p > .05); however, age was a significant predictor (β = −.13, p < .001), with switching time improving with age. The age by gender interaction in the fourth step was a significant predictor of color-word switching performance (ΔF(1, 640) = 3.89, p < .05), accounting for an additional 1% of the variance. Fig. 4 depicts age-related performance on the color-word interference switching task. Men showed improvement with increasing age (β = −.17) when compared with women (β = −.08).

Fig. 4.

Color-word interference test performance.

Fig. 4.

Color-word interference test performance.

In the first step of the regression analysis involving the card sorting test, IQ significantly predicted sorting performance, F(1, 645) = 201.83, p < .001, accounting for 24% of the variance. In the second step, age and gender accounted for 8% additional variance, ΔF(2, 643) = 38.59, p < .001. Performance increased with age, with greater number of card sort descriptions with greater age (β = .27, p < .05), and women outperforming men (β = .10, p < .05). Age-related performance on the card sorting test is presented in Fig. 5. The age by gender interaction in the fourth step was not a significant predictor of card sorting description score, p > .05.

Fig. 5.

Card sorting description score performance.

Fig. 5.

Card sorting description score performance.

To examine differences between age groups on switching tasks, post hoc Bonferroni t-tests were conducted for each task and presented in Table 2. These analyses showed that for the trail making test switching task, verbal fluency test switching task, and card sorting test description score, significant differences between age groups were found between the 8–9-year-old group and 10–11-year-old group, as well as the 10–11 year olds and the 12–13 year olds (p < .05). However, non-significant age differences were found for these tasks after the 12–13-year-old group (p > .05). On the design fluency test switching task, a significant age difference was found between the 10–11-year-old group and 12–13-year-old group (p < .05); non-significant differences were found between the 8–9-year-old group and 10–11-year-old group and the 14–15-year-old group and older (p > .05). For the color-word interference test switching task, a significant differences were found between groups for the 8–9-year-old group, the 10–11-year-old group, the 12–13-year-old group, and the 14–15-year-old group (p < .05), but not for any age groups after the 14–15-year-old group (p > .05).

Table 2.

Post-hoc Bonferroni comparisons between age groups for switching tasks

Dependent variable 8–9 vs. 10–11 10–11 vs. 12–13 12–13 vs. 14–15 14–15 vs. 16–17 16–17 vs. 18–24 18–24 vs. 25–30 
Trail making .000* .000* .193 1.00 .794 1.00 
Design fluency .330 .007* 1.00 .062 1.00 1.00 
Verbal fluency .042* .002* 1.00 .832 1.00 1.00 
Color-word interference .000* .000* .039* 1.00 1.00 1.00 
Card sorting .001* .001* 1.00 1.00 .221 1.00 
Dependent variable 8–9 vs. 10–11 10–11 vs. 12–13 12–13 vs. 14–15 14–15 vs. 16–17 16–17 vs. 18–24 18–24 vs. 25–30 
Trail making .000* .000* .193 1.00 .794 1.00 
Design fluency .330 .007* 1.00 .062 1.00 1.00 
Verbal fluency .042* .002* 1.00 .832 1.00 1.00 
Color-word interference .000* .000* .039* 1.00 1.00 1.00 
Card sorting .001* .001* 1.00 1.00 .221 1.00 

Note: *p < .05.

Discussion

It is generally accepted that executive functions increase with age, and thus performance on executive function tasks also improves with age; however, there has been limited research to date to support this idea and some ambiguity with regard to attainment of executive function skills. For example, much of the literature has focused on the age at which adult-level performance is reached (Casey et al., 1997; Huizinga & van der Molen, 2007; Welsh et al., 1991), failing to consider changes in performance into adulthood. Given the results of brain imaging studies that indicate neural connectivity in the frontal lobes continues into adulthood (Casey et al., 2000), studies on the development of executive functions need to incorporate this breadth of age.

The current study provides evidence for increased performance across age on all five executive function tasks, as well as significant gender differences on four tasks (trail making test switching, verbal fluency test switching, design fluency test switching, and card sorting recognition). Across these four tasks, women generally had greater executive function abilities in set-shifting and flexibility than men. These results are contrasted with previous research, which has been mixed on the presence of gender differences in executive function abilities, with some researchers reporting non-significant gender differences on verbal fluency, planning, and organizing abilities (Welsh et al., 1991), and others reporting greater female performance on verbal fluency and working memory tasks (Anderson et al., 2001). However, it is important to note that these previous studies have examined what is referred to in this study as the component task of verbal fluency (i.e., naming words beginning with a specific letter or within a specific category), not the more complex task of verbal fluency involving set-shifting. Additionally, unlike these previous studies, the present study controlled for the influence of IQ and component tasks, suggesting that gender differences emerge within more complex tasks involving set-shifting and cognitive flexibility.

Moderation analyses in the current study indicate that men and women have differential executive function developmental trajectories across ages. Although women generally showed greater performance across executive functioning tasks, three out of the five tasks showed significant age by gender interaction effects which explained more of the variance in set-shifting than IQ, component tasks, age, and gender alone. With regard to Design fluency switching performance, there were no age effects between ages 8–9 and 10–11, but across ages of 12–13 and 14–15 girls showed a faster rate of improvement in this area of complex visual cognition than their male counterparts. Girls continued to outperform boys on the design fluency switching task until the 16–17 years age group, at which point boys' performance is greater than girls.

Interestingly, a different pattern emerged for the verbal fluency and color-word interference switching tasks. On the verbal fluency switching task, in the younger age groups (i.e., 8–9 and 10–11-year olds), girls' performance was initially greater than that of boys; however, across the ages of 10–11 and 12–13 years boys showed a faster rate of improvement than girls. Similarly, on the color-word interference switching task boys showed a faster rate of improvement between the ages of 10–11 and 12–13 years. These results are consistent with later relative pubertal changes in men and significant increases in frontal white matter for boys during adolescence (Blakemore & Choudhury, 2006; Reiss et al., 1996), which may delay the emergence of complex executive functions and increased performance relative to earlier developing women. As has been suggested by previous studies (e.g., Blakemore & Choudhury, 2006), gender differences in performance may be the result of pubertal hormonal changes acting on receptors in the brain; however, such results are speculative and warrant further study.

It is also possible that the different developmental trajectories across switching tasks are related to varying levels of demand for each task that were not assessed for in the current study. For example, the trail making test requires a demand for fine motor ability that tasks of the verbal fluency test do not. The current study revealed varying rates of improvement between component tasks, which may suggest that differences in set-shifting trajectories may be due to non-executive functioning demands across tasks.

The results of the current study are consistent with previous studies that have argued that the executive functions increase from childhood into adolescence. Similar to Huizinga and van der Molen (2007), performance on set-shifting tasks improved into adolescence; however, these researchers found that performance on set-shifting tasks was not significantly different between 15 year olds and 21 year olds, which led them to conclude that improvements within this domain did not continue until adulthood. However, a limitation in their study was that performance on component tasks was not controlled for, making it difficult to determine whether the plateau in performance was due to the effects of more basic skills, such as perceptual and motor abilities. The present study found evidence for improvements on set-shifting tasks into early adolescence age groups, after controlling for IQ and performance on related component tasks.

IQ accounted for 24% of the variance in performance on card sorting, indicating that general intellectual ability is highly related to cognitive flexibility on this task. In comparison, IQ accounted for 2%–9% of the variance across set-shifting tasks, whereas component tasks accounted for an additional 22%–60% of the variance. Age and gender explained an additional 2%–4% of the variance in set-shifting. Thus, while performance on set-shifting tasks is highly dependent on basic-level ability, the results suggest that set-shifting performance did improve independently of skills on component tasks due to differences in age and gender. The relatively small effect of age and gender can further be explained by the finding that the majority of set-shifting tasks (e.g., trail making, design fluency, verbal fluency, and card sorting) did not show age differences in performance past age 13 for rate or speed of production. Age-related differences on the color-word interference test switching task continued into the 14–15 years age group, at which point age differences were no longer significant. It is likely that performance on this task showed later development due to a higher task load required for this task (i.e., inhibition was a component task for the switching task).

Performance on executive function tasks follows a gradual progression from child to adolescent age groups. Although it was predicted that executive functioning would show improvements into early adulthood, the current study found significant changes only through middle adolescence. Although the frontal lobes may continue to undergo structural changes into adulthood, it is possible that skill development does not follow the same developmental progression. Furthermore, it is likely that different domains of executive functioning show different trajectories of development. For example, previous studies on working memory have indicated that this skill develops until middle to late adolescence (Conklin et al., 2007). The present study found that other executive domains, such as set-shifting, tend to improve only until early adolescence. Because these more complex mental capabilities rely heavily on performance on more basic skills (e.g., visual scanning, verbal fluency, etc.), it is also necessary to account for changes within these component areas. The current study provides evidence for the cumulative and interacting effect of basic skill level, age, and gender in understanding the construct of executive functioning.

The finding that women perform better than men on some set-shifting tasks indicates that flexibility is generally better for women; however, this is interpreted with caution because men also tend to show better improvements over time in set-shifting ability for verbal and visual tasks. This latter finding suggests that men may develop this executive ability later than women, but then “catch up.” This finding has important implications for parents and educators, who may benefit from the knowledge that some men may have more difficulty in transitioning between different tasks in a classroom and may require greater structure within classrooms into high school.

Although this study contributes to the accumulating research on the development of executive functions, several limitations need acknowledgement. The present study focused on set-shifting performance; however, differences may exist in other executive function domains such as behavioral inhibition and complex attention. Further research is needed to investigate other domains of executive functioning. The differences found in the literature on what constitutes an executive function are indicative of the difficulty in measuring the constructs involved. Although there is greater variability in performance for younger children, and therefore modest reliability of the D-KEFS tasks for this age group, the D-KEFS was used in the present study because it is one of few existing comprehensive batteries of executive functioning validated for children through adult ages. The D-KEFS also offers an assessment of the component skills of executive functions (Homack, Lee, & Riccio, 2005).

The current study found evidence for increased performance on set-shifting tasks into adolescence; however, the amount of variance accounted for by age differed by task. It is possible that the tasks used measured different aspects of set-shifting ability (e.g., verbal, nonverbal, etc.). Furthermore, a large amount of variance in set-shifting ability was explained by performance on component tasks. Future research might include other measures of both basic-level abilities and executive functions to investigate the exact nature of this relationship. Another limitation of the current study is the lack of data on pubertal status. Although differential improvements found for men and women may have been the result of pubertal changes, these conclusions are speculative. The current study utilized a cross-sectional design, which limits the ability to draw conclusions about individual change over time. The use of a longitudinal study design in future research might be able to provide more information about the effect of individual changes over time on executive functioning development. Nevertheless, the present results may be viewed as an intriguing survey of set-shifting ability development from late childhood through early adulthood.

Conflict of interest

Dr. Delis receives royalties from the sales of the D-KEFS. Dr. Holdnack is an employee of Pearson.

Acknowledgements

Standardization data were from the Delis–Kaplan Executive Function System (D-KEFS). Copyright 2001 by Pearson, Inc. Used with permission. All rights reserved.

References

Anderson
V. A.
Anderson
P.
Northam
E.
Jacobs
R.
Catroppa
C.
Development of executive functions through late childhood and adolescence in an Australian sample
Developmental Neuropsychology
 , 
2001
, vol. 
20
 (pg. 
385
-
406
)
Baldo
J. V.
Shimamura
A. P.
Delis
D. C.
Kramer
J.
Kaplan
E.
Verbal and design fluency in patients with frontal lobe lesions
Journal of the International Neuropsychological Society
 , 
2001
, vol. 
7
 (pg. 
586
-
596
)
Barkley
R. A.
Behavioral inhibition, sustained attention, and executive functions: Constructing a unifying theory of ADHD
Psychological Bulletin
 , 
1997
, vol. 
121
 (pg. 
65
-
94
)
Barkley
R. A.
The executive functions and self-regulation: An evolutionary neuropsychological perspective
Neuropsychology Review
 , 
2001
, vol. 
11
 (pg. 
1
-
29
)
Blakemore
S. J.
Choudhury
S.
Development of the adolescent brain: Implications for executive function and social cognition
Journal of Child Psychology and Psychiatry
 , 
2006
, vol. 
47
 (pg. 
296
-
312
)
Brocki
K. C.
Bohlin
G.
Executive functions in children aged 6 to 13: A dimensional and developmental study
Developmental Neuropsychology
 , 
2004
, vol. 
26
 (pg. 
571
-
593
)
Casey
B. J.
Giedd
J. N.
Thomas
K. M.
Structural and functional brain development and its relations to cognitive development
Biological Psychology
 , 
2000
, vol. 
54
 (pg. 
241
-
257
)
Casey
B. J.
Trainor
R. J.
Orendi
J. L.
Schubert
A. B.
Nystrom
L. E.
Giedd
J. N.
A developmental functional MRI study of prefrontal activation during performance of a go-no-go task
Journal of Cognitive Neuroscience
 , 
1997
, vol. 
9
 (pg. 
835
-
847
)
Caviness
V. S.
Jr
Kennedy
D. N.
Richelme
C.
Rademacher
J.
Filipek
P. A.
The human brain age 7–11 years: A volumetric analysis based on magnetic resonance images
Cerebral Cortex
 , 
1996
, vol. 
6
 (pg. 
726
-
736
)
Conklin
H. M.
Luciana
M.
Hooper
C. J.
Yarger
R. S.
Working memory performance in typically developing children and adolescents: Behavioral evidence of protracted frontal lobe development
Developmental Neuropsychology
 , 
2007
, vol. 
31
 (pg. 
103
-
128
)
Delis
D. C.
Kaplan
E.
Kramer
J. H.
Delis–Kaplan executive function system: Examiner's manual.
 , 
2001
San Antonio, TX
The Psychological Corporation
Delis
D. C.
Lansing
A.
Houston
W. S.
Wetter
S.
Han
S. D.
Jacobson
M.
, et al.  . 
Creativity lost: The importance of testing higher-level executive functions in school-age children and adolescents
Journal of Psychoeducational Assessment
 , 
2007
, vol. 
25
 (pg. 
29
-
40
)
Denckla
M. B.
Lyon
G. R.
Measurement of executive function
Frames of reference for assessment of learning disabilities: New views on measurement issues
 , 
1994
Baltimore
Brooks
(pg. 
117
-
142
)
Durston
S.
Hilleke
E.
Hulshoff
P.
Casey
B. J.
Giedd
J. N.
Buttlelaar
J. K.
, et al.  . 
Anatomical MRI of the developing human brain: What have we learned?
Journal of the American Academy of Child and Adolescent Psychiatry
 , 
2001
, vol. 
40
 (pg. 
1012
-
1020
)
Giedd
J. N.
Blumenthal
J.
Jeffries
N. O.
Castellanos
F. X.
Liu
H.
Zijdenbos
A.
, et al.  . 
Brain development during childhood and adolescence: A longitudinal MRI study
Nature Neuroscience
 , 
1999
, vol. 
2
 (pg. 
861
-
863
)
Holmbeck
G. N.
Toward terminological, conceptual, and statistical clarity in the study of mediators and moderators: Examples from the child-clinical and pediatric psychology literatures
Journal of Consulting and Clinical Psychology
 , 
1997
, vol. 
65
 (pg. 
599
-
610
)
Homack
S.
Lee
D.
Riccio
C. A.
Test review: Delis–Kaplan Executive Function System
Journal of Clinical and Experimental Neuropsychology
 , 
2005
, vol. 
27
 (pg. 
599
-
609
)
Huizinga
M.
van der Molen
M. W.
Age-group differences in set-switching and set-maintenance on the Wisconsin card sort task
Developmental Neuropsychology
 , 
2007
, vol. 
31
 (pg. 
193
-
215
)
Kramer
J. H.
Quitania
L.
Dean
D.
Neuhaus
J.
Rosen
H. J.
Halabi
C.
Magnetic resonance imaging correlates of set shifting
Journal of the International Neuropsychological Society
 , 
2007
, vol. 
13
 (pg. 
386
-
392
)
Luciana
M.
Conklin
H. M.
Hooper
C. J.
Yarger
R. S.
The development of nonverbal working memory and executive control processes in adolescents
Child Development
 , 
2005
, vol. 
76
 (pg. 
697
-
712
)
Reiss
A. L.
Abrams
M. T.
Singer
H. S.
Ross
J. L.
Denckla
M. B.
Brain development, gender and IQ in children: A volumetric imaging study
Brain
 , 
1996
, vol. 
119
 (pg. 
1763
-
1774
)
Rueda
M. R.
Fan
J.
McCandliss
B.
Halparin
J. D.
Gruber
D. B.
Pappert
L.
Development of attentional networks in childhood
Neuropsychologia
 , 
2004
, vol. 
42
 (pg. 
1029
-
1040
)
Stroop
J. R.
Studies of interference in serial verbal reaction
Journal of Experimental Psychology
 , 
1935
, vol. 
18
 (pg. 
643
-
662
)
Stuss
D. T.
Biological and psychological development of executive functions
Brain and Cognition
 , 
1992
, vol. 
20
 (pg. 
8
-
23
)
Wechsler
D.
Wechsler abbreviated scale of intelligence: Examiner's manual.
 , 
1999
San Antonio, TX
The Psychological Corporation
Wecker
N. S.
Kramer
J. H.
Hallam
B. J.
Delis
D. C.
Mental flexibility: Age effects on switching
Neuropsychology
 , 
2005
, vol. 
19
 (pg. 
345
-
352
)
Welsh
M. C.
Pennington
B. F.
Grossier
D. B.
A normative-developmental study of executive function: A window on prefrontal function in children
Developmental Neuropsychology
 , 
1991
, vol. 
7
 (pg. 
131
-
149
)