A Longitudinal Examination of the Developmental Executive Function Hierarchy in Children With Externalizing Behavior Problems

Abstract

Objective: Using a 4-year longitudinal design, we evaluated two hypotheses based on developmental executive function (EF) hierarchy accounts in a sample of children with externalizing problems. Method: The participants performed EF tasks when they were between 8 and 12 years (M = 9.93), and again approximately 4 years later when they were between 12 and 15 years (M = 13.36). Results: Inhibition in middle childhood predicted working memory (WM) 4 years later. Further, deficits in inhibition and sustained attention were more prominent in middle rather than late childhood, whereas poor WM was salient throughout these periods. Conclusions: These findings support the hypotheses that EFs develop hierarchically and that EF deficits in ADHD are more prominent in actively developing EFs. They also emphasize ADHD as a developmental disorder.

Keywords

executive function development hierarchy ADHD longitudinal

Executive functions (EFs) are required in situations where it would be incorrect or inappropriate according to a long-term goal to let external stimuli guide behavior. EFs are thus linked to internal self-control of thoughts and actions. These functions include (but are not limited to) the ability to inhibit habitual but incorrect responses or irrelevant information (i.e., inhibitory control), to maintain information active in consciousness and work with the information to obtain a goal (i.e., working memory [WM]), and to flexibly switch between two or more mental sets or tasks (i.e., mental set-shifting; for example, Miyake et al., 2000). Recent developmental literature in the EF area has highlighted the need to move beyond the focus of describing developmental trajectories of EFs, and instead address mechanisms of development, answering questions such as, “Does the early phase of development of one component facilitate the development of other components?” (Best & Miller, 2010, p. 1653).

There are several EF theories emphasizing such developmental hierarchies, all of which have in common the idea that complex EFs characterized by protracted developmental curves build on simple ones with earlier developmental periods. Bjorklund and Harnishfeger (1990) have proposed that as children mature, increased efficiency in inhibitory processing leads to improved selective attention, which in turn leads to increased WM resources available for task-relevant information. In addition, Dempster (1992) suggested that inhibition is a primitive, basic processing dimension, the development of which enables children to acquire higher order knowledge. Barkley’s (1997) theory is one of the most influential theories defined by a hierarchical perspective on EF development. He proposed that the essential first step in the development of EFs is when children are able to inhibit their initial responses, providing a delay in which more complex EFs can operate. Although the majority of these theories primarily focus on the developmental relation between inhibition and WM, generalizations could perhaps also be made to sustained attention as a simple function and mental set-shifting as a more complex function. For example, Diamond (2006) suggested that mental set-shifting partly builds on inhibitory capacity, which is necessary to be able to inhibit the mental set and the associated response that one is shifting away from.

Importantly, neither of these theories are specific about the particular periods when these developmental relations between simple and complex EFs should be most readily observable. However, it could be argued that as long as the EFs under study are actively developing, their developmental relations should be detectible. For example, inhibition generally shows active developmental periods between the early preschool years and about 11 years (e.g., Brocki & Bohlin, 2004; Garon, Bryson, & Smith, 2008; Levin et al., 1991; Romine & Reynolds, 2005), whereas WM and mental set-shifting that also start developing relatively early show a slower and more protracted curve reaching adult maturation in middle adolescence (e.g., Garon et al., 2008; Gathercole, Pickering, Ambridge, & Wearing, 2004; Huizinga, Dolan, & van der Molen, 2006; Luciana, Conklin, Hooper, & Yarger, 2005; Luciana & Nelson, 1998). Thus, inhibition assessed during early and middle childhood should be predictable of WM at any succeeding point in development up until middle adolescence.

Barkley (1997) suggested that his theory of the developmental mechanisms of EFs (described above) applies to both typical development and development in children with ADHD, emphasizing that children with ADHD show a similar but delayed EF development. Importantly, according to Barkley, the developmental EF hierarchy should also have consequences for the view on the development of EF deficits in ADHD. The relative size of the deficits caused by the developmental delay should change during development, as specific EFs show active development during different periods of childhood. Deficits in simple, earlier developing EFs should thus be more prominent in early and middle childhood (than later in development), and should also adversely impact the development of complex EFs. The deficits in complex EFs, in turn, should persist further up in the development as these abilities continue to develop for a longer time. In addition, Barkley suggests that for deficits to be found, the relevant abilities must have reached sufficient maturation in normal children (otherwise even normal children will perform poorly). Thus, because the trajectories are more protracted for complex EFs, deficits in these functions may not be evident as early as deficits in simple EFs. Generally, there is support for the main EFs, inhibition, WM, and shifting being deficient in school-aged children with ADHD (Willcutt, Doyle, Nigg, Faraone, & Pennington, 2005).

It has been discussed whether EF deficits are specific to ADHD or whether other externalizing disorders that often coexist with ADHD (i.e., oppositional defiant disorder [ODD] and conduct disorder [CD]; Biederman, 2005) also show poorer EF performance. Although a few studies have shown EF deficits specifically in ODD/CD (i.e., even when controlling for ADHD; Hobson, Scott, & Rubia, 2011; Séguin, Boulerice, Harden, Tremblay, & Pihl, 1999; Toupin, Déry, Pauze, Mercier, & Fortin, 2000), the majority of the literature indicates that if EF deficits are found in ODD/CD, these are accounted for by comorbid ADHD symptoms (Barkley, Edwards, Laneri, Fletcher, & Metevia, 2001; Berlin & Bohlin, 2002; Brocki & Bohlin, 2006; Brocki, Nyberg, Thorell, & Bohlin, 2007; Clark, Prior, & Kinsella, 2000; Déry, Toupin, Pauze, Mercier, & Fortin, 1999; Hummer et al., 2011; Oosterlaan, Scheres, & Sergeant, 2005; Schoemaker et al., 2012; Thorell & Wåhlstedt, 2006; Wåhlstedt, 2009; Wåhlstedt, Thorell, & Bohlin, 2008; Waschbusch, 2002). Therefore, children with ODD/CD behavior problems were included in the comparison group in this study.

Two related hypotheses, which have received little empirical attention, can be derived from the hierarchical developmental EF theories. First, individual differences—within both the clinical and normal range—in inhibition and sustained attention in early and middle childhood should set the stage for the development of WM and mental set-shifting that occur throughout childhood and adolescence. Second, and consequently, deficits in inhibition and sustained attention in ADHD should be more prominent in early and middle childhood (approximately until 11 years), whereas WM and mental set-shifting deficits should be salient throughout middle childhood and adolescence, although the effects in middle childhood may be weaker if sufficient maturation has not yet been reached in the comparison children.

The findings in a longitudinal study of a sample at risk for externalizing behavioral problems were generally in line with the first prediction, although this study concerned preschool ages. The findings showed that a simple form of inhibition and selective attention at age 5 predicted more complex inhibition and WM at age 6 (Brocki, Eninger, Thorell, & Bohlin, 2010). Furthermore, using a normal sample, Berlin, Bohlin, and Rydell (2003) demonstrated that inhibitory control at 5 years was longitudinally related to WM at 8.5 years. The results of a previous cross-sectional study on a normal sample of 6- to 13-year-olds were partly in line with the second prediction of a developmentally sensitive relation between EF and ADHD symptoms (Brocki & Bohlin, 2006). Inhibition was found to only be related to ADHD symptoms up until approximately 10 years, whereas WM was related to ADHD symptoms only in the older ages (10-13 years). Increasing the validity of the testing of the hypotheses derived from the hierarchical developmental EF theories requires the use of a longitudinal design with relatively long time span and the inclusion of children diagnosed with ADHD.

The Present Study

The aim was to examine two hypotheses based on the proposed developmental EF hierarchy (Barkley, 1997; Bjorklund & Harnishfeger, 1990; Dempster, 1992) using a longitudinal design in a sample of children with the externalizing behavior problems of ADHD, ODD, and CD. Participants were assessed with EF tasks when they were between 8 and 12 years and approximately 4 years later when they were between 12 and 15 years.

First, using an individual difference approach across the entire sample, we evaluated the hypothesis that complex EFs showing prolonged active development throughout childhood and adolescence (WM and mental set-shifting) build on simpler EFs that show an earlier period of active development (sustained attention and inhibition). Support for this hypothesis would be provided by results showing that individual variation in simpler EFs at Time 1 (T1; M age = 9.93, that is, middle childhood) predicted more complex EFs at Time 2 (T2; M age = 13.36 years, that is, early adolescence), in combination with results showing that individual differences in complex EFs at T1 did not predict simpler ones at T2. The testing of this hypothesis was thus made in terms of the 4-year within-person development between T1 and T2. The age variation between individuals was controlled for and our conclusions hence regarded the mean age level at each time point.

Second, we explored Barkley’s (1997) hypothesis linking the developmental EF hierarchy to ADHD. Specifically, we studied whether deficits in sustained attention and inhibition are more prominent in ADHD in middle childhood (i.e., at T1) than early adolescence (i.e., at T2). We further predicted that deficits in WM and mental set-shifting should be salient throughout middle childhood and adolescence (i.e., at both T1 and T2), however, we also recognized the possibility that these deficits may be weaker in middle childhood due to insufficient maturation (see argument by Barkley, 1997). Furthermore, to study whether our first hypothesis suggesting that simple EFs underlie the development of complex EFs extends to neuropsychiatric EF deficits, we also studied whether ADHD deficits in simple EFs at T1 could explain these children’s later deficits in complex EFs at T2. Deficits were defined as poorer performance in children with a diagnosis of ADHD than in children with externalizing behavior problems but no diagnosis of ADHD.

Method

Participants

The Bergen Child Study is a prospective and longitudinal study of children’s mental health and development from primary school age to adolescence (Heiervang et al., 2007; Stormark, Heiervang, Heimann, Lundervold, & Gillberg, 2008). The first wave of the study was organized in three phases, where Phase 1 comprised screening of symptoms of mental health problems, Phase 2 comprised a diagnostic evaluation based on a parent interview, and Phase 3 a comprehensive clinical and neuropsychological assessment (for further description of the inclusion criteria in Phase 3, see Lundervold, Posserud, Ullebø, Sørensen, & Gillberg, 2011; Sørensen, Plessen, & Lundervold, 2012).

This study is a 4-year longitudinal follow-up of a subsample of children showing high symptom-levels (defined below) of an externalizing behavioral disorder (i.e., ADHD, ODD, and CD) in Phase 3. Two of the participating children were identified as having mental retardation and were therefore excluded from the sample. One additional child was much younger than the other children (7.11 years at T1 and 11.11 years at T2) which resulted in an overlap in the ages included in T1 and T2. This child was therefore removed from the data set. The final sample comprised 47 children (79% boys; 59% of the invited children). The children’s age ranged from 8.00 to 11.90 years (M = 9.93, SD = 1.01 years) at T1 and from 12.00 to 15.01 years (M = 13.36, SD = 0.86 years) when assessed at the follow-up approximately 4 years later (M interval= 3.70, SD = 0.49 years) at T2.

Diagnosis of Externalizing Behavior Disorders

Both at Phase 3 (T1 in the present study) and the follow-up 4 years later (T2 in the present study), diagnostic status regarding ADHD, ODD, and CD was assessed using a psychiatric diagnostic interview (the Kiddie-Sads-Present and Lifetime Version [K-SADS-PL]; Kaufman et al., 1997) generating Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association [APA], 1994) diagnoses. Licensed psychologists or psychiatrists, trained in using the instrument and blind to the screening protocols, conducted the interview, first with the parent(s) and later the same day with the child. The inclusion criterion in this study was high symptom-levels of externalizing behavioral disorders in Phase 3, which was defined by one or more definite symptoms of ADHD, ODD, or CD following the K-SADS algorithm (http://www.psychiatry.pitt.edu/sites/default/files/Documents/assessments/ksads-pl.pdf).

Regarding diagnostic status, the K-SADS-PL algorithm defines a disorder as being either (a) not present, (b) possibly present, (c) in remission, or (d) definitely present. To conservatively test group differences between children having and not having ADHD (according to the second aim), children receiving a definite ADHD diagnosis formed the ADHD group, and the comparison group consisted of all other children in the sample. Twenty-four and 22 children, respectively, were diagnosed with ADHD at T1 and T2. At T1, 9 children were diagnosed with ADHD-primarily inattentive subtype (ADHD-IA) and 15 with ADHD-combined subtype (ADHD-C). At T2, 13 children were diagnosed with ADHD-IA, 1 with ADHD-predominantly hyperactive-impulsive subtype (ADHD-H/I), 6 with ADHD-C, and 2 with ADHD-not otherwise specified. Due to the relatively long time span some children changed their diagnostic status between T1 and T2 (n = 8). For the analyses concerning the second goal involving differences between the ADHD and comparison group, the choice was made to only include children who remained in the same group across the two time points (n_{ADHD group} = 19; n_{comparison group} = 20). For further details on the diagnostic status of the participants in the two groups (including comorbidity), see Table 1.

Table 1.

Diagnostic Status of the ADHD Group (n = 19; 95% Boys) and Comparison Group (n = 20, 60% Boys) at T1 (M age = 9.93 Years) and T2 (M age = 13.36 Years).

	T1		T2
	ADHD group	Comparison group	ADHD group	Comparison group
Definite ADHD	19	0	19	0
Possible ADHD	—	12	—	5
In remission from ADHD	—	0	—	2
Definite ODD	5	4	5	0
Possible ODD	2	2	1	3
In remission from ODD	0	0	2	1
Definite CD	1	1	0	0
Possible CD	0	1	0	0
In remission from CD	0	0	1	0
Comorbid diagnosis (i.e., ODD and/or CD)	5	1	6	0

Note: ODD = oppositional defiant disorder; CD = conduct disorder. Values Indicate Number of Children.

Tasks and Measures

Sustained Attention

At both T1 and T2, sustained attention was measured by the Conners’ Continuous Performance Test II (CCPT-II; Conners, 2000), where the participants were instructed to respond to all letters (n = 360) presented on the computer screen, except for the letter X (n = 36), for 14 min. Split-half reliabilities and test–retest reliabilities for a 3-month interval have been shown to range between .73 and .95 and .55 and .84, respectively, for the different performance measures on the CCPT-II (Conners, 2000). Both indices suggest adequate reliability for this test. Measures from this task have been shown to be valid and show adequate variability at least to the age of 17 years (Conners, Epstein, Angold, & Klaric, 2003). The test is divided into six time blocks, which enables calculation of the slope of change in performance across time on task. This has been suggested to be essential if conclusions are to be drawn regarding sustained attention (Nigg, 2006; Van der Meere & Sergeant, 1988). At both T1 and T2, composite scores were calculated averaging standardized measures of standard error of reaction time (RT) on hits (RTSE; that is, RT variability), the slope of change of hit RT, and of hit RTSE (internal consistency for the three measures was .70 at T1 and .60 at T2). Lower values on the composite score indicate better ability. The values on this variable of children having a z score above 3 on failures to respond on go-trials were replaced by the sample mean (n = 2 at both T1 and T2).

Inhibition

Inhibition was assessed by two very similar paper versions of the Stroop (1935) task. A version developed by Lund-Johansen, Hugdahl, and Wester (1996) was used at T1, and the Delis–Kaplan Executive Function System (D-KEFS) Color-Word Interference test (Delis, Kaplan, & Kramer, 2001) was used at T2. In Block 1, the participants were presented with color patches and were asked to name the colors as fast as possible. In Block 2, color words printed in black were presented and the participants were asked to read the words as fast as they could. In Block 3, color words printed in incompatible color (e.g., the word BLUE printed in red color) were presented and the participants were supposed to name the colors (and ignore the word) as fast as possible. This incompatible condition was expected to produce a response conflict requiring inhibition. At both T1 and T2, inhibition was indexed by a variable representing number of errors made in the color-word incompatible condition, controlled for number of errors made in the color patch condition. This index was calculated as the residuals derived from a regression analysis with errors in incompatible condition as dependent variable and errors in color patch condition as independent variable. Split-half reliability for the D-KEFS Color-Word Interference Test has been reported in the range from .62 to .86 in different age groups (Homack, Lee, & Riccio, 2005). Lower values on this measure indicate better ability.

WM

At T1, WM capacity was assessed using the backward version of the Digit Span subtest of the Wechsler Intelligence Scale for Children (WISC-III; Wechsler, 1992). Participants were supposed to recall and immediately reproduce sequences of digits in the reversed order. The sequences of digits increased from two to nine with two trials at each sequence length. The test stopped when the participant was no longer able to correctly reproduce at least one of the two trials at a particular sequence length. This WISC-III subtest has been reported to show a reliability of .85 (Wechsler, 1992). The dependent variable was the number of correct trials.

At T2, WM was assessed by the Letter-Number Sequencing (LNS) task from WISC-IV (Wechsler, 2003). The child was presented with sequences of digits and letter (e.g., E-8-A-6) and was instructed to repeat first the digits in numerical order and then the letters in alphabetical order. The sequences increased from three to eight items, with three trials at each sequence length. The test stopped when the child was no longer able to answer correctly on at least one of the three trials at a particular sequence length. Internal consistency has been shown to be .90 and test–retest reliability was .75 (across age groups; Wechsler, 2003). The dependent variable on this task was the number of correct trials. Importantly, the two WM tasks come from the same paradigm, involving simultaneous maintenance and manipulation of information (Wechsler, 2007). These two tasks loaded onto the same WM factor in the WISC-IV battery, and showed a relatively high correlation in Swedish validation samples (r = .41; Wechsler, 2007), supporting the similarity of the operationalization of WM at T1 and T2.

Mental Set-Shifting

At T1, mental set-shifting was assessed by the Children’s Color Trails test (Llorente, Williams, Satz, & D’Elia, 2003). In the first block of this task (control block), the children were presented with a card containing 15 colored circles with numbers. The instruction was to start at the circle containing the number “1,” and to connect the circles with a pencil in ascending order as quickly and accurately as possible. In the second block (test block), each number was presented within both a pink and a yellow circle. The children were to start at the circle containing “1” in pink and make a trail, alternately connecting numbers in ascending order in pink and yellow (i.e., 1 in pink, 2 in yellow, 3 in pink, etc.) as quickly and accurately as possible. The test block thus involves constantly switching mental set between the tasks “connect numbers in pink” and “numbers in yellow.” The child was told when making an error and was instructed to correct it before continuing. The control block and test block, respectively, show test–retest reliability for a 4-month interval of .68 and .60 (Llorente et al., 2003). Total time to complete the test block, controlled for the time to complete the control block, derived as regression residuals, was used as the dependent measure.

At T2, mental set-shifting was measured by the switching condition in D-KEFS Color-Word Interference test (see description of the test above; Delis et al., 2001). The child was presented with incompatible color words printed on a page, 50% of which were bordered by a black rectangle. The child was instructed to alternate between naming the ink color of the incompatible color-word when the word was not surrounded by a rectangle, and reading the word when the word was surrounded by a rectangle. To obtain a pure measure of set-shifting from this task (which was also used to assess inhibition), regression residuals were derived by removing the variance in the time to complete the switching condition that was accounted for by the time to complete the incompatible block (when the incompatible ink color should always be reported, that is, the condition assumed to mainly assess inhibition). Lower values on the measures of mental set-shifting indicate better ability. The two mental set-shifting tasks both target the type of shifting defined by the task-switching paradigm (Cragg & Chevalier, 2012). In tasks within this paradigm, participants must alternate between tasks, based on either sequence position information (the Children’s Color Trails test) or cues (the switching condition in the D-KEFS Color-Word Interference test).

Statistical Analyses

The hypothesis concerning simple EFs as developmental predictors of complex EFs across the entire sample was studied using correlation analysis. The primary interest in the EF intercorrelations were (a) the relations of sustained attention and inhibition at T1 with WM and mental set-shifting at T2 and (b) the relations of WM and mental set-shifting at T1 with sustained attention and inhibition at T2.

Before studying the aim concerning developmental differences in ADHD-related EF deficits, we first ensured the nondeficient nature of the comparison group by comparing their EF performance at T1 (using independent t tests) to a subgroup of the children from Phase 3 with no psychiatric behavior problems (as assessed by the K-SADS-PL; n = 139; 59% boys; M_age = 9.75 years). The comparison group did not differ significantly from the subgroup of normal children on any of the studied EFs at T1 (ts < 1.21, ps > .22). This supports the nondeficient nature of our comparison group.

The ADHD-related aim was studied in separate one-way ANOVAs on T1 data and T2 data, with group (ADHD vs. comparison group) as independent variable and the different EFs as dependent variables. Note that these analyses were only conducted with the 39 children who remained in the same group (ADHD and comparison, respectively) across T1 and T2. Finally, we studied whether ADHD deficits in simple EFs at T1 could account for T2 deficits in more complex EFs, by conducting ANCOVAs with group (ADHD vs. comparison) as independent variable, each of the complex EFs measured at T2 as dependent variable, and each of the simple EFs at T1 as covariate.

All analyses concerning ADHD group differences were conducted with as well as without control for diagnoses of ODD and CD (in two separate variables) to more closely disentangle the role of comorbidity in our findings. Results are reported without control for ODD/CD, and any differences in findings with this control are noted in the text and tables. Because the aims concerned longitudinal predictions/comparisons, the age variation between individuals at each time point was controlled for in all above described analyses concerning both the first and second aim, by entering age at T1 and/or T2 (depending on analysis) as covariate/s.

Results

All variables met the standard criteria for univariate normality: skewness < 3 and kurtosis < 10 (Kline, 2005). Skewness values were <1.86, and kurtosis values were <5.09. No bivariate or multivariate outliers defined by Cook’s D > 1 were identified in any of the conducted analyses. Missing data were replaced by the mean on that particular variable. Between 1 and 2 missing values for each variable were replaced at T1, and between 0 and 5 missing values per variable were replaced at T2. Age was only significantly correlated with sustained attention at T2 (r = −.32, p < .05; other rs = .00-.20), indicating poorer sustained attention in younger children. However, to completely dismiss the probability that between-individual age variation could affect the longitudinal interpretations, age was controlled for in all analyses (as described in the “Statistical Analyses” section). There were no significant effects of sex on any of the cognitive variables, ts(45) = 0.03-1.07, ns.

Simple EFs as Developmental Predictors of Complex EFs in the Entire Sample

Descriptive data on all measures at T1 and T2 are presented in Table 2. Note that several measures are constructed as composite scores or statistical residuals, making the absolute values less descriptive. Table 3 presents intercorrelations between all T1 and T2 measures. Regarding the longitudinal predictions of more complex EFs by simpler EFs, Table 3 shows that inhibition at T1 predicted WM but not mental set-shifting at T2. Sustained attention at T1 predicted neither WM nor mental set-shifting at T2. Importantly, neither WM nor mental set-shifting assessed at T1 significantly predicted sustained attention or inhibition assessed at T2 (see Table 3).

Table 2.

Descriptive Statistics and ANOVA Results for the Measures Used at T1 (M age = 9.93 Years) and T2 (M age = 13.36 Years).

	Total	ADHD group	Comparison group
	M (SD)	LSM (SE)	LSM (SE)
	N = 47	n = 19^a	n = 20^a	F	Partial η²
T1 measures
Sustained attention	0.00^b (0.81)	0.22 (0.18)	−0.27 (0.17)	3.98^c	.10
Inhibition	0.00^b (5.63)	2.61 (1.30)	2.27 (1.27)	7.23*	.17
Working memory	3.61 (1.51)	3.31 (0.35)	4.31 (0.37)	4.16*	.10
Mental set-shifting	0.00^b (24.25)	6.24 (5.82)	−3.90 (5.67)	1.55	.04
T2 measures
Sustained attention	0.00^b (0.74)	0.13 (0.17)	−0.07 (0.17)	0.69	.02
Inhibition	0.00^b (2.91)	0.38 (0.69)	−0.30 (0.68)	0.49	.01
Working memory	5.43 (2.10)	4.96 (0.47)	6.33 (0.46)	4.39*	.11
Mental set-shifting	0.00^b (27.45)	7.94 (6.17)	−4.47 (6.01)	2.07^d	.05

Note: LSM = least square means (controlling for age). F values in italics are significant with control for ODD and CD.

Only children who remained in the same group across T1 and T2 were included in the analyses of group differences (see “Method” section).

This measure was either a composite score (average) of standardized variables or it was calculated as statistical residuals, both of which by default have a mean of 0.

p = .054. When controlling for ODD/CD, F = 4.29, p < .05, partial η² = .11.

This group difference became significant (F = 4.22, p < .05, partial η² = .11) when controlling for ODD/CD.

p < .05. **p < .01. ***p < .001.

Table 3.

Partial Intercorrelations (Controlling for Age) Between Executive Function Tasks at T1 (M age = 9.93 Years) and T2 (M age = 13.36 Years).

	T1				T2
	1	2	3	4	5	6	7	8
T1 measures
1. Sustained attention	—	.39**	−.26	−.14	.12	.09	−.18	.02
2. Inhibition		—	−.36*	.40**	.19	.19	−.34*	.20
3. Working memory			—	−.39**	−.02	−.18	.65***	−.30*
4. Mental set-shifting				—	−.17	.26	−.38**	.21
T2 measures
5. Sustained attention					—	.04	.22	−.01
6. Inhibition						—	−.23	.19
7. Working memory							—	−.30*
8. Mental set-shifting								—

Note: Correlations important for the hypothesis testing are marked in italics (N = 47).

p < .05. **p < .01. ***p < .001.

Developmental Change in the EF Deficits of Children With ADHD

Least square means and standard errors for the ADHD group and comparison group as well as ANCOVA results (controlling for age) are presented in Table 2. At T1, children with ADHD performed significantly worse than children without this diagnosis on inhibition, and WM, but not on mental set-shifting. In addition, the group difference in sustained attention was borderline significant, but became significant when controlling for ODD and CD diagnoses. At T2, the ADHD group performed significantly worse only on WM, and when controlling for ODD and CD diagnoses also the group difference in mental set-shifting became significant. Although the effects were generally stronger with control for ODD and CD, there were no other differences in significance level. Importantly, the effect sizes for the significant effects described above were much larger than the effect sizes for the nonsignificant effects.

Finally, we examined whether ADHD deficits in simple EFs at T1 could account for these children’s later deficits in complex EFs at T2. The significant group difference in T2 WM (see Table 2) disappeared both when controlling for T1 sustained attention, F(1, 34) = 2.87, p > .05, partial η² = .08, and T1 inhibition, F(1, 34) = 1.25, p > .05, partial η² = .04. However, inhibition seemed to account for most of the effect of sustained attention, because adding sustained attention together with inhibition as covariates did not further decrease the size of the group effect in WM, F(1, 33) = 1.19, p > .05, partial η² = .04. In contrast, the group difference in T2 mental set-shifting (that was significant when controlling for ODD and CD) was still significant (and the effect size did not decrease) both when controlling for T1 inhibition, F(1, 33) = 4.07, p = .052, partial η² = .11, and T1 sustained attention, F(1, 33) = 6.06, p < .05, partial η² = .16, as well as when controlling for both, F(1, 32) = 4.35, p < .05, partial η² = .12.

Discussion

Two hypotheses rooted in the theoretically proposed developmental EF hierarchy (Barkley, 1997; Bjorklund & Harnishfeger, 1990; Dempster, 1992) were evaluated in a sample of children with externalizing behavior problems (i.e., ADHD, ODD, and/or CD). First, based on individual differences across the whole sample, we examined whether simpler EFs that primarily show active maturation during early and middle childhood predicted the development of more complex EFs that continue to mature throughout childhood and adolescence. Second, we explored Barkley’s (1997) ADHD hypothesis, suggesting that ADHD-related deficits in simple EFs should be more prominent in middle childhood than early adolescence, and also that they should developmentally underlie the deficits in complex EFs which in turn should be salient throughout the studied periods.

Simple EFs as Developmental Predictors of Complex EFs in the Whole Sample

We received some support for the theoretical proposition that simple EFs underlie the development of complex EFs. We found that inhibition in middle childhood (i.e., at T1) predicted WM in early adolescence (i.e., at T2), and importantly that there were no longitudinal relations going in the opposite direction; that is, from the more complex to simpler EFs. Interestingly, the support for the developmental EF hierarchy was obtained in terms of the specific relation between inhibition and WM, the functions focused on in the theoretical accounts (Barkley, 1997; Bjorklund & Harnishfeger, 1990; Dempster, 1992). This raises the question of whether the EF hierarchy is specific to the developmental relation between these two key EFs or whether other EF components than the ones studied here may underlie the development of mental set-shifting.

Although the few extant previous empirical studies (Berlin et al., 2003; Brocki et al., 2010) have found some support for the developmental EF hierarchy hypothesis in terms of longitudinal correlations between earlier simple EFs and later complex EFs, the design of these studies did not allow for ruling out the possibility that the reverse was also true (i.e., that complex EFs earlier in development predict simple EFs later on). Showing this specificity in relational direction is pivotal for the validity of the support. In addition, we have extended the previous results, which were obtained for younger children (5 to 8.5 years in Berlin et al., 2003; 5 to 6 years in Brocki et al., 2010) even further by showing that the hierarchical developmental relations are also found later in development and across a 4-year period. Although dimensional accounts on neuropsychological conditions (such as Barkley’s, 1997, model of ADHD) emphasize quantitative rather than qualitative differences in traits/cognitive functions between clinical and normative populations, our findings evidently need to be verified also in typically developing samples before any firm conclusions on the developmental EF hierarchy can be drawn.

Due to the variability in age between individuals in the present study, we acknowledge the limitation in not being able to pinpoint narrower age periods for the developmental relations stipulated by the theory. We also emphasize that although there were only weak correlations between age and EFs (i.e., between-individual effects), which may be due to the sample including children (of different ages) both with and without EF deficits, the EFs of each child are still clearly presumed to develop from one time point to another regardless of deficits (see, for example, Barkley, 1997). This within-person development was the basis for our conclusions.

Developmental Change in the EF Deficits of Children With ADHD

Based on the developmental EF hierarchy, Barkley (1997) has emphasized the similarities in the mechanisms of typical EF development and the development of ADHD-related EF deficits. That is, because simpler EFs mature more rapidly earlier on in development, deficits in these functions should be more salient in early and middle childhood, whereas deficits in more complex functions should be evident all through their active period of development, that is, throughout childhood and adolescence. Our results were largely in line with these propositions. The effect sizes for the ADHD-related deficits in inhibition and sustained attention were larger (and only significant) at T1 (middle childhood) than at T2 (early adolescence), whereas the strength of the effect for poorer WM ability in children with ADHD seemed to be constant (and significant) throughout middle childhood and early adolescence (at both T1 and T2). Two previous cross-sectional studies on normal samples (6-13 years and 6-16 years, respectively) showed that the association between poorer WM and ADHD symptoms only was significant in age groups older than 9.7 years (Brocki & Bohlin, 2006; Tillman, Eninger, Forssman, & Bohlin, 2011). However, their youngest age groups included 6- to 7-year-olds, which may be too young for WM to be sufficiently developed in children without ADHD symptoms to reveal poorer performance in children with such symptoms (see Barkley, 1997).

We also found that the effect size for the deficit in mental set-shifting (when controlling for ODD/CD) was larger (and only significant) at T2 than at T1. This finding is partly inconsistent with our prediction that deficiencies in the complex EFs under study should be evident at both T1 and T2 when they are actively developing. At the same time, as emphasized by Barkley (1997), specific EF components must be sufficiently developed before deficits can be detected. Our finding could thus potentially mean that mental set-shifting is not sufficiently developed for a deficit to appear before early adolescence. This interpretation is supported by Diamond’s (2006) suggestion that mental set-shifting is at the top of the EF hierarchy, representing the essence of executive control, and may thus be an even more complex function than WM.

We also crucially showed that the WM deficits at T2 in children with ADHD were accounted for by the children’s poorer inhibitory control as well as sustained attention at T1, although this effect was mainly driven by inhibitory control. This suggests that deficits in WM developmentally build on deficits in inhibition. Our finding thus expands models of typical developmental EF hierarchies (Barkley, 1997; Bjorklund & Harnishfeger, 1990; Dempster, 1992) by showing that such developmental associations also apply to the neuropsychiatric EF deficits associated with ADHD. Interestingly, neither early deficits in inhibition nor sustained attention (or both) accounted for the later observed deficits in mental set-shifting, again raising the question of whether some other EF components may underlie the development of shifting ability and deficits therein.

Although the heterogeneous constitution of the comparison group is a limitation in the present study, we were indeed able to show that the comparison group did not differ from children with no psychiatric behavior problems on the studied EFs. Although it is still possible that a normal comparison group would yield overall stronger effects, we argue that the size of the effects relative to each other (which is the focus of our aim) would still be similar. However, replications using a normal comparison group will nonetheless be needed before any firm conclusions can be drawn. Furthermore, although it would be theoretically ideal to have a purer ADHD group, this is practically difficult to achieve, and the ecological validity of such a group would be poor.

Controlling for the comorbidity in the ADHD and comparison group generally resulted in stronger effects. However, only the effect for mental set-shifting at T2 went from nonsignificant without this control to significant with this control. This suggests that the ADHD diagnosis was specifically important for the group differences, specifically regarding mental set-shifting at T2.

Although we have argued that the WM and mental set-shifting tasks used at T1 tap the same EF domain as the tasks used at T2 (see “Method” section), it is important to acknowledge the limitation in not having the exact same measures at these two time points. Being able to statistically ensure measurement invariance is always valuable, and it is perhaps particularly so for examining the developmental EF hierarchy. This might result in different conclusions than ours. However, the maximization of measurement invariance in longitudinal design is complex; in some cases, the use of the exact same tasks is necessary, but in other cases, it is critical due to developmental issues to slightly adjust the difficulty of the tasks. Therefore, future empirical attempts to test the developmental EF hierarchy should be careful to choose tasks that capture processes as similar as possible at different developmental periods. Furthermore, the general downside of using a single task approach to assess each EF is that the full construct can then never be captured without error. Using another set of EF tasks could potentially produce results different than ours. We therefore also recommend future studies taking on this research question to consider a latent-variable approach, yielding more robust and error-free measures of each EF at the construct level.

Summary and Conclusion

The present study examined two hypotheses based on the theoretically proposed developmental EF hierarchy (Barkley, 1997; Bjorklund & Harnishfeger, 1990; Dempster, 1992) in a sample of children with externalizing behavior problems. We received some support for the hypothesis that complex EFs build on simpler ones by showing that inhibition in middle childhood (8-12 years) predicted WM 4 years later. In accordance with the second hypothesis suggesting that EF deficits in ADHD should be more prominent in the specific EFs that are actively developing at a certain period (Barkley, 1997), we concluded that deficits in the simpler EFs, sustained attention and inhibition, are more prominent in middle childhood than in late childhood/early adolescence, whereas poor WM appears to be salient throughout these periods. Deficits in mental set-shifting were only evident in late childhood/early adolescence, which may perhaps suggest that this function is not sufficiently developed for a deficit to become evident before early adolescence. Importantly, this finding emphasizes ADHD as a developmental disorder, not just with regard to behavioral symptoms (Hart, Lahey, Loeber, Applegate, & Frick, 1995) but also when it comes to cognitive function.

Footnotes

Acknowledgements

We thank the project group of the Bergen Child Study for making the data available for the present study. We also thank Uppsala Child and Baby Lab, and particularly Professor Gunilla Bohlin, at the Department of Psychology, Uppsala University, for insightful discussions.

Authors’ Note

The study was approved by the Regional Committee for Medical Research Ethics in Western Norway and by the Ombudsman for Privacy in Research, Norwegian Social Science Data Services Ltd.

Declaration of Conflicting Interest

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Bergen Child study was financially supported by the Norwegian Research Council, The Norwegian Directorate for Health and Social Affairs, Western Norwegian Regional Health Authority, the L. Meltzer legacy for the University of Bergen, and the City of Bergen.

Author Biographies

Carin Tillman is an associate professor at the Department of Psychology, Uppsala University. She has conducted research on executive functions and ADHD.

Karin C. Brocki is an associate professor at the Department of Psychology, Uppsala University. She conducts research on executive functions in typical development and in relation to symptoms of ADHD.

Lin Sørensen is an associate professor at the Department of Biological and Medical Psychology, University of Bergen. Her main research interests are cognitive and emotional/motivational control in children with symptoms of ADHD.

Astri J. Lundervold is a professor at the Department of Biological and Medical Psychology, University of Bergen. Her research interest includes characterization of cognitive and emotional behavior associated with normal function and neurodevelopmental disorders across the life span.

References

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Barkley

R. A.

(1997). ADHD and the nature of self-control. New York, NY: Guilford.

Barkley

R. A.

Edwards

Laneri

Fletcher

Metevia

(2001). Executive functioning, temporal discounting, and sense of time in adolescents with attention deficit hyperactivity disorder (ADHD) and oppositional defiant disorder (ODD). Journal of Abnormal Child Psychology, 29, 541-556.

Berlin

Bohlin

(2002). Response inhibition, hyperactivity, and conduct problems among preschool children. Journal of Clinical Child & Adolescent Psychology, 31, 242-251.

Berlin

Bohlin

Rydell

A.-M.

(2003). Relations between inhibition, executive functioning, and ADHD symptoms: A longitudinal study from age 5 to 8(1/2) years. Child Neuropsychology, 9, 255-266.

Best

J. R.

Miller

P. H.

(2010). A developmental perspective on executive function. Child Development, 81, 1641-1660.

Biederman

(2005). Attention-deficit/hyperactivity disorder: A selective overview. Biological Psychiatry, 57, 1215-1220.

Bjorklund

D. F.

Harnishfeger

K. K.

(1990). The resources construct in cognitive development: Diverse sources of evidence and theory of inefficient inhibition. Developmental Review, 10, 48-71.

Brocki

K. C.

Bohlin

(2004). Executive functions in children aged 6 to 13: A dimensional and developmental study. Developmental Neuropsychology, 26, 571-593.

10.

Brocki

K. C.

Bohlin

(2006). Developmental change in the relation between executive functions and symptoms of ADHD and co-occurring behavior problems. Infant and Child Development, 15, 19-40.

11.

Brocki

K. C.

Eninger

Thorell

L. B.

Bohlin

(2010). Interrelations between executive function and symptoms of hyperactivity/impulsivity and inattention in preschoolers: A two year longitudinal study. Journal of Abnormal Child Psychology, 38, 163-171.

12.

Brocki

K. C.

Nyberg

Thorell

L. B.

Bohlin

(2007). Early concurrent and longitudinal symptoms of ADHD and ODD: Relations to different types of inhibitory control and working memory. Journal of Child Psychology and Psychiatry, 48, 1033-1041.

13.

Clark

Prior

Kinsella

G. J.

(2000). Do executive function deficits differentiate between adolescents with ADHD and oppositional defiant/conduct disorder? A neuropsychological study using the Six Elements Test and Hayling Sentences Completion Test. Journal of Abnormal Child Psychology, 28, 403-414.

14.

Conners

C. K.

(2000). The Conners’ Continuous Performance Test II: Technical guide. Toronto, Ontario, Canada: Multi-Health Systems.

15.

Conners

C. K.

Epstein

J. N.

Angold

Klaric

(2003). Continuous performance test performance in a normative epidemiological sample. Journal of Abnormal Child Psychology, 31, 555-562.

16.

Cragg

Chevalier

(2012). The processes underlying flexibility in childhood. Quarterly Journal of Experimental Psychology, 65, 209-232.

17.

Delis

D. C.

Kaplan

Kramer

J. H.

(2001). D-KEFS executive function system. Technical manual. San Antonio, TX: Psychological Corporation.

18.

Dempster

F. N.

(1992). The rise and fall of the inhibitory mechanism: Toward a unified theory of cognitive development and aging. Developmental Review, 12, 45-75.

19.

Déry

Toupin

Pauze

Mercier

Fortin

(1999). Neuropsychological characteristics of adolescents with conduct disorder: Association with attention-deficit-hyperactivity and aggression. Journal of Abnormal Child Psychology, 27, 225-236.

20.

Diamond

(2006). The early development of executive functions. In Bialystock

Craik

F. I. M.

(Eds.), Lifespan cognition: Mechanisms of change (pp. 70-95). Oxford, UK: Oxford University Press.

21.

Garon

Bryson

S. E.

Smith

I. M.

(2008). Executive function in preschoolers: A review using an integrative framework. Psychological Bulletin, 134, 13-60.

22.

Gathercole

S. E.

Pickering

S. J.

Ambridge

Wearing

(2004). The structure of working memory from 4 to 15 years of age. Developmental Psychology, 40, 177-190.

23.

Hart

E. L.

Lahey

B. B.

Loeber

Applegate

Frick

P. J.

(1995). Developmental change in attention-deficit hyperactivity disorder in boys: A four-year longitudinal study. Journal of Abnormal Child Psychology, 23, 729-749.

24.

Heiervang

Stormark

K. M.

Lundervold

A. J.

Heimann

Goodman

Posserud

Gillberg

(2007). Psychiatric disorders in Norwegian 8 to 10 year olds: An epidemiological survey of prevalence, risk factors, and service use. Journal of the American Academy of Child & Adolescent Psychiatry, 46, 438-447.

25.

Hobson

C. W.

Scott

Rubia

(2011). Investigation of cool and hot executive function in ODD/CD independently of ADHD. Journal of Child Psychology and Psychiatry, 52, 1035-1043.

26.

Homack

Lee

Riccio

C. A.

(2005). Test review: Delis-Kaplan executive function system. Journal of Clinical and Experimental Neuropsychology, 27, 599-609.

27.

Huizinga

Dolan

C. V.

van der Molen

M. W.

(2006). Age-related changes in executive function: Developmental trends and a latent variable analysis. Neuropsychologia, 44, 2017-2036.

28.

Hummer

T. A.

Kronenberger

W. G.

Wang

Dunn

D. W.

Mosier

K. M.

Kalnin

A. J.

Mathews

V. P.

(2011). Executive functioning characteristics associated with ADHD comorbidity in adolescents with disruptive behavior disorders. Journal of Abnormal Child Psychology, 39, 11-19.

29.

Kaufman

Birmaher

Brent

Rao

Flynn

Moreci

Ryan

(1997). Schedule for affective disorders and schizophrenia for school-age children-present and lifetime version (K-SADS-PL): Initial reliability and validity data. Journal of the American Academy of Child & Adolescent Psychiatry, 36, 980-988.

30.

Kline

R. B.

(2005). Principles and practice of structural equation modeling (2nd ed.). New York, NY: Guilford.

31.

Levin

H. S.

Culhane

K. A.

Hartmann

Evankovich

Mattson

A. J.

Harward

Fletcher

J. M.

(1991). Developmental changes in performance on tests of purported frontal lobe functioning. Developmental Neuropsychology, 7, 377-395.

32.

Llorente

A. M.

Williams

Satz

D’Elia

(2003). Children’s color trails test. Professional manual. Odessa, FL: Psychological Assessment Resources.

33.

Luciana

Conklin

H. M.

Hooper

C. J.

Yarger

R. S.

(2005). The development of nonverbal working memory and executive control processes in adolescence. Child Development, 76, 697-712.

34.

Luciana

Nelson

C. A.

(1998). The functional emergence of neurofrontally-guided working memory systems in four-to-eight-year-old children. Neuropsychologia, 36, 273-293.

35.

Lundervold

A. J.

Posserud

M. B.

Ullebø

A. K.

Sørensen

Gillberg

(2011). Teacher reports of hypoactivity symptoms reflect slow cognitive processing speed in primary school children. European Child and Adolescent Psychiatry, 20, 121-126.

36.

Lund-Johansen

Hugdahl

Wester

(1996). Cognitive function in patients with Parkinson’s disease undergoing stereotaxic thalamotomy. Journal of Neurology, Neurosurgery & Psychiatry, 60, 564-571.

37.

Miyake

Friedman

N. P.

Emerson

M. J.

Witzki

A. H.

Howerter

Wager

T. D.

(2000). The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology, 41, 49-100.

38.

Nigg

J. T.

(2006). What causes ADHD? Understanding what goes wrong and why. New York, NY: Guilford.

39.

Oosterlaan

Scheres

Sergeant

J. A.

(2005). Which executive functioning deficits are associated with AD/HD, ODD/CD and comorbid AD/HD+ODD/CD? Journal of Abnormal Child Psychology, 33, 69-85.

40.

Romine

C. B.

Reynolds

C. R.

(2005). A model of the development of frontal love functioning: Findings from a meta-analysis. Applied Neuropsychology, 12, 190-201.

41.

Schoemaker

Bunte

Wiebe

S. A.

Espy

K. A.

Dekovic

Matthys

(2012). Executive function deficits in preschool children with ADHD and DBD. Journal of Child Psychology and Psychiatry, 53, 111-119.

42.

Séguin

J. R.

Boulerice

Harden

P. W.

Tremblay

R. E.

Pihl

R. O.

(1999). Executive functions and physical aggression after controlling for attention deficit hyperactivity disorder, general memory, and IQ. Journal of Child Psychology and Psychiatry, 40, 1197-1208.

43.

Sørensen

Plessen

Lundervold

A. J.

(2012). The impact of inattention and emotional problems on cognitive function in primary school children. Journal of Attention Disorders, 16, 589-599.

44.

Stormark

K. M.

Heiervang

Heimann

Lundervold

A. J.

Gillberg

(2008). Predicting nonresponse bias from teacher ratings of mental health problems in primary school children. Journal of Abnormal Child Psychology, 36, 411-419.

45.

Stroop

J. R.

(1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643-661.

46.

Thorell

L. B.

Wåhlstedt

(2006). Executive functioning deficit in relation to symptoms of ADHD and/or ODD in preschool children. Infant and Child Development, 15, 503-518.

47.

Tillman

C. M.

Eninger

Forssman

Bohlin

(2011). The relation between working memory components and ADHD symptoms from a developmental perspective. Developmental Neuropsychology, 36, 181-198.

48.

Toupin

Déry

Pauze

Mercier

Fortin

(2000). Cognitive and familial contributions to conduct disorder in children. Journal of Child Psychology and Psychiatry, 41, 333-344.

49.

Van der Meere

Sergeant

. (1988). Controlled processing and vigilance in hyperactivity: Time will tell. Journal of Abnormal Child Psychology, 16, 641-655.

50.

Wåhlstedt

(2009). Neuropsychological deficits in relation to symptoms of ADHD: Independent contributions and interactions. Child Neuropsychology, 15, 262-279.

51.

Wåhlstedt

Thorell

L. B.

Bohlin

(2008). ADHD symptoms and executive function impairment: Early predictors of later behavioral problems. Developmental Neuropsychology, 33, 160-178.

52.

Waschbusch

D. A.

(2002). A meta-analytic examination of comorbid hyperactive-impulsive-attention problems and conduct problems. Psychological Bulletin, 128, 118-150.

53.

Wechsler

(1992). Wechsler Intelligence Scale for Children (3rd ed.). London, England: Psychological Corporation.

54.

Wechsler

(2003). Wechsler Intelligence Scale for Children. Technical and interpretive manual (4th ed.). London, England: Harcourt Assessment.

55.

Wechsler

(2007). Wechsler Intelligence Scale for Children (4th ed., Swedish version). London, England: Harcourt Assessment.

56.

Willcutt

E. G.

Doyle

A. E.

Nigg

J. T.

Faraone

S. V.

Pennington

B. F.

(2005). Validity of the executive function theory of attention-deficit/hyperactivity disorder: A meta-analytic review. Biological Psychiatry, 57, 1336-1346.