1. Introduction

During the course of a child’s development, children become more capable in controlling their actions and thoughts. This increase of control is thought to be due to the development of one’s cognitive executive function (EF). EF is important for goal-directed behaviour, particularly in demanding or new situations that require a flexible and quick adjustment in behaviour to adapt to the environment.

Research (i.e. lesion studies) has indicated that EF relies on the prefrontal cortex (PFC). These studies indicate that lesions in the PFC lead to poor performance on the Wisconsin Card Sorting Task (WCST) and the Tower of London task (ToL). Children completing these tasks show the same pattern as patients with PFC damage; they perseverate on the WCST and need more turns to solve the ToL task. The extended maturation process of the PFC seems to be the cause of the slow EF development. This study looks for conclusive evidence regarding developmental trajectories of various EF components with regard to performance on regular neuropsychological EF tasks.

1.1. The decomposition of executive function

The organization of the EF remains a major theoretical issue. On the one hand researchers think that the EF is unitary (thus does not consist of distinct sub-components or sub-functions), and therefore damage to the PFC leading to behavioural and cognitive impairments are due to one dysfunctional system. On the other hand, EF is viewed as being non-unitary (thus multi-faceted), containing numerous distinct cognitive processes with a relatively central neural representation. Support for this theory can be found in neuroimaging studies: different parts of the EF seem to depend on different parts of PFC (f.e. maintaining information in the working memory (WM) recruits mostly lateral PFC, switching between tasks mostly medial PFC, and inhibiting responses the orbitofrontal cortex).

Task impurity means that a single indicator of a given construct (thus f.e. the operationalization of WM), can seldom be seen as a pure measure of that specific construct. This is because they are usually contaminated by systematic and random error. In neuroimaging and behavioural studies that use various EF tasks, task impurity deters the interpretation of the results, as EF components often involve other cognitive processes (which can also be non-EF processes).

Miyake and colleagues (2000) found a way to address this problem, namely by using various tasks in order to measure the EF components and therewith assuming a latent variables method in order to extract the variance that is shared in those tasks. This method decreases task impurity and for that reason offers a lot more information in developmental studies.

The results of Miyake and colleagues study (2000) showed that Working Memory, Shifting, and Inhibition could be seen as separable constructs.

Also, EF component processes predicted performance differentially (Inhibition predicted ToL performance and Shifting predicted WCST performance).

1.2. The development of executive function

Although it has been found that the development of EF begins in early childhood and continues into adolescence, EF tasks differ in their developmental trajectories. Studies show that adult-level performance on various EF tasks are achieved at numerous different ages during the course of childhood and adolescence. For example, the capacity of the WM develops throughout childhood into adolescence, and studies on task shifting abilities indicate there is a decrease in the price of shifting between tasks as children age (whereby adult performance levels are achieved at approximately 12 years of age). Inhibitory control has also been found to increase during the course of childhood (reaching adult level performance at approximately 12 years of age).

Direct interpretation of developmental trends in EF component processes are hindered by the following factors:

1. The same EF component is measured using different tasks across various studies.

2. It remains uncertain whether the strategies children use when they carry out EF tasks differs between ages (do children of different ages use the same strategy and are we therefore measuring the same construct or not. If not, this can lead to measurement invariance).

3. There are several developmental studies that focus only on a single EF component process. This makes reliable assessment of possible developmental changes across EF component processes difficult (differing rates may be because of different samples and not because of different components).

Therefore, homogeneous age groups and application of a latent variables method are necessary for a reliable assessment of EF’s developmental pattern. This makes it possible to determine which EF component processes are shared. Earlier research has observed an alike factor structure in 8 – 13 year old children as Miyake and colleagues (2000) found earlier in adults. This study aims to contribute to these results and adopts a multi-group design. This will provide a more graded assessment regarding developmental change in EF component processes.

1.3. The present study

In this study the conceptual framework of Miyake and colleagues (2000) was adopted in order to determine developmental change in EF. The primary goal is to review age-related changes in Working Memory (WM), Shifting, as well as Inhibition.

Three homogeneous age groups were tested (7-,11-, and 15-year old children), as well as one group of young adults (21 years of age), whereby three experimental tasks for each EF component we used (thus 9 tasks in total).

Definitions:

• WM: all of the cognitive processes which briefly retain information that can easily be retrieved and used to carry out any mental task.

• Shifting: the ability to shift back and forth between multiple tasks

• Inhibition: the ability to intentionally constrain (or inhibit) automatic, dominant, or pre-potent reactions.

Two approaches in analysing the data were adopted.

1. A standard analysis of variance (ANOVA) approach

2. A latent variable approach (multi-group confirmatory factor analyses)

The following was examined:

1. The organization of EF in both children as well as young adults. This was done by investigating if the indicators from the WM, Shifting and Inhibition tasks measured identical constructs across age.

2. If EF organization changes during development.

3. The contribution of EF component processes across age groups on WCST and ToL task performance.

2. Method

2.1. Sample

• 71 seven-year-old children (39 female, mean age 7.2 (range 6-8))

• 108 eleven-year-old children (62 female, mean age 11.2 (range 10-12))

• 111 fifteen-year-old adolescents (58 female, mean age 15.3 (range 14-16))

• 94 twenty-one year old young-adults (72 female, mean age 20.8 (range 18-26))

The participants were recruited from local schools and universities. Children and university students with health problems, psychiatric problems, or neurological damage were excluded. Participants all had normal vision. Informed consent was obtained. Intelligence was assessed using a non-verbal IQ test, the Standard Progressive Matrices. The gender distribution was not equal, there were many more females in the young-adult group. This did not change main effects or interactions regarding the particular task manipulations, thus IQ and gender were not reported in the analyses.

2.2. Experimental tasks

The tasks used were designed to assess the EF components Working Memory, Shifting, and Inhibition. There were also three complex EF tasks that were administered: the WCST, the ToL, and the Random Number Generation task. All tasks were speeded choice reaction time (RT) tasks.

2.2.1. Working memory

• The Tic Tac Toe task (Xs and Os) requires participants to remember visual information concerning the orientation of a pattern of figures actively in their working memory. This task has two phases: a memorizing and a recognition phase.

• The Mental Counters task requires participants to remember numerical information actively in their working memory.

• The Running Memory: stimuli pictures of fruit and of animals are presented and participants indicate if a current pair matched the last presentation of that pair.

2.2.2. Shifting

• Local-Global task whereby participants respond to randomly presented squares or rectangles.

• Dots-Triangles task whereby varying numbers of either green triangles or red dots are presented on the screen in a 4x4 grid.

• Smiling Faces task: in this task the stimuli are schematic faces (man/woman, smiling/unsmiling) presented in a 2x2 grid.

2.2.3. Inhibition

• Stop-signal task whereby participants respond as quickly as possible to a green or red arrow pointing either to the left or to the right.

• Eriksen Flankers task whereby the participant has to respond to a right versus left pointing arrow in the centre of the screen, which is flanked by four congruent arrows (→→→→→ or ←←←←←) or by four incongruent arrows (→→←→→ or ←←→←←).

• Stroop task whereby stimuli are pictures of “smileys” with either a red or a blue contour (the colour task) and with either a right-side-up or upside-down orientation (the orientation task).

2.2.4. Complex EF tasks

• Wisconsin Card Sorting Task (WCST)

• Tower of London task (ToL).

2.3. Procedure

There were two testing sessions of approximately 1.5. hrs. Six tasks were administered per session. Task order was counterbalanced across participants, however, the WCST and ToL were administered as the last tests in the test battery. At the end of the session, the children completed the Raven SPM individually in the classroom, and the 15- and 21-year olds completed the test using a pencil and paper version of the test.

3. Results

Three sets of analyses were performed:

1. ANOVA’s to assess the developmental trajectories for each task.

2. Multi-group CFA’s in order to determine when latent components WM, Shifting, and Inhibition reached the adult level.

3. Regression analyses in order to assess the latent factor contribution on the WSCT and ToL performance.

4. Discussion

This study examined developmental trajectories of EF components WM, Shifting, and Inhibition with regard to performance on EF tasks (WCST and ToL). Both standard analyses and multi-group latent variable modeling were adopted, thus allowing for the modelling of underlying (latent) variables across age groups.

4.1. Analysis of manifest task performance

In accordance with several earlier developmental studies, it was found that developmental trajectories of different EF component processes vary. Earlier studies have found that developmental changes in the WM (behavioural and brain functioning) co-occur (f.e. that development of WM co-occur with the maturation of the lateral PFC). This study also found on two tasks (the Tic-Tac-Toe-task and the Mental Counters task) that an adult level of performance was not achieved prior to 12 years of age. Accuracy of performance on the Running Memory WM task increased with age, but there was no task load effect found. The authors do not have an explanation for this.

All three task shifting tasks indicated that the shift costs decrease until the age of 15. Therefore, it seems that the capability to shift tasks sets reaches a young-adult level of performance in adolescence.

Consistent with previous studies, the two tests used to test the inhibition of motor responses showed distinct developmental trends. Performance of the Stop-signal and Eriksen Flankers task improved at a quick rate until 11 years of age. From the age onwards, performance was consistent with that of the 15- and 21- year olds. The Stroop task only showed weak developmental trends during the period of early childhood into adulthood.

The WCST and the ToL showed that the performance on these tasks were in line with earlier found developmental trends. The proportion of perseverative errors and conceptual level on the WCST reached the young-adulthood level in 11-year-old children. However, the number of categories completed in this task increased into young adulthood. The ToL task indicated that number of extra moves and planning time reaches the young-adult level in 11-year-old children, although the proportion of perfect solutions keeps increasing with age into young adulthood.

Overall, these analyses of performance indicate that WM, Shifting and Inhibition reach adult levels between the ages of 11 and 15, although some areas of inhibitory control (i.e. the Stroop task) does not reach adult-level until after 15 years of age. With regard to the WCST and ToL, adult performance levels are reached between the ages of 11 and 21.

4.2. Study Limitations

• Missing data: was retained as missing and was not imputed (this is known to lead to loss in statistical power)

• The WM factor was defined only by two indictors (Tic-Tac-Toe-task and the Mental Counters task), however it is desirable to have at least three indicators.

4.3. Conclusion

This study provides support for EF as being non-unitary (thus multi-faceted), which means that EF contains numerous distinct cognitive processes. Differences in developmental trends in WM and Shifting were found: WM continues to develop into young-adulthood and Shifting already reaches developed levels during adolescence. Ability to inhibit response on the Stop-signal and Eriksen Flanker task improves rapidly until 11-years of age. On the Stroop task, response speed only revealed little developmental change, however, accuracy of response did increase rapidly during the course of childhood and into young-adulthood.

Adult-level of performance was achieved between the ages of 11 and 15 on the WCST and ToL. There is, however, an increase in the proportion of perfect solutions on the ToL into young-adulthood. These findings are in concurrence with earlier findings that have demonstrated developmental improvements in EF component processes, as well as with the recent findings emerging from neuroimaging studies regarding cognitive development. Neurophysiological studies that show that anatomical development of PFC areas only reach maturity in young-adulthood are in support of this notion.