BJMB
Brazilian Journal of Motor Behavior
Research Article
!
Beavan et al
2019
VOL.13
N.2
64 of 75
Age-Related Differences in Executive Functions Within High-Level Youth Soccer Players
ADAM BEAVAN
1,2
| JAN SPIELMANN
3
| JAN MAYER
3
| SABRINA SKORSKI
1
| TIM MEYER
1
| JOB FRANSEN
4
|
1
Institute of Sports and Preventive Medicine, Saarland University, Saarbrücken, GERMANY.
2
Think Tank, DFB-Akademie (Deutscher Fußball-Bund), Frankfurt, GERMANY.
3
TSG 1899 Hoffenheim, Zuzenhausen, GERMANY.
4
Sport and Exercise Discipline Group, Faculty of Health, University of Technology, Sydney, AUSTRALIA.
Correspondence to: Adam Beavan. Institute of Sports and Preventive Medicine, Saarland University, Campus, Building B 8-2, 66123, Saarbrücken.
email: adam.beavan@uni-saarland.de
https://doi.org/10.20338/bjmb.v13i2.131
HIGHLIGHTS
• Positive relationship between Executive
Functions (EF), age and playing experience.
• Magnitude of change in EF is larger between
the younger cohorts.
• Older athletes can better negate unimportant
information from incongruent precues.
• EF sum score calculating all tests together
can differentiate between age-groups.
ABBREVIATIONS
ANOVA Analysis of variance
EF Executive Functions
Exp Experience playing soccer
MANOVA Multivariate Analyses of Variance
PCRTT Precued response time task
RSTT Reactive stress tolerance task
SSRT Stop signal reaction time
PUBLICATION DATA
Received 08 Apr 2019
Accepted 14 Jul 2019
Published 01 Aug 2019
BACKGROUND: It is less-common for athletes to be assessed on their ability to detect and process implicit
sources of information.
AIM: This study aimed to investigate age-group differences in executive functions within youth soccer players,
with the inclusion of a new implicit precued choice response time task.
METHOD: Seventy-four male soccer players: U12 (n=15), U13 (n=17), U17 (n=21) and U19 (n=21) representing
a representing a youth academy of an elite German Bundesliga club participated in this study. Players
conducted a battery of computer-based cognitive function tests: a precued choice response time task (PCRTT),
a stop signal reaction time task (SSRT), a multiple-object-tracking task (Helix), and a reactive stress tolerance
task (RSTT).
RESULTS: The MANOVAs revealed a multivariate effect of age group on the RSTT (p<0.001, ES=0.38) and the
SSRT (p<0.001, ES=0.20). A one-way ANOVA revealed an age group effect for response accuracy in the Helix
(p=0.01, ES=0.14). Lastly, a within-subjects effect of congruency on the PCRTT (p<0.001, ES=0.41) and a
between-subjects effect of age group (p=0.008, ES=0.15) was observed.
CONCLUSION: The results provided support for including an implicit precueing task, while the overall testing
demonstrated that the magnitude of the increase in executive functions between ages was greater across the
younger age groups compared to the older age groups.
KEYWORDS: Football | Cognitive | Inhibitory Control | Implicit Precue
INTRODUCTION
In a sporting context, executive functions (EF) are a sub category within the
theoretical frame work of the cognitive component approach, and are often described as
‘game intelligence’
1
. Vestberg, Gustafson, Maurex, Ingvar, Petrovic
2
first noted that the
existing body of research lacked understanding of the importance of general cognitive
abilities within an athletic population. The authors proceeded to test high and low division
adult soccer players on a series of non-sport specific cognitive function tests. The results
revealed that soccer players outperformed the norm group for both men and women, and
high division players outperformed the low division players. Since Vestberg and colleagues’
paper on EF in sport, interest in measuring EF has grown.
One EF that talented soccer players consistently outperformed their lower-level
counterparts on is response inhibition (i.e. the suppression of an ongoing motor response)
2,3
, among others. Thus, enhanced response inhibition may be a contributor to successful
sporting performance in more talented players across all age groups, and therefore
BJMB! ! ! ! ! ! ! ! !!!!!!!!Research Article!
Brazilian(Journal(of(Motor(Behavior!
Beavan et al
2019
VOL.13
N.2
65 of 75
advocates for more research to investigate this EF. The importance of response inhibition
in sport may be attributed to the role that it plays in the decision-making process
4
. The
ability to inhibit a response results in players making fewer errors by being able to
suppress acting on a decision; which is typical in soccer when a defender suddenly guards
the intended receiver of a pass, and a new decision must immediately be created.
Response inhibition in the EF research has commonly been assessed using simple or two-
choice motor response tasks
5
. However, a simple motor response may not be
representative of the stimulus-response a team-sport athlete encounters in-situ.
Accordingly, a multiple-choice motor response task test may better reflect performance, as
players must decide rapidly which decisions to act upon and which decisions to suppress
while presented with a variety of options
6
. Moreover, not only is the task complexity
simplified, the current response inhibition tests such as the stop-signal reaction test are
explicit in nature. It may be speculated that the vast majority of stimuli which athletes are
exposed to are hidden within the sporting environment (i.e. implicit rather than explicit), as
it is impossible to consciously attend to every stimulus. Many stimuli go unnoticed during a
game that may non-consciously change and/or challenge the athlete’s sporting
performance
7
. Therefore, the development of a new EF test that measures the impact that
implicitly perceived visual cues on response time has value.
Understanding the influence that non-attentively perceived cues have on motor
performance requires the contribution of the paradigm in cognitive science known as
‘precueing’. Precueing is the effect that a presented stimulus has on participants’
subsequent decision-making or motor behaviour, albeit an explicit or implicit stimulus
8
. A
precue can influence a decision at a non-conscious level, leaving the participant with no
subjective experience of having their decisions altered or to some extent, delayed
7
. For
instance, in an attempt to prepare the player in possession of the ball for the movement
that will occur next, a teammate may point towards their intended direction prior to the
initiation of a run. However, whether the player in possession of the ball consciously or
non-consciously registers the teammate’s hand gesture prior to the run is not always
certain.
The first studies on the effects of advanced visual information have demonstrated
that if this information provides accurate information about the subsequent stimulus
(congruent), it improved reaction times in comparison to non-cued trials
8
. Opposingly,
response times were impaired if the precue and stimulus contradicted each other
(incongruent)
9
. Although precueing has been extensively researched in mainstream
psychology; the transition of research into a sporting domain may improve the
understanding of response inhibition in athletes
10
.
Despite the advances of knowledge of EF in athletes, there is another noteworthy
limitation. Previous methodologies have used a relatively high variation of participants’ age
distribution within each group. For example, Vestberg, Reinebo, Maurex, Ingvar, Petrovic
11
grouped players age ranging from 12-19 years together, and it has not yet been
investigated whether more specific age-group (i.e. stratified by distinctive birth years)
differences are revealed in a homogenous population of high-level athletes. From research
sourced from a cognitive science domain, EF are still developing rapidly during the
adolescent phase
12
. In course of normal aging, early adolescents experience an increased
effectiveness to engage in deliberate, goal-orientated thought and action, and these
changes are have been reported to be significantly improved between children (mean age
= 8 years old) and young adults (mean age = 22.3)
13
, yet more specific age groups are not
BJMB! ! ! ! ! ! ! ! !!!!!!!!Research Article!
Brazilian(Journal(of(Motor(Behavior!
Beavan et al
2019
VOL.13
N.2
66 of 75
provided. Furthermore, the enhanced ability to differentiate between goal appropriate
responses and goal inappropriate responses that must be supressed also continues to
improve throughout the adolescent phase
14
, reflected by reduced reaction times on
measures of response inhibition. Accordingly, these studies provide support towards not
grouping players with differently developed EF coupled with various levels of domain-
specific experience. Contrastingly, identifying specific age group reference values may
provide more of a justification of which age groups share similar or distinctive EF to be
combined in future studies if required.
Therefore, the aims of this study were threefold. First, to investigate age-group
differences on EF tests in a homogenous population of talented youth soccer players. It is
hypothesised that performance on EF tests will be greater in the older groups, as more
domain specific experience is expected to transfer into better EF performance. The second
aim was to examine the influence of an implicit precue on response times in a precued
response time task (PCRTT) as measuring implicit response processes compared to
explicit measures may be more appropriate to sports where fast and accurate responses
are required. It is further hypothesised that the increase in domain specific experience will
also transfer into older players to act on correct information whilst also negate unimportant
information, demonstrated by faster reaction times on the PCRTT. The third aim was to
develop an overall EF sum score, allowing practitioners to more easily interpret and
convey the results of tests to coaches and players alike.
MATERIAL AND METHODS
Participants
Seventy-four youth male soccer players (means ± SD; Age; years of experience
playing soccer = Exp) from four age groups: U12 (n = 15; Age = 10.3 ± 0.6; Exp = 6.4 ±
1.7), U13 (n = 17; Age = 11.2 ± 0.5; Exp = 7.6 ± 1.7), U17 (n = 21; Age = 15.2 ± 0.3; Exp:
11.6 ± 2.5) and U19 (n = 21; Age = 16.7 ± 0.5; Exp = 12.9 ± 2.2) representing a youth
academy of an elite German Bundesliga club participated in this study. Prior to
commencement of this study, informed consent for all players was received, and the
Institutional Ethics Committee approved this study.
Procedures and apparatus
Players conducted a battery of cognitive function tests. Each group was assessed
on a separate day in the same week during pre-season. Each player was assigned to a
cognitive assessment and rotated to the next free assessment. One staff member
remained at each assessment station to give standardized instructions and monitor each
player’s performance. Each assessment had a standardized familiarisation protocol prior to
commencing the experimental trials.
Vienna Test System: Determination Test
The determination test (Schufried GmbH, Austria) is a complex multi-stimuli
reaction test involving the combination of five different coloured stimuli and two acoustic
signals (2000 Hz high and 100 Hz low tone) for finger pressing, and two pedal stimuli for
the feet. These stimuli corresponded to the pressing of appropriate buttons on the
response panel and foot pedals. The determination test aims to measure reactive stress
BJMB! ! ! ! ! ! ! ! !!!!!!!!Research Article!
Brazilian(Journal(of(Motor(Behavior!
Beavan et al
2019
VOL.13
N.2
67 of 75
tolerance and the associated reaction speed. The participant must remain composed
whilst the quick succession of the single pairing of stimuli and response lasting four
minutes. ‘Correct responses’ describes the total number of accurate responses within the
four minutes, and ‘response time’ is the median response time (s) from the appearance of
a stimulus to pressing of the correct button.
Vienna Test System: Response Inhibition Test
The response inhibition test (Schufried GmbH, Austria) uses a stop signal
paradigm. In each trial, the player is presented with an arrow either pointing left or right, to
which he must respond by pressing the corresponding button. Each arrow is displayed for
one second, and the time before the subsequent arrow appears is also one second.
Seventy-six stimuli are ‘go trials’, with the other 24 stimuli having a tone at a pitch of
1000Hz for 100 ms (stop signal). The player must then supress the already initiated
response, known as ‘stop trials’. The time between the presentation of the stimulus and the
tone is dependent on the player’s performance, being that if the player responds correctly
to a stop signal trial, the interval for the next stop stimuli will occur 50 ms later, and vice
versa. Therefore, the correct response to the stimuli will continually progress in difficulty
(minimum 50 ms; maximum 350 ms). The dependent variable that reflects the latency of
the inhibitory process is stop signal reaction time (SSRT). The SSRT is calculated by
deducting the mean stop signal delay from the mean reaction time (s).
Helix
The Helix (SAP, Walldorf, Germany) is a multiple object tracking assessment in
which participants are asked to track multiple players at once. The player stands facing a
180º curved screen (7 m width x 2.16 m height) and must track four out of eight players.
Simulated players run around a soccer field for eight seconds in a randomized fashion and
return to back to the start line up. Players must then choose the four players they had to
track. Players had four practice trials, and ten marked trials. The maximum score is 40.
Precued Choice Response Time Task
Participants were required to press the button on a joystick panel associated with
a stimulus circle presented on the laptop screen as fast and accurate as possible. The
PCRTT developed using Unity software (Unity, Version 5.4.0f3, 2016). Four blank stimulus
circles were presented in a horizontal line, with one circle turning yellow in colour after a
randomised (2-4 second) fore-period length. Each circle each had a diameter of 512 pixels
and an edge width of 5 pixels on a 13.2-inch display. Prior to the appearance of the
stimulus, a three second countdown timer was shown. After the appearance of the four
stimulus circles, a small dot was appeared for 43 ms in the centre of one stimulus circle,
86 ms prior to the circle turned yellow. The duration of the precue was based on prior
research supporting that precue duration below the 100 ms threshold are suitable to be
used as unconscious precues
15
, and a 43 ms precue has been identified as an
appropriate precue length in research involving cognitive responses
16
.
Twenty-four trials were conducted. Twelve trials had the small dot appear in the
same circle as the yellow dot (congruent) and the other twelve trials had the dot appear in
a different circle as the yellow dot (incongruent). Response time (given in ms) was
measured as the duration between the appearance of the stimulus circle (turned yellow) on
BJMB! ! ! ! ! ! ! ! !!!!!!!!Research Article!
Brazilian(Journal(of(Motor(Behavior!
Beavan et al
2019
VOL.13
N.2
68 of 75
the computer screen and the moment the button was pressed by the participant. A visual
depiction of the task used can be found in Figure 1.
Figure 1. Depiction of the Precued Choice Response Time Task.
Statistical Analysis
Following data collection, participant responses were initially analysed according
to their accuracy. Responses that did not correspond with the stimulus circle (i.e. when a
false response was given) were considered incorrect and the response time of the
respective trial was discarded due to the low frequency of incorrect responses (n = 53).
Furthermore, to highlight instances in which the participants missed the controller button or
did not press it sufficiently, an outlier labelling rule was used following the methods
outlined by Hoaglin, Iglewicz, Tukey
17
, and applied on an individual basis to limit within
subject variance. Furthermore, the interquartile range was multiplied by 1.5, and trials with
response times beyond the 25
th
and 75
th
percentiles ± the inter-quartile range were
considered outliers and therefore discarded (n = 108). The remaining raw responses (n =
1615) from this test were grouped according to ‘condition’ (i.e. congruent or incongruent
trials), and the mean of the correct responses from each participant in each condition was
computed.
Normalized values were calculated from z-scores for all items as per: Normalized
score = 100+(Z-score*15). When larger numbers represented poorer scores, the z-scores
were inversed before normalization, so a higher value was associated with a better score.
These normalized values were then used in two factor analyses to develop a total
executive function sum score (EF sum score) for all players. An exploratory factor analysis
used principal component analysis with a varimax rotation to determine the number of
factors revealed within all EF assessments to assess the feasibility of one overarching EF
factor, a second confirmatory factor analysis then investigated item loadings when all items
BJMB! ! ! ! ! ! ! ! !!!!!!!!Research Article!
Brazilian(Journal(of(Motor(Behavior!
Beavan et al
2019
VOL.13
N.2
69 of 75
were forced to load onto a single factor. Items were discarded when they were deemed to
be ‘unimportant’, i. e. when their communality was found to be lower than 0.40. From the
final factor analysis, a new EF variable was developed using each individual item’s factor
loading as a weighting system.
Finally, (i) one Repeated Measures Analysis of Variance, (ii) two Multivariate
Analyses of Variance (MANOVA) and (iii) two one-way Analysis of Variance (ANOVA)
were used to investigate age-group differences in: (i) PCRTT response time where
congruent-incongruent scenarios were included as a within-subjects variable, (ii)
Determination Test performance with response time and correct responses entered as
dependent variables and Vienna Test performance with start-stop response time and
response time as dependent variables, and (iii) Helix performance score and the newly
developed EF sum score. Bonferroni corrections were used to investigate multiple
comparisons between age groups and partial eta squared effect sizes were used
throughout to investigate the magnitude of any observed effects using Cohen
18
guidelines
for interpreting effect sizes: 0.01-0.06 = small effect, 0.06-0.14 = moderate effect and
>0.14 = large effect. In all analyses, partial eta squared effect sizes were calculated and
the significance level was set at p <0.05. All statistical analyses were performed using IBM
SPSS Statistics for Windows, Version 24.
RESULTS
The descriptive statistics and results of (M)ANOVAs for the EF tests can be found
in Table 1.
Table 1 – Means ± standard deviations (95% confidence intervals), and results of (M)ANOVAs for the executive function tests.
Test
Variable
U12 (n = 15)
U13 (n = 17)
U17 (n = 21)
U19 (n = 21)
PCRTT
Congruent RT (ms)
0.590±0.059
(0.562-0.619)
0.578±0.065
(0.554-0.603)
0.547±0.047
(0.523-0.571)
0.537±0.053
(0.512-0.561)
Incongruent RT (ms)
0.612±0.056
(0.582-0.641)
0.607±0.069
(0.582-0.632)
0.569±0.052
(0.544-0.594)
0.558±0.049
(0.533-0.582)
Determination
Test
Correct Answers (n)
215.07±25.04
(201.20-228.93)
238.78±32.477
(222.63-254.93)
263.90±29.64
(250.41-277.40)
291.14±41.220
(272.38-309.91)
RT (ms)
0.835±0.066
(0.798-0.871)
0.758±0.096
(0.710-0.805)
0.651±0.055
0.626-0.677)
0.619±0.065
(0.589-0.648)
Response
Inhibition Test