Does Executive Function Capacity Moderate the Outcome of Executive Function Training in Children with ADHD?

Abstract

Objective: Executive functioning (EF) training-interventions aimed at ADHD-symptom reduction have limited results. However, EF-training might only be effective for children with relatively poor EF capacity. This randomized double-blind placebo-controlled study examined if pre-training EF capacity moderates the outcome of an EF-training intervention on measures of near- (EF performance) and far transfer (ADHD symptoms and parent-rated EF behavior) immediately after treatment and at three months follow-up.
Methods: Sixty-one children with ADHD (aged 8-12) were randomized to either an EF-training condition where working memory (WM), inhibition and cognitive-flexibility (CF) were trained, or to a placebo condition. Single moderation models were used.
Results: All significant moderation outcomes had small effect-sizes. After Bonferroni correction there were no significant moderators of treatment outcome.
Conclusions: Children with poor EF capacity do not benefit more from EF training than from placebo training. Training only EF-impaired children will probably not improve outcomes of EF training-studies.

Full text

Vol.:(0123456789)
1 3
ADHD Attention Deficit and Hyperactivity Disorders
https://doi.org/10.1007/s12402-019-00308-5
ORIGINAL ARTICLE
Does executive function capacity moderate theoutcome ofexecutive
function training inchildren withADHD?
SebastiaanDovis1,2 · MarijaMaric1,2 · PierJ.M.Prins1,2 · SaskiaVanderOord1,2,3
Received: 1 October 2018 / Accepted: 15 May 2019
© Springer-Verlag GmbH Austria, part of Springer Nature 2019
Abstract
Executive functioning (EF) training interventions aimed at ADHD-symptom reduction have limited results. However, EF
training might only be effective for children with relatively poor EF capacity. This randomized double-blind placebo-con-
trolled study examined whether pre-training EF capacity moderates the outcome of an EF-training intervention on measures
of near transfer (EF performance) and far transfer (ADHD symptoms and parent-rated EF behavior) immediately after
treatment and at 3-month follow-up. Sixty-one children with ADHD (aged 8–12) were randomized either to an EF-training
condition where working memory, inhibition and cognitive flexibility were trained, or to a placebo condition. Single mod-
eration models were used. All significant moderation outcomes had small effect sizes. After Bonferroni correction, there
were no significant moderators of treatment outcome. Children with poor EF capacity do not benefit more from EF training
than from placebo training. Training only EF-impaired children will probably not improve outcomes of EF training studies.
Keywords ADHD· Cognitive training· Moderation· EF training· Children
Introduction
Theories of ADHD suggest that deficits in executive func-
tioning are at the core of the ADHD syndrome and play a
pivotal role in explaining the problems children with ADHD
encounter in daily life (e.g., Barkley 2006; Nigg 2006; Rap-
port etal. 2001). Via dorsal frontostriatal brain circuits,
executive functions (EFs) allow individuals to regulate
their behavior, thoughts and emotions and, thereby, enable
self-control (Durston etal. 2011). Evidence indeed suggests
that impairments in EF are related to deficits in attention,
hyperactivity and impulsivity (e.g., Crosbie etal. 2013;
Sarver etal. 2015; Tillman etal. 2011), and with associated
problems such as deficient academic and social functioning
(Titz and Karbach 2014; Kofler etal. 2018a, c). Moreover,
research suggests that EF capacity and its associated levels
of brain activity are not static, but may be altered by task
repetition or training (Klingberg 2010). Of the different EFs
especially working memory, and to lesser extent inhibition
and set-shifting are impaired in individuals with ADHD
(Martinussen etal. 2005; Willcutt etal. 2005) Therefore,
in the past decade, EF training interventions with often as
central aim training ofespecially working memory have
received considerable interest.
However, recent meta-analyses (Cortese etal. 2014;
Dovis etal. 2015a; Hodgson etal. 2014; Rapport etal. 2013;
Sonuga-Barke etal. 2013; also see Chacko etal. 2013) sug-
gest that these EF training interventions in children with
ADHD mainly improve performance on measures of near
transfer (measures similar to the trained tasks in terms of
format and processing requirements), but have very limited
effects on measures of far transfer (i.e., measures that assess
different constructs or domains, such as ADHD symptoms
or parent-rated EF behavior in everyday life): In most pla-
cebo-controlled EF training studies transfer to measures of
untrained EF has been limited at best, and effects on parent-
or teacher-rated behavior (e.g., ADHD or EF) are generally
not found (Dovis etal. 2015b).
* Saskia Vander Oord
saskia.vanderoord@kuleuven.be
1 Developmental Psychology, University ofAmsterdam,
Nieuwe Achtergracht 129B, 1001NKAmsterdam,
TheNetherlands
2 Cognitive Science Center Amsterdam, University
ofAmsterdam, Nieuwe Achtergracht 129B,
1001NKAmsterdam, TheNetherlands
3 Clinical Psychology, KU Leuven, Tiensestraat 102, bus 3720,
3000Leuven, Belgium

S.Dovis et al.
1 3
Nonetheless, when clinicians, parents or teachers have
questions to whether a particular child with ADHD could
benefit from EF training, it is difficult to provide them with
a well-founded answer. This is mainly because current pla-
cebo-controlled EF training studies only focus on overall
treatment efficacy (i.e., “did my intervention work or not?”;
Maric etal. 2015), whereas variables that could influence
the relationship between treatment and outcome, including
“for whom” a certain treatment achieves its effects, remain
largely unstudied. These so-called treatment moderators are
“pretreatment or baseline variables that identify subgroups
of patients within the population who have different effect
sizes” (Kraemer etal. 2006, p. 1286). A treatment mod-
erator that is of particular interest for EF training studies is
children’s pre-training EF capacity. Evidence indicates that
ADHD is a heterogeneous disorder, with not all children
with ADHD having deficits in EF (e.g., Dovis etal. 2015c;
Fair etal. 2012; Nigg etal. 2005). It is suggested that espe-
cially EF-impaired children will benefit from EF training, as
they have more room for improvement (Diamond and Lee
2011; Diamond 2012), whereas in EF-unimpaired children
with ADHD, EF training will probably have less impact on
ADHD symptoms, as their symptoms are less likely to origi-
nate from impairments in EF.
To date, many placebo-controlled EF training studies
have been conducted. However, to our knowledge, none of
these studies in ADHD samples have investigated whether
the relation between EF training and improvements in
ADHD symptoms or parent-rated EF behavior is moder-
ated by children’ pre-training EF capacity (Van der Oord
and Daley 2015; for two non-placebo-controlled studies see
Hunt etal. 2014; Van der Donk etal. 2016). Identifying
such treatment moderators using decent placebo-controlled
comparisons may well be key to individualized and more
effective non-pharmacological treatments for children with
ADHD.
The goal of the present study is to determine whether pre-
training EF capacity is a moderator of near (EF performance)
and far transfer effects (ADHD symptoms and parent-rated
EF behavior) of a gamified, 5-week, home-based, EF train-
ing intervention titled Braingame Brian (BGB; Dovis etal.
2015b; Prins etal. 2013; Van der Oord etal. 2014). BGB
targets multiple EFs that are commonly impaired in children
with ADHD: visuospatial working memory (WM), response
inhibition, and cognitive flexibility (e.g., see Willcutt etal.
2012). Training multiple EFs has been suggested to be a
potentially more effective strategy to improve EF-related
ADHD behavior than single EF training (e.g., Cortese etal.
2014; Van Dongen-Boomsma etal. 2014). This is not only
because multiple EFs are involved in daily functioning (e.g.,
Isquith etal. 2013), but also because evidence suggests that
most children with ADHD show deficits in multiple EFs
(Fair etal. 2012), and that these EFs are largely related to
different brain regions (i.e., training one EF, will not auto-
matically result in improvement of another; e.g., McNab
etal. 2008; Schecklmann etal. 2013; Smith etal. 2006; for
a discussion of the unity and diversity of EFs see Miyake
and Friedman 2012).
To answer the current research questions we re-analyzed
part of the dataset from a recently published double-blind,
placebo-controlled study of BGB (see Dovis etal. 2015b).
In that study participants were randomized to one of three
conditions (i.e., versions of BGB): (1) a full-active condition
where visuospatial WM, inhibition and cognitive flexibility
were trained, (2) a partially active condition where inhibi-
tion and cognitive flexibility were trained and the WM train-
ing was presented in placebo mode, or (3) to a full placebo
condition. Overall short-term (1–2weeks) and long-term
(3months) treatment efficacy was evaluated. Regarding
near transfer, this study showed that visuospatial short-term
memory (STM) and WM only improved in the full-active
condition, inhibition only improved in the full-active and
partially active condition, and cognitive flexibility was not
improved in any condition. Regarding far transfer, both par-
ent- and teacher-rated ADHD symptoms and parent-rated
EF behaviors in everyday life improved in all conditions, but
no treatment x time interactions were found. These findings
are similar to those of other placebo-controlled EF training
studies in children with ADHD (Chacko etal. 2014a; Green
etal. 2012; Klingberg etal. 2002; Klingberg etal. 2005;
Kray etal. 2012). It was concluded that mainly nonspecific
treatment factors—as opposed to the specific effects of train-
ing EFs—seem related to far transfer effects (Dovis etal.
2015b). However, this and other placebo-controlled studies
did not account for potential moderators (i.e., pre-training
EF capacity) influencing treatment outcomes. These will be
investigated in the current study.
In the current study, to limit the number of analyses and to
assess moderation for the potentially most optimal condition
(the full-active condition), no specific hypothesis regarding
moderation for the partially active condition was formulated.
Therefore, we only compared the full-active condition to
the placebo condition. For clarity, from here onwards the
full-active condition will be referred to as the EF training
condition. Moderators of the short-term (1–2weeks post-
training) and long-term (3-month post-training) effects of
the EF training were evaluated using moderation analyses
(a conceptual moderation model illustrating proposed mod-
eration relations is presented in Fig.1).We expected that
pre-training EF performance would moderate change in out-
come measures of near transfer and far transfer (i.e., children
with poor pre-training EF would benefit more from EF train-
ing than from placebo, Diamond and Lee 2011; Diamond
2012). However, as largest differences between children
with ADHD and typically developing children are gener-
ally found in the EF working memory, this EF is the most

Does executive function capacity moderate theoutcome ofexecutive function training inchildren…
1 3
likely candidate for a being a significant moderator of EF
training effects.
Methods
This double-blind, placebo-controlled study is part of a large
study investigating the efficacy of BGB (Dovis etal. 2015b),
parts of it have been used in Sebastian’s Dovis Phd Thesis
(Dovis 2014). Not all measures that were used in that pre-
vious study are included in the current study as they are
not relevant for the current research questions. For details
regarding these measures, the original trial design, etc., see
the trial register: http://www.trial regis ter.nl/trial reg/admin /
rctvi ew.asp?TC=2728) and Dovis etal. (2015b).
Participants
Study settings
Fourteen outpatient mental-healthcare centers within pre-
dominantly urban type of communities in the Netherlands
were used for recruiting of children.
Eligibility criteria
Participants were all children in the age range between 8
and 12years with (a) a prior DSM-IV-TR (American Psy-
chiatric Association 2000) diagnosis of ADHD combined-
type 9 (b) absence of any autism spectrum disorder accord-
ing to a child psychologist or psychiatrist, (c) a score
within the clinical range (95th to 100th percentile) on the
Disruptive Behavior Disorder Rating Scale (DBDRS; Pel-
ham etal. 1992; Dutch translation: Oosterlaan etal. 2000),
more specifically the ADHD scales of both the parent- and
teacher-rated versions, (d) a confirmed diagnosis of ADHD
combined-type on the ADHD section of the Diagnostic
Interview Schedule for Children, parent version (PDISC-
IV; Shaffer etal. 2000). The structured diagnostic inter-
view PDISC-IV is based on the DSM-IV and has adequate
psychometric properties, (d) absence of conduct disorder
(CD) based on the CD sections of the structured diagnostic
interview the PDISC-IV, (e) an IQ score ≥ 80, which was
determined by a short version of the Dutch Wechsler Intel-
ligence Scale for Children (WISC-III; Kort etal. 2002).
This short version consisted of two subtests, Vocabulary
and Block Design, that were used to estimate full-scale IQ
Fig. 1 Conceptual moderation model. Note: BRIEF behavior rat-
ing inventory of executive function questionnaire, CBTT corsi block
tapping task, DBDRS disruptive behavior disorder rating scale, Far
transfer measures that assess constructs or domains different from
the trained tasks, Near transfer measures similar to the trained tasks
in terms of format and processing requirements, RCI reliable change
index (pre to post and pre to follow-up RCIs were used), STM short
term memory, TMT trail making task, WM working memory

S.Dovis et al.
1 3
(FSIQ). This composite score has satisfactory reliability.
Moreover, it correlates highly with FSIQ (Sattler 2001),
(f) absence of any neurological disorder, sensory (color
blindness, vision) or motor impairment as reported by the
parents, (g) not taking any medication except for meth-
ylphenidate or dextroamphetamine. When children were
taking medication, children discontinued their regular
methylphenidate dose at least 24h before each test ses-
sion, allowing for a complete washout (Greenhill 1998).
Children taking dextroamphetamine discontinued their
medication 48h before each test session (Wong and Ste-
vens 2012); finally, (h) parents were requested to keep the
dose of their medication for ADHD unchanged between
the date of the intake and the 3-month follow-up session,
and parents consented to not initiate or participate in other
psychosocial treatments during the course of the study. For
treatment group comparisons of baseline demographics
and clinical characteristics, see Table1.
Treatment conditions
General characteristics oftheintervention
“Braingame Brian” (BGB; Dovis etal. 2015b) is a home-
based, computerized EF training, which is embedded in a
game world. The main character of the game is “Brian”.
Throughout the game, Brian, a young inventor, helps and
befriends inhabitants of the game world. He does this by
creating elaborate inventions (e.g., a delivery-rocket for the
grocery-store owner); throughout the game they become
more elaborate. The game has 25 training sessions. Within
each training session, the player can create inventions by
completion of the tasks in the training session: each train-
ing session consists of a WM task, a cognitive flexibility
task, and an inhibition task. The duration of every ses-
sion is about 35–50min (30min for task completion and
an optional amount of time for exploring the elaborate
game world). For all participants, an identical additional
Table 1 Baseline demographics
and clinical characteristics by
treatment group
CD conduct disorder; DBDRS disruptive behavior disorder rating scale; FSIQ full scale IQ; M: F, Male
Female; ODD, oppositional defiant disorder; PDISC-IV, diagnostic interview schedule for children, parent
version;
a Continuous data were investigated using ANOVAs. Nominal data were investigated using Pearson’s Chi
squared tests
b Three children were taking dextroamphetamine (two in the EF training condition, and one in the placebo
condition)
Measure Treatment group
EF training Placebo
(n = 31) (n = 30)
M SD M SD F/χ2Group comparisona
Gender (M: F) 25 : 6 – 24 : 6 – .04 ns (p = .949)
Age (years) 10.6 1.4 10.5 1.3 .58 ns (p = .752)
FSIQ 101 11.5 101 11.6 .05 ns (p = .850)
DBDRS parent
Inattention 22.0 3.6 21.9 4.6 .23 ns (p = .924)
Hyperactivity/impulsivity 21.3 3.8 20.5 5.1 .69 ns (p = .458)
ODD 11.6 5.8 11.7 5.9 .40 ns (p = .937)
CD 2.9 3.1 3.2 2.9 .20 ns (p = .701)
DBDRS teacher
Inattention 16.1 5.6 18.0 4.8 1.54 ns (p = .153)
Hyperactivity/impulsivity 13.8 6.2 16.6 6.0 1.84 ns (p = .082)
ODD 7.4 6.0 8.6 6.6 .49 ns (p = .466)
CD 1.1 1.7 1.9 2.5 1.22 ns (p = .184)
PDISC-IV
ODD diagnosis, N (%) 17 (55%) – 15 (50%) – 1.24 ns (p = .705)
ADHD medication–b, N (%) 20 (65%) – 22 (73%) – .56 ns (p = .475)
Computer game experience
(hours per week)
8.6 5.0 11.6 8.4 1.17 ns (p = .105)
Dyscalculia, N (%) 0 (0%) – 0 (0%) – – –
Dyslexia, N (%) 2 (7%) – 5 (17%) – 2.03 ns (p = .211)

Does executive function capacity moderate theoutcome ofexecutive function training inchildren…
1 3
standardized external reward system for completing ses-
sions was used to even further enhance the child’s motiva-
tion for doing the training (for more details see Dovis etal.
2015b). This consisted of receiving game-related stickers,
reward ribbons and medals for completing sessions.
EF training condition
In this condition WM, inhibition and cognitive flexibility
were all in active training mode. Training mode was that
after each block of training tasks, the level of difficulty of
the training task was adjusted automatically to the child’s
level of performance. Also in training mode (a) the WM
task (Dovis etal. 2008a) consisted of five training lev-
els: the first level aims at training visuospatial short-term
memory (STM) only, whereas the other four levels aim at
combinations of visuospatial STM, updating and manipu-
lation of information (i.e. these four levels aimed at both
STM and the central executive). Every level was trained
for 5 of the 25 sessions. The difficulty level increased as
the amount of information that had to be remembered,
updated and manipulated amounted, (b) the inhibition task
(Dovis etal. 2008b) aimed at decreasing the time needed
to inhibit a prepotent response (as in the stop signal reac-
tion time measured by the Stop task; Logan etal. 1997).
On most trials the child responded to a go stimulus by
pressing left or right within a specific time-frame (a green
colored response window between 550 and 850ms; see
Fig.1), thereby creating a prepotent response tendency.
On 25% of the trials, somewhere after the go stimulus and
before the middle of the response window, a stop signal
was presented (a tone and a visual cue). After the stop
signal, the child had to inhibit the prepotent response (stop
trials). Difficulty level increased by shortening the time
for inhibition of this response, (c) the cognitive-flexibility
task (Dovis etal. 2008b) aimed at decreasing the time
a child needs to adapt his/her behavior when task rules
change (i.e., switch cost). The child sorted objects with
various shapes and colors (e.g. blue or red colored plung-
ers and wheels) to either the left or the right according to
a specific rule. This rule was either to sort according to
shape or to sort according to color. In 25% of the trials, the
rule switched (switch trials). Difficulty level increased by
shortening the switch time between the two rules (for more
details of the three training tasks see Van der Oord etal.
2014). To assess whether the training actually improved
task performance on the EFs, improvement on training
performance from beginning to end of training was com-
puted; results showed there was a significant improvement
during the training on inhibition, cognitive flexibility and
for all levels of working memory (see Dovis etal. 2015a,
b, c).
Placebo condition
WM, inhibition and cognitive flexibility were all in placebo
mode in the placebo condition. For the inhibition task and
the cognitive-flexibility task, the stop trials and switch tri-
als were replaced by go trials and non-switch trials (i.e.,
no-stop trials and switch trials were presented), and there
was no adjustment of the difficulty level. Placebo mode in
the WM task was that the difficulty level was not adjusted
to the child’s level of performance and set to a maximum of
two (no more than two items had to be remembered); also
only the WM tasks’ first level was presented for all 25 ses-
sions. The number of trials in placebo mode was increased
to match the training time in training mode; for each EF
domain there was 10-min training per session.
Measures
Near transfer measures
Corsi block tapping task (CBTT)
The CBTT (Corsi 1972) assesses visuospatial STM and WM
capacity. The CBTT consists of nine cubes/blocks positioned
on a board. A similar task to Kessels etal. (2000) was used
(same size of board and blocks, distances between blocks),
and the same procedure was used as in Geurts etal. (2004).
The experimenter tapped a sequence of blocks. The child
is asked to reproduce the sequence in the same (CBTT for-
ward) or in reversed order (CBTT backward). The minimum
sequence length was three, and the maximum sequence
length was eight blocks. Each sequence length was pre-
sented for three trials. The total score is the total amount of
sequences correctly reproduced. Total scores on the CBTT
forward and CBTT backward were used as outcomes for
visuospatial STM and visuospatial WM (for more details,
see the statistical analyses section). The CBTT shows good
reliability (Schellig 1997).
Stop task
The Stop task was used to measure inhibition (Logan etal.
1997). Two types of trials were presented: go trials and stop
trials. During go trials, a go stimulus (an arrow) pointing either
to the right or left was presented. Participants were instructed
to press a response button corresponding to the direction of
the stimulus as quickly and as accurately as possible. Stop
trials were identical to the go trials, but in addition a stop sig-
nal was presented (a tone and a visual cue). Once a stop trial
was presented, the participant had to withhold his/her ongo-
ing response. The delay between the go signal and stop signal
was dynamically varied (in steps of 50ms) so that inhibition

S.Dovis et al.
1 3
was successful in 50% of the stop trials. At this point, the go
process and stop process are of equal duration, which makes
it possible to estimate the stop signal reaction time (SSRT;
Logan, 1997), the latency of the stop process. First two prac-
tice blocks were administered, followed by four experimental
blocks (of 64 trials each). SSRTs were used as inhibition out-
come (for more details, see the statistical analyses section).
Test retest reliability of the SSRT in children with ADHD is
.72 (Soreni etal. 2009).
Trail making test (TMT)
The TMT of the Delis–Kaplan Executive Function System
(D-KEFS; Delis etal. 2007) aims at measuring cognitive flex-
ibility. The TMT is a timed task that requires the individual
to connect a series of letters and numbers in ascending order
while alternating between numbers and letters. Outcomes for
the current study were scaled contrast scores—the contrast
between the scaled non-switch trials (number sequencing and
letter sequencing) and the scaled switch trials (number–letter
switching) (i.e., switch cost; for more details, see the statisti-
cal analyses section). Test–retest reliabilities range from .20
to .77 (Delis etal.).
Far transfer measures
DBDRS (parent andteacher versions)
The DBDRS has four DSM-IV scales; inattention, hyperac-
tivity/impulsivity, oppositional defiant disorder (ODD), and
CD. The child’s behavior is rated by parents and teachers on a
4-point Likert-type scale. Adequate psychometric properties
have been reported (Oosterlaan etal. 2000). The scores on
the inattention and hyperactivity/impulsivity scales were used
ADHD behavior outcomes.
Behavior rating inventory ofexecutive function
questionnaire (BRIEF; Gioia etal. 2000)
EF behavior in everyday life was assessed with the Dutch ver-
sion of the BRIEF. The BRIEF has 75 questions and eight EF
sub-domains: Inhibit, Shift, Emotional Control, Initiate, WM,
Plan/Organize, Organization of Materials, and Monitor. The
test has adequate psychometric properties (Smidts and Huiz-
inga 2009). T scores on the EF sub-domains WM, Inhibit and
Shift (cognitive flexibility) were used as outcomes.
Moderators
Executive functioning
Pre-training total score on the CBTT backward, pre-training
SSRT, and the pre-training scaled contrast score on the TMT
was used as indicators of working memory, inhibition, and
cognitive flexibility capacity, respectively.
Procedure
The faculty’s IRB (the Ethics Review Board of the Faculty of
Social and Behavioral Sciences of the University of Amster-
dam) approved the study. First, written informed consent
was obtained from the parents (on behalf of the participating
children). Next, parents and teachers filled in the DBDRS.
A 6-month version of the DBDRS was administered for
this first screening (regarding the child’s behavior over the
past 6 months). At the pre-test, post-test and follow-up, a
2-week version of the DBDRS was administered (regarding
the child’s behavior over the past 2-weeks). When inclusion
criterion was met on the DBDRS, children and parents were
invited to an intake session. The intake session consisted
of questions regarding demographics (see Table1), and the
PDISC-IV, and the short-form of the WISC-III. If follow-
ing this intake session inclusion criteria were met, parent
and child were invited to the pre-test session and the startup
session. Also they were allocated to one of the treatment
conditions using the process of randomization by minimiza-
tion (Altman & Bland 2005) on the basis of age, gender, IQ,
medication-use (yes/no), and parent- and teacher-rated inat-
tention and hyperactivity/impulsivity symptoms (using the
6-month DBDRS). At pre-test session, outcome measures
were administered. Further, the teacher filled in the 2-week
version of the DBDRS in the same week of the pre-test ses-
sion. The pre-test was planned approximately 1–2weeks
before the startup session of the training. The startup session
was an instruction on the computer, training program and
the external reward system. Also a schedule was established
for implementing the intervention and for weekly coaching
calls. The research assistant that had done a startup session
with a particular family could not test or have further contact
with that family or the teacher (to preserve blinding). Dur-
ing the commencement of the 5-week training, a research
assistant blind to the treatment condition made weekly calls
to monitor progress in the training, motivation and compli-
ance, and assisted with solving technical and game-related
problems. There was an explicit instruction for parents and
children not to discuss the content of the training tasks with
this person. If this person did receive information revealing
the treatment condition, he/she was replaced and could no
longer have contact with the family or the teacher. Between
1 and 2weeks after the last training session the post-test was
planned. The teacher filled in the DBDRS in the same week.
The follow-up was scheduled 3 months after the post-test,
and the teacher completed the DBDRS in the same week
as the follow-up test. Experimenters were blind to condi-
tion in all testing sessions. The effectiveness of blinding, at

Does executive function capacity moderate theoutcome ofexecutive function training inchildren…
1 3
post-test, was assessed by asking the parents to report the
condition they thought their child was assigned to.
Moderation models andstatistical analyses
Single moderation models were used to test whether pre-
training EF (using the pre-training total score on the CBTT
backward, the pre-training SSRT on the Stop task, and the
pre-training scaled contrast score on the TMT) moderated
near and far transfer outcomes of EF training.
Prior to conducting the moderation analyses, for each near
transfer measure and far transfer measure, reliable change
indices (RCI; Jacobson & Truax 1991; Wise 2004) were
calculated and used as measures of pre- to post-, and pre- to
follow-up training change. These RCIs of the near and far
transfer measures were subsequently subjected to modera-
tion analyses, using the PROCESS modeling tool (Hayes
2012), with treatment condition (EF training vs. placebo)
as independent variable, and pre-training EF task scores as
moderators (see Fig.2). The “R2-chng” parameter from the
“R-square increase due to interaction” output from the PRO-
CESS tool was used as a measure of effect size. This param-
eter (hereinafter referred to as R2-change) can be interpreted
as the percentage of variance in the outcome measure that
is due the interaction between the independent variable and
the moderator. Significant moderation effects were further
explored using the Johnson and Neyman method (available
in the PROCESS tool). This method is used to determine
for which values of the moderator the independent variable
significantly predicts the outcome (Field 2013; Hayes 2012).
Given the relation between age and EF (e.g., Westerberg
etal. 2004), EF task scores that were used as modera-
tor were adjusted for age using a regression procedure.
That is, in the entire sample we regressed EF task scores
on age, and the discrepancy between observed and pre-
dicted data was taken as the age-adjusted task score. These
age-adjusted EF task scores were used in the moderation
analyses.
An intent-to-treat (ITT) approach, using single impu-
tations, was used (also see Dovis etal. 2015b). That is,
for each treatment group stochastic regression imputation
was used to predict the missing post-training and follow-
up values. The missing post-test values were based on
the non-missing pre-training and post-training scores of
each treatment group. The missing follow-up values were
based on the non-missing pre-training scores, post-training
scores, follow-up scores, and pre-training—post-training
difference scores of each treatment group (although the
overall percentage of missing data was low—only around
5% was missing—it must be noted that stochastic regres-
sion imputation can increase the probability of making
type I errors).
For each near and far transfer measures, RCI data points
were excluded from analyses (i.e., treated as outliers) if the
absolute value of the standardized residual was greater than
3, or when both of the following criteria were met: (1) a
standardized residual with an absolute value greater than 2,
and (2) a Cook’s distance ≥ 1 (Field 2013). Based on this cri-
terion, one data point was excluded for each of the analyses
that contained one of the following outcome variables: the
pre- to follow-up RCI of the CBTT backwards, the pre- to
post- and the pre- to follow-up RCI of the SSRT, the pre- to
post-RCI of the TMT, the pre- to post-RCI of the parent-
rated BRIEF WM sub-domain, and the pre- to follow-up
RCI of the teacher-rated DBDRS attention scale. Overall,
6 different data points, from 6 different participants, were
excluded (which is only 0.5% of the total amount of data
points).
Fig. 2 The inhibition training
task with the green colored
time-frame (response window)
in the upper middle of the
screen

S.Dovis et al.
1 3
Results
Groups did not differ with respect to any of the baseline
demographics or clinical characteristics (see Table1).
Compliance to treatment was high; of the 31 participants
assigned to the EF training condition, 30 (96.7%) met com-
pliance criteria (completing 25 training sessions within
5weeks). Of the 30 participants assigned to the placebo
condition, 28 (93.3%) met compliance criteria. Further,
three participants (5%) of our total sample (i.e., 1 child in
the training condition and 2 children in the placebo condi-
tion) were, although they completed the training, lost to
post-training testing, and another three participants were
lost to follow-up testing (i.e., 2 children in the training
condition and 1 child in the placebo condition, reason:
unable to schedule or contact). There were no significant
differences on baseline demographics and clinical char-
acteristics between these children and those that did par-
ticipate in the post-training/follow-up assessments. Means
and SDs of the variables involved as well as other details
can be found in Dovis etal. (2015b).
No participant (child, parent, teacher, experimenter, or
coach) was unblinded at any point during the conduct of
the trial, and parents were not able to guess the condition
wherein their child was included (there was no significant
association between the conditions wherein participants
were actually included and the conditions whereof par-
ents afterward reported that their child was assigned to;
see Dovis etal. 2015b). Further, it was tested whether
children improved on the training tasks during the EF
training. Within the EF training condition, paired t-tests
showed a significant difference (improvement) between
the Start Index (result of day 2 and 3 of training) and the
Max Index (result of the 2 best training days) for the inhi-
bition training (p < .001), the cognitive flexibility training
(p < .001), and for all the levels of the WM training (all
p-values< .001). For more details see Dovis etal. (2015b).
Moderation analyses
The results of the moderation analyses are presented in
Table2. These analyses generated four significant modera-
tion effects (see Table2). However, none of these modera-
tion effects survived (Bonferroni) correction for multiple
testing (p values needed to be < .0013 [.05/38] to survive,
whereas actual p-values ranged between .017 and .046).
This suggests that the robustness of these effects is limited.
Nonetheless, to provide more insight into the direction
and effect size of these findings (are they in the expected
direction? are our results related to a lack of power?) the
moderation effects are described in more detail below.
Pre‑training WM
Pre-training WM performance moderated pre- to follow-up
change (RCI) in parent-rated hyperactive/impulsive behav-
ior, b = − 0.37, 95% CI [− 0.73, − 0.008], t = 2.04, p = .046
(also see Table2). R2-change was .040, indicating that only
4% of the variance in the RCI of parent-rated hyperactive/
impulsive behavior could be explained by the interaction
between treatment condition (EF training vs. placebo) and
the moderator (pre-training WM performance). Follow-up
analyses using the Johnson and Neyman method showed
that there only was a significant negative relationship
between treatment condition and the pre- to follow-up
RCI of the P-DBDRS hyperactivity/impulsivity scale in
children with high pre-training WM performance (1.25 SD
above the age corrected mean score on the CBTT back-
wards), whereas this relationship was non-significant in
children with lower pre-training WM performance (see
Fig.3).
These results suggest that, with regard to follow-up
treatment change in parent-rated hyperactivity/impulsiv-
ity behavior, children with very good pre-training working
memory benefit less from the EF training condition than
from the placebo condition. However, the R2-change param-
eter indicates that this effect was small.
Pre‑training response inhibition
Pre-training inhibition performance moderated pre- to
follow-up treatment change in inhibition performance (as
measured by the RCI of the SSRT; see Table2), b = 0.01,
95% CI [0.004, 0.024], t = 2.08, p = .042). R2-change was
.049, indicating that only 4.9% of the variance in the treat-
ment change in inhibition performance could be explained
by the interaction between treatment condition (EF train-
ing vs. placebo) and the moderator (pre-training inhibi-
tion performance). Follow-up analyses using the Johnson
and Neyman method showed that there only was a sig-
nificant positive relationship between treatment condition
and the pre- to follow-up RCI of the SSRT in children
with medium to high pre-training SSRTs (note: higher
SSRTs means worse inhibition performance), whereas
this relationship was non-significant in children with lower
pre-training SSRTs (lower than 0.5 SD below the mean
[mean = 196ms; SD = 58ms]; see Fig.4).
These results suggest that, with regard to follow-up
treatment change in response inhibition, only children with
medium to poor pre-training inhibition benefit more from
the EF training condition than from the placebo condition.
However, the R2-change parameter indicates that this effect
was small.

Does executive function capacity moderate theoutcome ofexecutive function training inchildren…
1 3
Pre‑training cognitive exibility
Pre-training cognitive flexibility performance moderated
pre- to follow-up treatment change (RCI) in cognitive
flexibility performance, b = 0.15, 95% CI [0.027, 0.265],
t = 2.45, p = .017 (see Table2). R2-change was .071, indicat-
ing that only 7.1% of the variance in the treatment change
in cognitive flexibility performance could be explained by
the interaction between treatment condition (EF training vs.
placebo) and the moderator (pre-training cognitive flexibility
performance). Follow-up analyses using the Johnson and
Neyman method showed a significant negative relationship
between treatment condition and the pre- to follow-up RCI
of the TMT score in children with low pre-training cog-
nitive flexibility (lower than 1.25 SD below the mean), a
non-significant relationship in children with moderately low
Table 2 Moderation outcomes (EF training condition versus placebo condition)
Age corr., age corrected performance; Bkw., backward; BRIEF, behavior rating inventory of executive function questionnaire; CBTT, corsi block
tapping task; CF, cognitive flexibility; DBDRS, disruptive behavior disorder rating scale; Mod, moderator; P-, parent-rated; Pre, Pre-test; Pre-tr.,
Pre-training; R2-cng, “R2-change” parameter; RCI, reliable change index (for all outcome measures RCI scores were used); shift, cognitive flexibil-
ity; SSRT, stop signal reaction time; T-, teacher-rated; TMT, trail making task; treatment, treatment condition
*p < .05; **p < .01; ***p < .001; †p < .07
Outcome measure (RCI) Pre- versus post-test Pre- versus follow-up test
Coefficient (b) Coefficient (b)
Independent
variable (Treat-
ment)
Mod Treatment x Mod
(Moderation
effect)
R2-cng Independent
variable (Treat-
ment)
Mod Treatment x Mod
(Moderation
effect)
R2-cng
Mod = Pre-tr. WM (age
corr. CBTT bkw. total
score)
Near transfer
CBTT backward .59* − .13 .25 .057 .46* − .28*** .12 .012
CBTT forward 1.01** − .07 .11 .008 .89** − .01 .09 .020
Far Transfer
P-DBDRS att .33 − .21 − .17 .005 .24 − .02 .22 .010
P-DBDRS hyp/imp .03 − .03 − .24 .015 − .24 − .06 − .37* .040
T-DBDRS att .45 .08 .06 .001 .38 .29* .41† .058
T-DBDRS hyp/imp .22 .07 .35 .043 − .07 .23 .45 .054
P-BRIEF WM .21 − .17 − .30 .048 .01 − .12 − .24 .038
Mod = Pre-tr. Inh. (age
corr. SSRT)
Near transfer
SSRT 1.10 *** .01** .005 .009 1.14*** .02*** .01* .049
Far transfer
P-DBDRS att .46 − .001 .001 .001 .32 .01 .007 .006
P-DBDRS hyp/imp .005 − .004 − .001 < .001 − .23 − .003 − .002 .001
T-DBDRS att .42 .002 < .001 < .001 .22 .003 .014 .041
T-DBDRS hyp/imp .17 − .001 − .005 .004 − .23 − .002 < .001 < .001
P-BRIEF Inhibition − .65 − .01 − .16 .003 − .75 .07 − .14 .002
Mod = Pre-tr. CF (age corr.
TMT)
Near transfer
TMT − .08 − .20*** .09 .021 − .03 − .24*** .18* .071
Far transfer
P-DBDRS att .46 .03 .04 < .001 .27 − .03 − .06 .001
P-DBDRS hyp/imp .06 − .08 − .06 .001 − .20 .03 − .10 .004
T-DBDRS att .41 − .06 .18 .015 .20 − .006 .02 < .001
T-DBDRS hyp/imp .21 − .15* .32* .057 − .22 .01 − .24 .025
P-BRIEF Shift − .65 − .07 − .18 .008 − .75 − .002 − .27 .016

S.Dovis et al.
1 3
Fig. 3 Pre-training WM per-
formance moderation on pre to
follow-up change in parent-rated
hyperactive/impulsive behavior
Fig. 4 Pre-training inhibition
performance moderation on pre
to follow-up change in inhibi-
tion performance

Does executive function capacity moderate theoutcome ofexecutive function training inchildren…
1 3
to moderately high pre-training cognitive flexibility, and a
significantly positive relationship in children with very high
pre-training cognitive flexibility (higher than 1.5 SD above
the mean; see Fig.5).
These results suggest that, with regard to post-treatment
change in cognitive flexibility, children with very poor pre-
training cognitive flexibility benefit more from the placebo
condition than from the EF training condition, whereas chil-
dren with very good pre-training cognitive flexibility show
a worse outcome in the placebo condition than in the EF
training condition. However, the R2-change parameter indicates
that this effect was small.
Pre-training cognitive flexibility performance also mod-
erated pre- to post-treatment change (RCI) in teacher-rated
hyperactive/impulsive behavior, b = 0.32, 95% CI [0.028,
0.611], t = 2.20, p = .03 (see Table2). R2-change was .057,
indicating that only 5.7% of the variance in the RCI of
teacher-rated hyperactive/impulsive behavior could be
explained by the interaction between treatment condition
(EF training vs. placebo) and the moderator (pre-training
cognitive flexibility performance). Follow-up analyses using
the Johnson and Neyman method showed that children with
very good pre-training cognitive flexibility (2 SD above the
mean) benefited more from the EF training condition than
from the placebo condition. However, inspection of Fig.6
suggests that pre-training cognitive flexibility capacity
only has impact on teacher-rated hyperactivity/impulsiv-
ity in the placebo condition. In the placebo condition better
pre-training cognitive flexibility seems to be associated with
worse hyperactivity/impulsivity outcomes. However, the R2-
change parameter indicates that this effect was small.
In sum, although pre-training inhibition performance and
pre-training cognitive flexibility performance were signifi-
cant moderators of near transfer, and pre-training WM per-
formance and pre-training cognitive flexibility performance
were significant moderators of far transfer, these modera-
tion effects were often not in the expected direction, did not
survive Bonferroni correction for multiple testing and were
characterized by small effect sizes.
Discussion
The aim of this placebo-controlled study was to determine
whether pre-training EF capacity of children with ADHD
moderates the outcome of an EF training intervention on
measures of near transfer (EF performance) and far trans-
fer (parent- and teacher-rated ADHD symptoms and par-
ent-rated EF behavior in everyday life). We expected that
children with poorer pre-training EF capacity would ben-
efit more from EF training than from a placebo training, as
they have more EF-related room for improvement (Diamond
and Lee 2011; Diamond 2012), whereas in children with
good pre-training EF capacity, EF training would probably
have no more impact on ADHD symptoms than a placebo
Fig. 5 Pre-training cogni-
tive flexibility performance
moderation on pre to follow-up
change in cognitive flexibility
performance

S.Dovis et al.
1 3
training, as their symptoms are less likely to originate from
impairments in EF.
However, our results are not in line with these expec-
tations. That is, although we found that pre-training
inhibition performance and pre-training cognitive flex-
ibility performance were significant moderators of near
transfer (pre- to follow-up treatment change in inhibition
performance and cognitive flexibility performance), and
pre-training WM performance and pre-training cognitive
flexibility performance were significant moderators of far
transfer (treatment change in parent-rated and teacher-
rated hyperactive/impulsive behavior, respectively), these
moderation effects were often not in the expected direc-
tion, did not survive Bonferroni correction for multiple
testing, and were characterized by small effect sizes. This
suggests that these effects are not robust and are unlikely
to be of clinical significance. To illustrate the latter, the
effect sizes indicated that only 4–7% of the variance in
the observed treatment change could be explained by the
interaction between the type of treatment (EF training vs.
Placebo) and pre-training EF. Although the non-robustness
of our effects might be explained by our relatively small
sample size, using a larger sample is unlikely to change
the effect sizes and the conclusions regarding the clinical
significance of the effects. In sum, these results suggest
that children’s pre-training EF capacity is not a clini-
cally significant moderator of the relation between type
of treatment (EF training vs. Placebo) and improvements
on measures of near transfer (EF performance) and far
transfer (parent- and teacher-rated ADHD symptoms and
parent-rated EF behavior in everyday life). Hence, com-
pared to a placebo training, children with poor EF capacity
do not seem to benefit more from EF training than children
with good EF capacity.
EF training interventions in children with ADHD mainly
improve performance on measures of near transfer, but
have very limited effects on measures of far transfer (Cor-
tese etal. 2014; Dovis etal. 2015a; Hodgson etal. 2014;
Sonuga-Barke etal. 2013; also see Chacko etal. 2013). Con-
sequently, it has been suggested that these findings might
have been more positive if only those children with ADHD
who actually have EF impairments were selected for training
(e.g., Cortese etal. 2014). However, our current findings do
not support this suggestion and imply that the strategy of
training only those children who have EF impairments will
probably not change the conclusions of these meta-analyses.
Furthermore, our findings do not change the conclusion
from our previous placebo-controlled study (Dovis etal.
2015b) stating that changes in EF performance seem unre-
lated to the changes in ADHD symptoms and EF behavior
(EF performance only improved in the EF training condition,
whereas the far transfer indices improved irrespective of the
type of treatment received; see Dovis etal. 2015b), and are
in line with the notion that improvement of EF might not
be the mechanism of change when it comes to improving
ADHD symptoms or EF behavior in everyday life.
Fig. 6 Pre-training cognitive
flexibility performance modera-
tion on pre to post treatment
change in teacher-rated hyperac-
tive/impulsive behavior

Does executive function capacity moderate theoutcome ofexecutive function training inchildren…
1 3
If not improvement in EF, what else could this mecha-
nism of change be? The improvements in ADHD and EF
behaviors are probably not caused by a Hawthorne effect,
nor by effects of multiple testing or the passage of time, as a
previous study investigating the EF training (Van der Oord
etal. 2014) found no improvement on parent- and teacher-
rated ADHD and EF behaviors in a wait-list control group.
Nonetheless, at this point we can only speculate about the
nature of the underlying mechanism(s) of change. It must
be something that is common to both treatment conditions.
For instance, in both the EF training and the placebo condi-
tions, training tasks were gamified and parents were pro-
vided with a standardized external reward system to keep
children motivated to adhere to treatment. If children were
indeed motivated to adhere to this 25-session, home-based
treatment, which is consistent with the high compliance rate
in our study, then one could imagine that parents may have
had less need for negative interactions and more opportuni-
ties for positive interactions with their child. To elaborate on
the latter; the achievements in the game (e.g., creating new
inventions) and in the standardized external reward system
(e.g., earning stickers, ribbons and medals) may have made
it easier for parents to detect and use these opportunities
for positive interactions with their child. Evidence suggests
that decreased negative and increased positive parent–child
interactions can improve ADHD-related behavior, even in
the classroom (e.g., see Hinshaw 2007; Matos etal. 2009).
Future EF training studies should include process measures
to further investigate this and other potential mechanisms of
change (such as effects of expectancies, self-fulfilling proph-
ecies, or attribution; see Dovis etal. 2015b; Hinshaw 2007).
In its current form, regardless of children’s pre-training
EF capacity, EF training seems not more effective than
a placebo training in improving symptoms of ADHD or
EF behavior in everyday life. Nonetheless, there are still
opportunities that need further exploration. For example,
to increase chances of finding far transfer effects that result
from EF training specifically, training tasks should be
made more ecologically valid (e.g., by using EF training
tasks that resemble the complexity of problematic situa-
tions in everyday life) and should be intertwined with rel-
evant real-life EF-taxing activities (e.g., completing chores
in everyday life could be an additional goal in the EF
training; for more suggestions see Gathercole 2014; also
see Van der Donk etal. 2016). Also potentially training
focused on enhancing mainly the central executive com-
ponent of working memory may be more effective as the
central executive is most disturbed in ADHD and related
to deficits in functioning (Chacko etal. 2014b; Rapport
etal. 2013), for a promising example see (Kofler etal.
2018b). Furthermore, the domains of far transfer that were
investigated in this study were limited to indirect meas-
ures of behavior (e.g., ADHD behavior as rated by parents
and teachers). Future studies should also include more
direct measures of behavior or potentially more relevant
far transfer measures. More relevant far transfer measure
than EF and ADHD ratings of parents may be social and
academic functioning, research shows clear associations
between working memory capacity and these domains
(Kofler etal. 2018a, c). For example, a placebo-controlled
WM-training study (Green etal. 2012) found no specific
treatment effects on parent-rated behavior (teacher-rated
behavior was not investigated), but found specific effects
on aspects of experimenter-observed off-task behavior dur-
ing an academic task. Finally, future studies should use
larger sample sizes. Given the performed moderation anal-
yses, our sample size was relatively small (N = 61). This
suggests that the null findings in this moderation study
should be interpreted with caution (due to the possibility
of type II error). Nonetheless, all null findings were char-
acterized by small effect sizes suggesting that a replication
study using a larger sample is not likely to find more clini-
cally relevant results.
With regard to operationalization of our moderators, a
potential limitation of the current study is that we used the
scores on the CBTT (forward and backward) as measure-
ment of WM, with as limitation that this measure seems to
be mainly associated with the STM component of WM, but
less with its central executive (CE) component (Kessels etal.
2008). Given the evidence that children with ADHD seem
to be impaired on both the STM and CE component of WM
(e.g., see Dovis etal. 2015a, b, c) and the fact that the WM-
training paradigm of the EF-training condition was designed
to target both the STM and CE component of WM, it would
have been interesting to investigate our research questions
with a more CE-oriented WM task such as the Chessboard
task (e.g., Dovis etal. 2013) or the N-Back task (Kane etal.
2007). Further, although the theoretical reasons for using
the contrast score from the D-KEFS TMT as the measure
of cognitive flexibility (task switching) are strong, it must
be noted that its test–retest reliabilities are low (see Craw-
ford etal. 2008). One could argue that including the results
from the original switch trials (scores from D-KEFS TMT 4)
might reduce this limitation as these “non-contrasted” scores
are comprised of only one source of measurement error
instead of two (Crawford etal. 2008). However, evidence
suggests that these “non-contrasted” scores also have low
test–retest reliability (r = .20; Delis etal. 2001) but, in con-
trast to the TMT contrast scores, have low construct validity
(see Sánchez-Cubillo etal. 2009: they found that the TMT
switch trials primarily reflect working memory, whereas the
TMT contrast scores primarily reflect task switching). We
therefore chose to only use the TMT contrast scores as meas-
ure of cognitive flexibility/task switching in our moderation
analyses. Finally multiple measures of one EF construct is
preferred; however, given that multiple EFs were trained in

S.Dovis et al.
1 3
this study and children already had long pre- and post-test
sessions, adding more EF measures was not feasible for the
participants
Based on our current findings, what would be our answer
when clinicians, parents or teachers ask us whether a par-
ticular child with ADHD could benefit from EF training?
It would probably be something like this: “In general, per-
formance on outcome measures of working memory and
inhibition seem to improve more than after placebo training
(Dovis etal. 2015b). However, since many of these outcome
measures are very similar to the training tasks themselves
we do not know if and to what extend this improvement is
the result of a learned strategy instead of improved cognitive
capacity (Thompson etal. 2013). ADHD symptoms and EF
behavior in everyday life might also improve (according to
parents ADHD symptoms improve in about 39–55% of the
cases, and EF behavior improves in about 26–55% of the
cases; according to teachers ADHD symptoms improve in
about 16–39% of the cases; see Dovis etal. 2015b), but the
same improvement is found after placebo training. More-
over, these outcomes seem independent of the child’s EF
capacity. That is, compared to a placebo training, children
with poor EF capacity do not seem to benefit more from
EF training than children with good EF capacity. In sum,
these findings suggest that if the ADHD or EF behavior of
the child improves after EF training, this is probably not the
result of the actual improvement of EFs, but of some other
yet unknown mechanism of change. At this point, we can
only speculate about the nature of this unknown underlying
mechanism(s) of change (e.g. effects of expectancies, self-
fulfilling prophecies, attribution, or improved parent–child
interactions); however, improvement of EFs seems to have
little to do with it.”
In conclusion, we found that children’s pre-training EF
capacity is not a clinically significant moderator of the rela-
tion between type of treatment (EF training vs. Placebo)
and improvements on measures of near transfer (EF perfor-
mance) and far transfer (parent- and teacher-rated ADHD
symptoms and parent-rated EF behavior in everyday life).
Hence, it does not seem to be the case that especially chil-
dren with poor pre-training EF capacity benefit more from
EF training than from placebo training.
Acknowledgements We are grateful to Jeugdriagg Noord Holland
Zuid, GGz Noord Holland Noord (Centrum voor Kinder- en Jeug-
dpsychiatrie), Regionaal Centrum voor Kinder en Jeugdpsychiatrie
Gooi en Vechtstreek (RCKJP), Bosman GGz, Stichting De Praktijk,
Stichting Kram, PuntP, Academisch Behandelcentrum UvA Minds,
and Kinderpraktijk VIS, to multimedia company ShoSho for the gami-
fication of Brian Game Brian, to Hilde Huizenga and Joost Agelink
van Rentergem Zandvliet for their comments and statistical advice, to
Marloes van der Arend, Tim van den Broek, Josje de Bont, Annette
Brouwer, Tycho Dekkers, Lucie van den Eertwegh, Rebecca Goedee,
Roza van der Heide, Lisanne Klink, Astrid Nauta, Inge Meulenberg,
Muriël Musa, Pascale Riaskoff, Elise Tilma, Marije Voermans, Ida de
Vries, and Pamina Warmbrunn for their help with data collection, and
to all participating children and families.
Compliance with ethical standards
Conflict of interest S.D., S.VDO and M.M. declare no competing in-
terests relating to this paper. P.J.M.P. was member of Stichting Gaming
& Training (until 2017), a nonprofit organization that facilitates the
development and implementation of the above-mentioned EF training;
“Braingame Brian.”
References
Altman DG, Bland JM (2005) Treatment allocation by minimisation.
BMJ 330:843
American Psychiatric Association (2000) Diagnostic and statistical
manual of mental disorders, 4th revised edition (DSM-IV-TR).
American Psychiatric Association, Washington, DC
Barkley RA (2006) Attention-deficit hyperactivity disorder. A hand-
book for diagnosis and treatment, 3rd edn. Guilford Press, New
Yor k
Chacko A, Feirsen N, Bedard A-C, Marks D, Uderman JZ, Chimiklis A
(2013) Cogmed working memory training for youth with ADHD:
a closer examination of efficacy utilizing evidence-based criteria.
J Clin Child Adolesc Psychol 42:769–783
Chacko A, Bedard AC, Marks DJ, Feirsen N, Uderman JZ, Chimik-
lis A, Ramon M (2014a) A randomized clinical trial of cogmed
working memory training in school-age children with ADHD: a
replication in a diverse sample using a control condition. J Child
Psychol Psychiatry 55:247–255
Chacko A, Kofler M, Jarrett M (2014b) Improving outcomes for youth
with ADHD: a conceptual framework for combined neurocogni-
tive and skill-based treatment approaches. Clin Child Fam Psychol
Rev 17(4):368–384. https ://doi.org/10.1007/s1056 7-014-0171-5
Corsi PM (1972) Human memory and the medial temporal region of
the brain. Diss Abstr Int 34:819B
Cortese S, Ferrin M, Brandeis D, Buitelaar J, Daley D, Dittmann RW,
Sonuga-Barke EJS (2014) Cognitive training for attention-deficit/
hyperactivity disorder: meta-analysis of clinical and neuropsycho-
logical outcomes from randomized controlled trials. J Am Acad
Child Adolesc Psychiatry 54:164–174
Crawford JR, Sutherland D, Garthwaite PH (2008) On the reliability
and standard errors of measurement of contrast measures from the
D-KEFS. J Int Neuropsychol Soc 14:1069–1073
Crosbie J, Arnold P, Paterson A, Swanson JM, Dupuis A, Li X,
Schachar RJ (2013) Response inhibition and ADHD traits: cor-
relates and heritability in a community sample. J Abnorm Child
Psychol 41:497–507
Delis DC, Kaplan E, Kramer JH (2001) Delis-Kaplan executive func-
tion system (D-KEFS) examiner’s manual. The Psychological
Corporation, San Antonio, Texas, USA
Delis DC, Kaplan E, Kramer JH (2007) Delis-Kaplan executive func-
tion system: examiner’s manual, trial making test. Harcourt
Assessment, San Antonio, TX
Diamond A (2012) Activities and programs that improve children’s
executive functions. Curr Dir Psychol Sci 21:335–341
Diamond A, Lee K (2011) Interventions shown to aid executive func-
tion development in children 4–12years old. Science (New York,
N.Y.) 333(6045):959–964
Dovis S (2014) Cognitive control and motivation in children with
ADHD: How reinforcement interacts with the assessment and
training of executive functioning. Doctoral thesis, University of
Amsterdam

Does executive function capacity moderate theoutcome ofexecutive function training inchildren…
1 3
Dovis S, Ponsioen A, Geurts H, Ten Brink E, Van der Oord S, Prins
PJM (2008a) Executive function training tasks: working mem-
ory. Theoretical background and parameters. In: Task force
ADHD & computer. University of Amsterdam, Amsterdam
Dovis S, Geurts H, Ponsioen A, Ten Brink E, Van der Oord S, Prins
PJM (2008b) Executive function training tasks: inhibition and
cognitive flexibility. Theoretical background and parameters.
In: Task force ADHD & computer. University of Amsterdam,
Amsterdam
Dovis S, Van der Oord S, Wiers RW, Prins PJM (2013) What part of
working memory is not working in ADHD? Short-term memory,
the central executive and effects of reinforcement. J Abnorm
Child Psychol 41(6):901–917. https ://doi.org/10.1007/s1080
2-013-9729-9
Dovis S, Agelink van Rentergem J, Huizenga HM (2015a) Does
cogmed working memory training really improve inattention
in daily life? A reanalysis [letter to the editor]. PLoS ONE
10:e85992
Dovis S, Van der Oord S, Huizenga HM, Wiers RW, Prins PJM
(2015b) Prevalence and diagnostic validity of motivational
impairments and deficits in visuospatial short-term memory
and working memory in ADHD subtypes. Eur Child Ado-
lesc Psychiatry 24(5):575–590. https ://doi.org/10.1007/s0078
7-014-0612-1
Dovis S, Van der Oord S, Wiers RW, Prins PJM (2015c) Improving
executive functioning in children with ADHD: training multiple
executive functions within the context of a computer game. A
randomized double-blind placebo controlled trial. PLOS ONE
10:e0121651
Durston S, van Belle J, de Zeeuw P (2011) Differentiating frontostri-
atal and fronto-cerebellar circuits in attention-deficit/hyperactivity
disorder. Biol Psychiatry 69:1178–1184
Fair DA, Bathula D, Nikolas MA, Nigg JT (2012) Distinct neuropsy-
chological subgroups in typically developing youth inform
heterogeneity in children with ADHD. Proc Natl Acad Sci
109:6769–6774
Field A (2013) Discovering statistics using SPSS, 4th edn. SAGE Pub-
lications, London
Gathercole SE (2014) Commentary: working memory training and
ADHD—Where does its potential lie? reflections on Chacko etal.
(2014). J Child Psychol Psychiatry 55:256–257
Geurts HM, Verté S, Oosterlaan J, Roeyers H, Sergeant JA (2004)
How specific are executive functioning deficits in attention deficit
hyperactivity disorder and autism? J Child Psychol Psychiatry
45:836–854
Gioia GA, Isquith PK, Guy SC, Kenworthy L (2000) Behavior rat-
ing inventory of executive function. Psychological Assessment
Resources, Odessa
Green CT, Long DL, Green D, Iosif A-M, Dixon JF, Miller MR, Sch-
weitzer JB (2012) Will working memory training generalize to
improve off-task behavior in children with attention-deficit/hyper-
activity disorder? Neurotherapeutics 9:639–648
Greenhill LL (1998) Childhood attention deficit hyperactivity disor-
der: pharmacological treatments. In: Nathan PE, Gorman J (eds)
A guide to treatments that work. Oxford University Press, New
York, pp 42–64
Hayes AF (2012) PROCESS: a versatile computational tool for
observed variable mediation, moderation, and conditional process
modeling [White paper]. Retrieved from http://www.afhay es.com/
publi c/proce ss201 2.pdf
Hinshaw SP (2007) Moderators and mediators of treatment outcome
for youth with ADHD: understanding for whom and how interven-
tions work. J Pediatr Psychol 32:664–675
Hodgson K, Hutchinson AD, Denson L (2014) Nonpharmacological
treatments for ADHD: a meta-analytic review. J Atten Disord
18:275–282
Hunt AD, Kronenberger WG, Dunn DW, Gibson BS, Gondoli DM
(2014) Predictors of change in short-term memory span following
working memory training. Acta Neuropsychologica 12:445–458
Isquith PK, Roth RM, Gioia G (2013) Contribution of rating scales to
the assessment of executive functions. Appl Neuropsychol Child
2:125–132
Jacobson NS, Truax P (1991) Clinical significance: a statistical
approach to defining meaningful change in psychotherapy
research. J Consult Clin Psychol 59:12–19
Kane MJ, Conway ARA, Miura TK, Colflesh GJH (2007) working
memory, attention control, and the N-back task: a question of con-
struct validity. J Exp Psychol Learn Mem Cognit 33(3):615–622.
https ://doi.org/10.1037/0278-7393.33.3.615
Kessels RPC, van Zandvoort MJ, Postma A, Kappelle LJ, de Haan EH
(2000) The Corsi block-tapping task: standardization and norma-
tive data. Appl Neuropsychol 7:252–258
Kessels RPC, van den Berg E, Ruis C, Brands AMA (2008) The back-
ward span of the corsi block-tapping task and its association with
the WAIS-III digit span. Assessment 15(4):426–434. https ://doi.
org/10.1177/10731 91108 31561 1
Klingberg T (2010) Training and plasticity of working memory. Trends
Cognit Sci 14:317–324
Klingberg T, Forssberg H, Westerberg H (2002) Training of work-
ing memory in children with ADHD. J Clin Exp Neuropsychol
24:781–791
Klingberg T, Fernell E, Olesen PJ, Johnson M, Gustafsson P, Dahl-
ström K, Westerberg H (2005) Computerized training of working
memory in children with ADHD—a randomized, controlled trial.
J Am Acad Child Adolesc Psychiatry 44:177–186
Kofler MJ, Harmon SL, Aduen PA, Day TN, Austin KE, Spiegel JA,
Sarver DE (2018a) Neurocognitive and behavioral predictors of
social problems in ADHD: a Bayesian framework. Neuropsychol-
ogy 32(3):344–355. https ://doi.org/10.1037/neu00 00416
Kofler MJ, Sarver DE, Austin KE, Schaefer HS, Holland E, Aduen
PA, Lonigan CJ (2018b) Can working memory training work
for ADHD? Development of central executive training and com-
parison with behavioral parent training. J Consult Clin Psychol
86(12):964–979. https ://doi.org/10.1037/ccp00 00308
Kofler MJ, Spiegel JA, Soto EF, Irwin LN, Wells EL, Austin KE
(2018c) Do working memory deficits underlie reading problems
in attention-deficit/hyperactivity disorder (ADHD)? J Abnorm
Child Psychol 8:8–9. https ://doi.org/10.1007/s1080 2-018-0447-1
Kort W, Compaan EL, Bleichrodt N, Resing WCM, Schittekatte M,
Bosmans M, Vermeir G, Verhaeghe P (2002) WISC-III NL Han-
dleiding (Dutch Manual). NIP, Amsterdam
Kraemer HC, Frank E, Kupfer DJ (2006) Moderators of treatment out-
comes. JAMA 296:1286
Kray J, Karbach J, Haenig S, Freitag C (2012) Can task-switching train-
ing enhance executive control functioning in children with atten-
tion deficit/-hyperactivity disorder? Front Hum Neurosci 5:180
Logan GD, Schachar RJ, Tannock R (1997) Impulsivity and inhibitory
control. Psychol Sci 8:60–64
Maric M, Prins PJM, Ollendick TH (eds) (2015) Moderators and
mediators of youth treatment outcomes. Oxford University Press,
New York
Martinussen R, Hayden J, Hogg-Johnson S, Tannock R (2005) A
meta-analysis of working memory impairments in children with
attention-deficit/hyperactivity disorder. J Am Acad Child Adolesc
Psychiatry 44(4):377–384. https ://doi.org/10.1097/01.chi.00001
53228 .72591 .73
Matos M, Bauermeister JJ, Bernal G (2009) Parent-child interac-
tion therapy for Puerto Rican preschool children with ADHD
and behavior problems: a pilot efficacy study. Fam Process
48:232–252
McNab F, Leroux G, Strand F, Thorell LB, Bergman S, Klingberg
T (2008) Common and unique components of inhibition and

S.Dovis et al.
1 3
working memory: an fMRI, within-subjects investigation. Neu-
ropsychologia 46:2668–2682
Miyake A, Friedman NP (2012) The nature and organization of indi-
vidual differences in executive functions: four general conclu-
sions. Curr Dir Psychol Sci 21:8–14
Nigg JT (2006) What causes ADHD? Understanding what goes wrong
and why. Guilford Press, New York
Nigg JT, Willcutt EG, Doyle AE, Sonuga-Barke EJS (2005) Causal
heterogeneity in attention-deficit/hyperactivity disorder: Do we
need neuropsychologically impaired subtypes? Biol Psychiatry
57:1224–1230
Oosterlaan J, Scheres A, Antrop I, Roeyers H, Sergeant JA (2000)
Vragenlijst voor Gedragsproblemen bij Kinderen (VvGK) (Dutch
translation of the DBD Rating Scale). Swets & Zeitlinger, Lisse,
The Netherlands
Pelham WE, Gnagy EM, Greenslade KE, Milich R (1992) Teacher
rating of DSM-III-R symptoms for disruptive behavior disorders.
J Clin Child Psychol 8:259–262
Prins PJM, Brink E Ten, Dovis S, Ponsioen A, Geurts HM, de Vries M,
Van der Oord S (2013) “Braingame brian”: toward an executive
function training program with game elements for children with
ADHD and cognitive control problems. Games Health J 2:44–49
Rapport MD, Chung KM, Shore G, Isaacs P (2001) A conceptual
model of child psychopathology: implications for understanding
attention deficit hyperactivity disorder and treatment efficacy. J
Clin Child Psychol 30:48–58
Rapport MD, Orban SA, Kofler MJ, Friedman LM (2013) Do programs
designed to train working memory, other executive functions, and
attention benefit children with ADHD? A meta-analytic review
of cognitive, academic, and behavioral outcomes. Clin Psychol
Rev 33(8):1237–1252. https ://doi.org/10.1016/j.cpr.2013.08.005
Sánchez-Cubillo I, Periáñez JA, Adrover-Roig D, Rodríguez-Sánchez
JM, Ríos-Lago M, Tirapu J, Barceló F (2009) Construct validity
of the trail making test: role of task-switching, working memory,
inhibition/interference control, and visuomotor abilities. J Int Neu-
ropsychol Soc 15(3):438–450
Sarver DE, Rapport MD, Kofler MJ, Raiker JS, Friedman LM (2015)
Hyperactivity in attention-deficit/hyperactivity disorder (ADHD):
Impairing deficit or compensatory behavior? J Abnorm Child Psy-
chol 43:1219–1232
Sattler JM (2001) Assessment of children: cognitive applications, 4th
edn. Author, San Diego
Schecklmann M, Ehlis A-C, Plichta MM, Dresler T, Heine M, Boreatti-
Hümmer A, Fallgatter AJ (2013) Working memory and response
inhibition as one integral phenotype of adult ADHD? A behav-
ioral and imaging correlational investigation. J Atten Disord
17:470–482
Schellig D (1997) Block-tapping-test. Swets and Zeitlinger, Lisse
Shaffer D, Fisher P, Lucas CP, Dulcan MK, Schwab-Stone ME (2000)
NIMH diagnostic interview schedule for children version IV
(NIMH DISC-IV): description, differences from previous ver-
sions, and reliability of some common diagnoses. J Am Acad
Child Adolesc Psychiatry 39:28–38
Smidts DP, Huizinga M (2009) BRIEF executieve functies gedragsvra-
genlijst: Handleiding (BRIEF Dutch manual). Hogrefe Uitgevers,
Amsterdam
Smith AB, Taylor E, Brammer M, Toone B, Rubia K (2006) Task-
specific hypoactivation in prefrontal and temporoparietal brain
regions during motor inhibition and task switching in medication-
naive children and adolescents with attention deficit hyperactivity
disorder. Am J Psychiatry 163:1044–1051
Sonuga-Barke EJS, Brandeis D, Cortese S, Daley D, Ferrin M, Holt-
mann M, Sergeant JA (2013) Nonpharmacological interventions
for ADHD: systematic review and meta-analyses of randomized
controlled trials of dietary and psychological treatments. Am J
Psychiatry 170:275–289
Soreni N, Crosbie J, Ickowicz A, Schachar RJ (2009) Stop signal and
Conners’ continuous performance tasks: test–retest reliability
of two inhibition measures in ADHD children. J Atten Disord
13:137–143
Thompson TW, Waskom ML, Garel K-LA, Cardenas-Iniguez C,
Reynolds GO, Winter R, Gabrieli JDE (2013) Failure of work-
ing memory training to enhance cognition or intelligence. PLoS
ONE 8:e63614
Tillman CM, Eninger L, Forssman L, Bohlin G (2011) The rela-
tion between working memory components and ADHD symp-
toms from a developmental perspective. Dev Neuropsychol
36(2):181–198
Titz C, Karbach J (2014) Working memory and executive func-
tions: effects of training on academic achievement. Psychol Res
78:852–868
van der Donk MLA, Hiemstra-Beernink AC, Tjeenk-Kalff AC, Van
der Leij A, Lindauer RJL (2016) Predictors and moderators of
treatment outcome in cognitive training for children with ADHD.
J Atten Disord
Van der Oord S, Daley D (2015) Moderators and mediators of treat-
ments for youth with ADHD. In: Maric M, Prins PJM, Ollendick
TH (eds) Moderators and mediators of youth treatment outcomes.
Oxford University Press, New York, pp 123–145
Van der Oord S, Ponsioen AJGB, Geurts HM, Ten Brink EL, Prins PJM
(2014) A pilot study of the efficacy of a computerized executive
functioning remediation training with game elements for children
with ADHD in an outpatient setting: outcome on parent- and
teacher-rated executive functioning and ADHD behavior. J Atten
Disord 18:699–712
Van Dongen-Boomsma M, Vollebregt MA, Buitelaar JK, Slaats-Wil-
lemse D (2014) Working memory training in young children with
ADHD: a randomized placebo-controlled trial. J Child Psychol
Psychiatry 55:886–896
Westerberg H, Hirvikoski T, Forssberg H, Klingberg T (2004) Visuo-
spatial working memory span: a sensitive measure of cognitive
deficits in children with ADHD. Child Neuropsychol J Norm
Abnorm Dev Child Adolesc 10:155–161
Willcutt EG, Doyle AE, Nigg JT, Faraone SV, Pennington BF
(2005) Validity of the executive function theory of attention-
deficit/hyperactivity disorder: a meta-analytic review. Biol
Psychiatry 57(11):1336–1346. https ://doi.org/10.1016/j.biops
ych.2005.02.006
Willcutt EG, Nigg JT, Pennington BF, Solanto MV, Rohde LA, Tan-
nock R etal (2012) Validity of DSM–IV attention deficit/hyper-
activity disorder symptom dimensions and subtypes. J Abnorm
Psychol 121:991–1010
Wise EA (2004) Methods for analyzing psychotherapy outcomes: a
review of clinical significance, reliable change, and recommenda-
tions for future directions. J Pers Assess 82(1):50–59
Wong CG, Stevens MC (2012) The effects of stimulant medication
on working memory functional connectivity in attention-deficit/
hyperactivity disorder. Biol Psychiatry 71:458–466
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.