SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
Suboptimal decision making by children with ADHD in the face of risk: Poor risk
adjustment and delay aversion rather than general proneness to taking risks
Lin Sørensen1,2,3, Edmund Sonuga-Barke4,5; Heike Eichele1,3, Heidi van Wageningen1,
Daniel Wollschlaeger6, & Kerstin J. Plessen3,7,8
1
Department of Biological and Medical Psychology, University of Bergen, Bergen,
Norway
2
Department of Child and Adolescent Psychiatry, Haukeland University Hospital,
Bergen, Norway
3
K. G. Jebsen Center for Research on Neuropsychiatric Disorders, Bergen, Norway
4
Department of Experimental Clinical and Health Psychology, Gent University, Gent,
Belgium
5
Department of Psychology, University of Southampton, Southampton, UK
6
Institute for Medical Statistics, Epidemiology and Informatics, University Medical
Center, Johannes-Gutenberg-University Mainz, Mainz, Germany
7
Child and Adolescent Centre of Mental Health, Capital Region, Copenhagen,
Denmark
8
Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark
Word count: Abstract: 225
1
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
2
Corresponding author: Dr. Lin Sørensen, Department of Biological and Medical
Psychology, University of Bergen, Jonas Liesvei 91, 5009 Bergen, Norway. Phone: +47
555896207. E-mail: [email protected].
Acknowledgements:
We thank Anne Øfsthus and Randi Hopsdal for their technical assistance, and Steinunn
Adolfsdottir for comments on the manuscript. We also thank the children and parents who
participated in this study. This work was supported by a post-doc grant to L.S. from the
Western Norway Health Authority and by the Research Council of Norway (190544/H110)
and National Norwegian ADHD network grants to K.J.P.
Author interests to disclose:
Professor Sonuga-Barke declares the following competing interest during the three years prior
to September 2014: Fees for speaking, consultancy, research funding and conference support
from Shire Pharma; Speaker fees from Janssen Cilag, Medice & Qbtech; Book royalties from
OUP and Jessica Kingsley; Consultancy from Neurotech solutions.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
3
Abstract
Background: Suboptimal decision making in the face of risk (DMR) in children with ADHD
may be mediated by deficits in a number of different neuropsychological processes. We
investigated DMR in children with ADHD using the Cambridge Gambling Task (CGT) to
distinguish difficulties in adjusting to changing probabilities of choice outcomes (so called
risk adjustment) from general risk proneness, and to distinguish these two processes from
delay aversion (the tendency to choose the least delayed option) and impairments in the
ability to reflect on choice options. Based on previous research we predicted that suboptimal
performance on this task in children with ADHD would be primarily due to problems with
risk adjustment and delay aversion rather than general risk proneness. Method: Drug naïve
children with ADHD (n = 36), 8 to 12 years, and an age-matched group of typically
developing children (n = 34) performed the CGT. Results: As predicted children with ADHD
were not more prone to making risky choices (i.e., risk proneness). However, they had
difficulty adjusting to changing risk levels and were more delay aversive – with these two
effects being correlated. Conclusions: Our findings add to the growing body of evidence that
children with ADHD do not favor risk taking per se when performing gambling tasks, but
rather may lack the cognitive skills or motivational style to appraise changing patterns of risk
effectively.
Keywords: ADHD; gambling task; delay aversion; risk taking, Cambridge Gambling
Task.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
4
Suboptimal decision making by children with ADHD in the face of risk: Poor risk adjustment
and delay aversion rather than general proneness to taking risks.
Attention-Deficit/Hyperactivity Disorder (ADHD) is a neurodevelopmental disorder,
affecting around 5% of children (Faraone et al., 2015). In everyday life children with ADHD
are sometimes described as risk takers and thrill seekers, when they place themselves in
potentially dangerous situations or choose alternatives where the outcomes are uncertain. This
behavioral style can persist as a problem into adolescence and adulthood and become
associated with reckless driving (Thompson, Molina, Pelham, & Gnagy, 2007), substance
uses/abuses (Rooney, Chronis-Tuscano, & Yoon, 2012), risky sexual behavior (Sarver,
McCart, Sheidow, & Letourneau, 2014) and pathological gambling (Faregh & Derevensky,
2011; Waluk, Youssef, & Dowling, in press) - all factors that reduce quality of life (Beitz,
Salthouse, & Davis, 2014). It has been suggested that this tendency reflects deficits in
decision making in the face of risk (DMR). Neuropsychologically, DMR is a complex process
that requires the integration of external information with internal value systems during choice
between response options with differences in outcome probability (Clark et al., 2008; Ochsner
& Gross, 2014). Gambling paradigms are frequently used to assess DMR (Bechara et al.,
1994; Bechara, 2007; Deakin, Aitken, Robbins, & Sahakian, 2004). Thus, impaired DMR
could be due to a range of cognitive and motivational factors (Buelow & Suhr, 2009; SonugaBarke et al., in press). To understand what particular process drives poor DMR in ADHD it is
important to distinguish these different elements. We hypothesize that the poor DMR may
stem from three different dysfunctional processes in ADHD: Risk proneness (thrill seeking),
reflection impulsivity (poor inhibitory control), or delay aversion (motivational impulsivity)
First, poor DMR could be due to the attractiveness of risk for children with ADHD
(Groen, Gaastra, Evans, & Tucha, 2013). Indeed, individuals with ADHD are often regarded
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
5
as innate thrill seekers and gamblers. However, in a recent review (Groen et al., 2013) only
half of the fourteen studies testing DMR in children with ADHD, using laboratory-based
gambling paradigms, showed them to be significantly more risk favoring than typically
developing children. The inconsistencies between studies may be explained by their use of
different gambling paradigms (Groen et al., 2013) or by different approaches to calculating
risk taking tendencies (Buelow & Suhr, 2009). In particular, a distinction can be drawn
between “explicit” and “implicit” gambling paradigms. This relates to whether the acting
outcome probabilities are described to participants or not. For instance, the probability of
wins or losses under different trail conditions is learnt on a trial and error basis during the
Iowa Gambling Task (IGT): Children are unaware of the probabilities of the contingencies
when they start performing at the beginning of the task (see Brand, Recknor, Grabenhorst, &
Bechara, 2007 for DMR under ambiguity). The IGT findings to date show that it is only in the
second half of this task (when probabilities are more likely to have been learnt, but also when
their persistence on the task is tested), that children with ADHD display abnormal
performance (Garon, Moore, & Waschbusch, 2006; Geurts, Oord, & Crone, 2006; Hobson,
Scott, & Rubia, 2011; Toplak, Jain, & Tannock, 2005). This highlights the importance of
making information about choice probabilities explicit from the start of the task if one is
interested in risk taking per se rather than contingency learning.
However, given the complexity of many risk/gambling situations, children with
ADHD may struggle to effectively adjust their choices to take account of information about
the probabilities of different outcomes – even when that information is explicit. In this case,
individuals with ADHD would not be risk takers per se, but rather generally show impaired
decision making on complex tasks.
Second, they may therefore make less rational decisions because of an inability to
reflect sufficiently carefully on possible outcomes - making rapid choices between different
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
6
options. This may be due to inhibitory problems (Barkley, 1997) that limit the time available
to evaluate outcome options effectively. Both children (Solanto et al., 2007) and adults
(Young, Morris, Toone, & Tyson, 2007) with ADHD make decisions more quickly than
controls and these effects persist, even when the level of task difficulty is increased (Young et
al., 2007), on the Tower of London Task that predominantly loads on inhibitory control
(Miyake et al., 2000).
Third, suboptimal DMR may alternatively stem from delay aversion (i.e., the
motivation to escape/avoid delay; Sonuga-Barke, 1994; 2003; 2005) with their choices being
dominated by the immediacy of the outcome rather than its probability. There is now
considerable evidence that children with ADHD prefer smaller, immediate rewards (or
gratifications) over larger delayed rewards (Marco et al., 2009; Paloyelis, Asherson, Mehta,
Faraone, & Kuntsi, 2010; Solanto et al., 2001). A negative emotional state (unpleasant delayinduced feelings that create an inner tension) mediated by hyperactivation of the brain’s
punishment centers (i.e., in the amygdala and insula) has been suggested to mediate the
aversiveness of delay in ADHD individuals (Sonuga-Barke, 2003; 2005). Indeed signals of
upcoming imposed delay elicit increased hemodynamic responses in these brain regions in
patients with ADHD (Lemiere et al., 2012; Wilbertz et al., 2013).
The Cambridge Gambling Task (CGT) is a standardized, well-characterized approach
to assessing decision making under uncertainty (Clark et al., 2008; Deakin et al., 2004;
Rogers et al., 1999) in which probabilities of different outcomes are presented explicitly.
Crucially for the current analysis, it was designed in a way that makes it possible to
disentangle the different components listed above that may account for suboptimal DMR in
ADHD by including these components as process scores. First, risk adjustment is defined as
the ability to modify choices in the light of information about the probability of different
outcomes and hence track the optimal outcome on each trial (Figure 1). Second, risk
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
7
proneness is the total number of points that are gambled on the most improbable outcome reflecting the overall tendency to take risks. Third, reflection time is the time taken to think
about the options during the decision phase and is calculated on the basis of the deliberation
time between stimulus presentation and choice outcome. Finally, delay aversion reflects the
proportion of choices made for the option available most immediately irrespective of the
outcome. In the CGT both reflection time and delay aversion are regarded as linked to the
concept of impulsivity - however, delay aversion is conceptualized as a motivational driver of
impulsivity (delayed rewards: Deakin et al., 2004; Clark et al., 2008), while reflection time is
more cognitive in nature perhaps linked to inhibitory control (Deakin et al., 2004).
Only two studies (Coghill, Seth, & Matthews, 2014; DeVito et al., 2008) have used the
CGT to study DMR in boys with ADHD. Coghill et al. (2014) found that the risk adjustment
score loaded on an overall decision making factor, which was impaired in the ADHD group.
Further boys with ADHD displayed greater CGT delay aversion. DeVito et al. (2008)
likewise reported poorer risk adjustment and greater delay aversion in a drug-stabilized
ADHD group.
The aim of the current study was thus to test the relationship between these four
putative processes and suboptimal DMR in drug-naïve children with ADHD. Our hypothesis
was that children with ADHD would be less able to effectively adjust their response to risky
options in the light of information about outcome probabilities (risk adjustment) and, building
on the results of previous studies, we predicted that this would be primarily linked to delay
aversion. We thus expected that children with ADHD would score lower on risk adjustment
and higher on delay aversion. Further, we expected that delay aversion would predict the
outcome of risk adjustment when taking into account risk proneness and reflection time. We
also expected that the effect of ADHD on delay aversion would be neuropsychological
specific - in that the effect on delay aversion would not be explained by inhibitory control or
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
8
working memory deficits. Furthermore, they would be clinically significant and not explained
by the presence of comorbid disorders. Delay aversion is a motivational deficit that is
suggested to be different from cognitive impulsivity (i.e., inhibitory control) (Deakin et al.,
2004; Marco et al., 2009: Sonuga-Barke, 2003). This assumption also applies to the relation
of delay aversion and working memory, however, a recent study (Patros et al., 2015)
identified problems with working memory as a possible mediator of choice-impulse – delay
aversion – in ADHD. It is also possible that comorbid disorders characterized by motivational
symptoms, such as oppositional defiant disorder (ODD) or anxiety, may explain the effect of
delay aversion in ADHD. Previous studies have shown these symptom constellations to affect
DMR in ADHD (Garon et al., 2006; Hobson et al., 2011).
The current study extends the two previous studies in a number of ways. First, it is the
first study to examine the inter-relationship between the constituent elements of impaired
DMR in ADHD – including the process scores of risk proneness, reflection time, and delay
aversion and the DMR outcome score of risk adjustment. Second, we tested a mixed gender
sample of drug-naïve patients whereas the two previous studies only included boys with
ADHD either drug-naïve (Coghill et al., 2014) or drug-stabilized (DeVito et al., 2008). Third,
we were able to examine how different neuropsychological processes such as working
memory and inhibitory control measures on other tasks were related to those outcomes.
Coghill et al., (2014) included a broad set of tasks whereas DeVito et al., (2008) did not. This
is important because Coghill et al. (2014) showed that CGT delay aversion also loaded on
inhibitory control, whereas delay aversion measured with the choice delay task has previously
been linked to working memory problems in ADHD (Patros et al., 2015). Fourth, we
investigated the influence of comorbid disorders in ADHD on the CGT scores. Based on the
findings from Coghill et al. (2014), where the presence of comorbid disorders in the ADHD
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
9
group did not have an effect on the CGT scores, we expected case control differences on the
CGT scores would be specific to ADHD and not driven by comorbidity.
Method
Participants
Two groups of children (n = 70) performed the CGT: 36 were drug-naïve children
with an ADHD diagnosis and 34 were typically developing children. The age of the children
ranged from 8 to 12 years of age (mean age = 10.10; SD = 1.18), and the two groups did not
differ significantly in age or gender distribution (62% boys in the total sample) (Table 1). The
local Regional Ethic Committee approved the study, and parents of the children gave
informed consent. Exclusion criteria were use of psychostimulants, a Full-Scale Intelligent
Quotient (FSIQ) < 80, prior head injury, suspicion of an autism spectrum disorder, prior
epileptic attacks, or being born before 36 weeks gestation. Patients without a prior ADHD
diagnosis but who had symptoms of ADHD were referred to the child and adolescent
psychiatric outpatient clinics of the municipality of Bergen and seen by members of the team.
Typically developing children were recruited from schools in regions with a geographical
overlap with the referral clinics. The children were carefully diagnosed using the “Schedule
for Affective Disorders and Schizophrenia for School-Age Children – Present and Lifetime
Version” (K-SADS) (Kaufman et al, 1997) according to the Diagnostic and Statistical Manual
of Mental Disorder – Fourth Edition criteria (Ambrosini, 2000; Americal Psychiatric
Association, 1994).
Measures
The Cambridge Gambling Task (CGT; Cambridge Neuropsychological Test
Automated Battery, CANTAB; www.camcog.com; Rogers et al., 1999; Clark et al., 2008).
The participants were instructed that a yellow token was hidden behind either a blue or red
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
10
box, presented in an array of 10 boxes at the top of a computer screen (Figure 1a)).
Furthermore, they were instructed to guess which colored box the token was behind and to
indicate their choice by pushing either a red or a blue square at the bottom of the screen.
Thereafter, they were instructed to make a bet on the likelihood of their decision being correct
or not. The possible bets were displayed on the right-hand side of the screen in sequence, and
the instruction was to touch the image of the box on the screen to place a bet. Points were
presented in 5-second increments/decrements. In four of the test blocks the bets were
presented in an ascending order of 5, 25, 50, 75, and 95% portion of the points that the
children would “earn” on each trial (displayed on the left hand side of the computer screen),
and in the other four in a descending order (95, 75, 50, 25, and 5% portion of the points). The
order of presentation (ascending or descending) was counterbalanced across individuals
within the groups.
The task was performed on a desktop PC, and the responses were recorded via a
touch-sensitive screen. The CGT comprises 4 practice trials and 8 test blocks (including 9
trials) with the red:blue box ration varying from 1:9 to 9:1, in a pseudorandom order, and with
the total points re-set to 100 points at the start of each block. In the current study, the CGT
generated scores for the four key CGT variables - risk adjustment, risk proneness, reflection
time, and delay aversion were calculated (Figure 1b)). Risk adjustment was designated the
outcome and risk proneness, reflection time, and delay aversion as the process variables.
These scores were centralized as z-scores (see Table 1 for the CGT raw scores). We also
included the test duration time as a variable.
_____________________________________
Please insert Figure 1a) and b) about here
_____________________________________
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
11
Working memory tasks. The letter-number sequencing task from the WISC-IV
(Wechsler, 2003) and the spatial span task from the WISC-IV integrated battery (Kaplan,
Fein, Kramer, Morris, Delis, & Maerlender, 2004) were used to assess verbal and visuospatial working memory (Baddely, 1986), respectively. The spatial span task consists of two
conditions, one forward and one backward. In the forward condition the children have to
recall and touch blocks in the same order as that presented by the examiner. In the backward
condition of the spatial span the children have to recall and touch blocks in the opposite order
than presented by the examiner. In performing the letter-number sequencing task, the children
have to recall, rearrange, and reproduce a sequence of letters and numbers presented aloud by
the examiner. We included the score generated from the performance on the backward
condition in the spatial span task as a measure of visuo-spatial working memory, and the total
score from the letter-number sequencing task as a measure of verbal working memory.
Inhibitory control task. The stop signal task from the CANTAB (www.camcog.com;
Chamberlain et al., 2011) was included to measure problems with inhibitory control (Aron,
2003). The stop signal task is a computer-based task where the children decide if a presented
arrow on the screen points to the right- or left-hand side, indicated by pressing either a right or
left button on a response box (i.e., go trials). In 25% of the trials a sound is presented which
signals that they shall inhibit the response on which direction the presented arrows points to.
The stop signal reaction time (SSRT) is the score that measures inhibitory control. The SSRT
is the time interval between the mean reaction time on go trials and the stop-signal delay (the
time interval from the stimuli –arrow- is presented on the screen until the stop signal is
presented). The stop signal delay dynamically adjusts to each child’s mean reaction time on
go trials and this adjustment secures that all the children succeed on at least 50% of the trials
when the stop signal is presented.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
12
Intellectual Function— Full-Scale IQ (IQ).
IQ was assessed using the Wechsler Intelligence Scale for Children – Fourth Edition
(WISC-IV; Wechsler, 2003). The mean IQ in the sample was 99.11 (SD = 11-19; range: 82134).
Diagnostic assessment.
The diagnostic assessment of all the children participating in the study followed the
algorithm of the K-SADS (Kaufman et al., 1997). This is a reliable semi-structured interview
evaluating current and past episodes of psychopathology in children according to the
Diagnostic and Statistical Manual of Mental Disorder-Fourth Edition criteria (Ambrosini,
2000; American Psychiatric Association, 1994; Kaufman et al., 1997). Clinical professionals
interviewed the children and their parents with the K-SADS interview, and a child psychiatrist
and a clinical psychologist confirmed the diagnostic evaluation. Only children with a primary
diagnosis of ADHD were included (n = 36).
Procedure
The children came with their parents to the testing and evaluation session. One researcher
followed the family throughout the assessment. They gave information to the children and
their parents about the procedure of the assessment. The children and parents then gave
written consent to participate in the study. The children were then administered the cognitive
assessments, administered by trained test assistants, while the parents were interviewed with
the K-SADS. After cognitive testing, children were interviewed with the K-SADS separately.
All children received the same battery of tests in a fixed order, which took approximately 3
hours to complete, They were allowed breaks as needed with one break set in the middle of
the testing. The CGT was administered as the last test of the battery, because piloting of the
test battery revealed that children seemed to like to perform this test. The ethical approval of
the study allowed for the reimbursement of £100 to the parents and their children -
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
13
compensating the parents for the time off from work and travel expenses. The children also
received small rewards (t-shirt and a little toy) after they finished the cognitive testing.
Statistical analyses
Statistical analyses were computed with IBM SPSS, version 23 and run in 4 stages. We first
tested the hypothesis that children with ADHD would score lower on risk adjustment and
higher on delay aversion. We then tested the hypothesis that delay aversion would predict risk
adjustment accounting for risk proneness and reflection time. Finally, we tested the
hypothesis that the ADHD effect on delay aversion on DMR would not be explained by
problems with inhibitory control or working memory. The step-wise approach was
implemented with the following procedure: In stage A, 4 univariate Analyses of Co-variance
(ANCOVAs) were run to test for between-group effects on the CGT scores of risk
adjustment, risk proneness, reflection time, and delay aversion. All 4 models were adjusted
for age and gender. In stage B, to explore whether the group differences found in stage A
were due to intelligence or other neuropsychological deficits, IQ, working memory (letternumber sequencing and backward spatial span), and inhibitory control (SSRT) were added as
further covariates into each of the 4 ANCOVA models (Mayers, 2013; Tabachnick & Fidell,
2007). In stage C, a linear hierarchical regression was run to test which of the CGT process
scores (risk proneness, reflection time, and delay aversion) predicted the outcome score of
risk adjustment while always adjusting for the covariates of age and gender. In stage D, in
order to explore whether the predictive effect of the process variables on the risk adjustment
were due to intelligence or other neuropsychological deficits, the linear hierarchical
regression analysis of stage C was repeated while additionally adjusting for IQ, working
memory (letter-number sequencing and backward spatial span), and inhibitory control (SSRT)
as covariates. Our hypothesis was that the delay aversion score would predict risk adjustment.
In stage C and D, we therefore entered risk proneness and deliberation time into the first step
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
14
of the hierarchical model together with the remaining covariates (stage C: Age and gender;
stage D: Age, gender, IQ, the working memory scores and inhibitory control score), and the
delay aversion score in the second step. In supplementary analyses, we ran 4 ANCOVAs to
investigate the effect of comorbid disorders in the ADHD group on the CGT scores (risk
adjustment, risk proneness, reflection time, and delay aversion) while always adjusting for
age and gender. First, we investigated the effect of externalizing comorbid disorders
(ODD/CD) on the CGT scores, and secondly, the effect of internalizing comorbid disorders
(anxiety). Further, also as a supplementary analysis, we tested if the prediction of delay
aversion on risk adjustment by delay aversion was specific to those with an ADHD diagnosis
by adding an interaction variable between ADHD and delay aversion in the second step of the
regression model described under stage C (main statistical analyses). A significant effect of
the interaction variable would suggest that the effect of delay aversion was specific to those
with an ADHD diagnosis (Aiken & West, 1991).
We adjusted for multiple analyses by using Bonferroni correction of alpha level () in
the between-group analyses (ANCOVAs); p = .05/4, which gives a α corrected p level of
.013. Outliers were defined using a plus/minus 3 standard deviation threshold from the sample
mean and were replaced with a score of plus/minus 2 standard deviation from the sample
mean (Field, 2005). Two outliers were detected, one child with ADHD showed a very long
reflection time (z = 4.12) and one typically developing child showed a very high score in risk
adjustment (z = 3.97).
Results
In the group of children with ADHD, 23 children fulfilled the diagnostic criteria for
the combined subtype, 11 had the predominantly inattentive subtype, and 2 the
hyperactive/impulsive subtype (Leckman, Sholomskas, Thompson, Belanger, & Wessman,
1982). Further, 23 of the 36 children with ADHD fulfilled the criteria for at least one
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
15
comorbid disorder. The comorbid disorders present were: Oppositional defiant disorder: n =
17; conduct disorder: n = 3; phobia disorders: n = 9; and general anxiety disorder: n = 2. One
child in the control group fulfilled the diagnostic criteria for specific phobia. On the cognitive
measures, the children with ADHD had significantly lower IQ and working memory (letternumber sequencing and spatial span) scores than the group of typically developing children,
whereas no such significant difference appeared in relation to inhibitory control (SSRT)
(Table 1).
__________________________________
Please insert Table 1 about here
__________________________________
CGT delay aversion and risk proneness scores correlated significantly with age younger children were more delay averse and less risk prone than the older ones
(Supplemental Table 1). Moreover, boys were more risk prone than girls. Delay aversion
correlated with lower IQ, whereas high IQ was associated with better risk adjustment.
Reflection time was not significantly associated with age, gender, or IQ. Lower risk
adjustment correlated significantly only with delay aversion and delay aversion correlated
significantly with lower risk proneness. Reflection time did not correlate with other measures.
Delay aversion correlated with a lower test duration time (r = -.84, p < .001), whereas greater
risk proneness (r = .36, p < .01), longer reflection time (r = .32, p < .01) and better risk
adjustment (r = .32, p < .01) correlated with longer test duration.
In stage A, we tested our hypothesis that children with ADHD would score lower on
risk adjustment and higher on delay aversion. This hypothesis was confirmed in that children
with ADHD displayed poorer risk adjustment (F(1,66)= 11.38 , p = .001, ηp2 = .15) and
more delay aversion (F(1,66)= 8.24, p = .006, ηp2 = .11) (Figure 2). The groups did not
differ with regard to reflection time or risk proneness. Age covaried with delay aversion
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
16
(F(1,66)= 10.69 , p = .002, ηp2 = .14) and risk proneness (F(1,66)= 4.82 , p = .03, ηp2 = .07),
whereas adding gender as a covariate did not affect the results. In stage B, we investigated the
hypothesis that the ADHD effect on risk adjustment and delay aversion would not be better
explained by other cognitive factors. IQ, working memory scores (spatial span backward and
letter-number sequencing), and inhibitory control score (SSRT) did not covary with the CGT
scores and were therefore excluded from the model.
____________________________________
Please insert Figure 2 about here
___________________________________
In the linear hierarchical regression analysis of stage C, we tested the hypothesis that
delay aversion would predict risk adjustment when taking into account risk proneness and
reflection time. This hypothesis was confirmed in that entering delay aversion in the second
step of the model explained 19% of additional variance of the risk adjustment score (Table 2)
compared with the first step including risk proneness and reflection time, age and gender as
predictors. Delay aversion and risk proneness significantly predicted risk adjustment in the
second step of the model. In stage D, we checked if the effect of delay aversion on risk
adjustment as an outcome could not be better explained by other cognitive factors. As such,
adding IQ, working memory (letter-number sequencing and backward spatial span), and
inhibitory control (SSRT) as covariates to the model did not contribute significantly to
explaining the variance in risk adjustment and were excluded from the model.
We additionally tested the influence of comorbid disorders on the effects of ADHD on
risk adjustment and delay aversion. ANCOVAs revealed no significant differences on the
CGT scores between ADHD with comorbid externalizing disorders (ODD/CD; n = 17) and
without such comorbidity (n = 19), or those with comorbid internalizing disorders (anxiety; n
= 11) and those without such comorbidity (n = 25) (Supplemental Table 2). Further, we also
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
17
tested whether the relationship between delay aversion and risk adjustment was specific to
ADHD by adding a term of the delay aversion and ADHD interaction in the regression model
described in stage C. This term was not significant indicating that the effect of delay aversion
on risk adjustment was not limited to the ADHD group (Supplemental Table 3).
___________________________________
Please insert Table 2 about here
__________________________________
Discussion
The aim of the current study was to use the CGT to tease apart the distinctive processes
underpinning suboptimal DMR in children with ADHD. Our findings suggest that the
children with ADHD were not unusually risk prone when performing the CGT, however,
consistent with previous results, children with ADHD had problems adjusting their choices to
the changing information available about the probabilities of winning. The results suggest that
these problems with risk adjustment were specifically linked to delay aversion. Further,
poorer risk adjustment and delay adverse behavior seemed to be specifically linked to the
ADHD diagnosis and not to comorbid disorders in this group. This is the first study showing
an inter-relationship between factors in DMR in children with ADHD, and that aversiveness
towards delay seems to be the most substantial contribution to suboptimal DMR (as indexed
by risk adjustment) in these children (Figure 3).
First, there was no evidence that overall, children with ADHD were risk prone – or
favored the risky option in general for its own sake. Contrary to what is often assumed and
presented in the literature these findings imply that the association between ADHD and risk
prone behaviors may not be caused by an intrinsic valuation of risk proneness in itself. The
ability of the CGT to distinguish between different elements of the DMR process provide
crucial evidence to move the field away from the assumption that ADHD children are
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
18
inherent risk takers–with three studies, including the current one, refuting the notion (Coghill
et al., 2014; DeVito et al., 2008).
Second, although not risk seekers, children with ADHD did have difficulties adapting
their choices to the changing patterns of outcome probabilities programmed during the task –
a finding also consistent with the two previous CGT studies in ADHD (Coghill et al., 2014;
DeVito et al., 2008). We found consistent evidence that this was reflected by children with
ADHD to show delay aversion as defined in the CGT: They tended to choose the least
delayed option of the two available independent of the level of associated risk. This finding
was predicted and in line with the two previous studies (Coghill et al., 2014; DeVito et al.,
2008). However, we found that delay aversion predominantly predicted the risk adjustment
score compared to being risk prone or using less reflection time, which is a novel finding,
since the two previous studies in ADHD (Coghill et al., 2014; DeVito et al., 2008) did not
study the relationship between delay aversion and risk adjustment. This may suggest that the
failure to adjust choice to optimize outcomes was linked to a different motivational style
rather than to a cognitive deficit. The finding of increased delay aversion is consistent with a
large body of literature showing that decision making processing in children with ADHD is
affected by a preference for immediate reward to escape delay (Marco et al., 2009; Paloyelis
et al., 2010; Solanto et al., 2001). Similar to Coghill et al. (2014) we found that this increased
delay aversion and the problems with adjusting risks was specific to the ADHD diagnosis and
not to the presence of comorbid disorders in this group of children.
The concept of delay aversion, however, has been operationalized somewhat
differently in different studies. In recent conceptualizations delay aversion is distinguished
from other possible causes such as inhibitory problems in a number of different ways – in
particular by being explicitly linked to the escape of delay rather than an impulsive drive for
immediate reward (Sonuga-Barke, 2003). Children with ADHD tend to choose immediate
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
19
rewards more consistently when this choice gives escape from delay compared to when the
same choice does not follow by an escape of delay (Marco et al., 2009; Sonuga-Barke et al.,
1992). Delay aversion is therefore hypothesized to be a dynamic trait in children with ADHD
that is highly context dependent (Sonuga-Barke, 2005). In contrast, problems with inhibitory
control comprise impulsivity that may be associated with a preference for immediate reward
(Barkley, Edwards, Laneri, Fletcher, & Metevia, 2001) independent of contextual factors (i.e.,
the possibility to escape delay; Barkley, 1997). It is important to note that the CGT delay
aversion score does not distinguish the drive for immediate reward from the need to escape
delay (Deakin et al., 2004; Rogers et al., 1999), and appears to load both on the factors of
delay aversion and on inhibitory control (Coghill et al., 2014). We thus confirmed that delay
aversion was associated with a shorter duration of test time, and this association showed a
large effect, which means that a predominant tendency to choose to bet on the immediate
points led to an earlier “escape” from this specific test setting. Further, the children’s degree
of inhibitory control did not associate with delay aversion in the between-group analysis. Also
the children with ADHD did not use less reflection time before making their choices – a
pattern linked to inhibitory problems, such as previously shown on the decision making task
of Tower of London (Solanto et al., 2007; Young et al., 2007). Moreover, other aspects of
cognitive functioning such as working memory and the children’s level of IQ also did not
associate with delay aversion in the between-group analyses. These findings strengthen the
claim that the delay aversion measure as operationalized in the CGT was the main contributor
to the poorer risk adjustment in the children with ADHD in the current study and that it is not
related to a more general tendency to reflect on problems (Deakin et al., 2004; DeVito et al.,
2008), working memory problems, or to a lower IQ. We have suggested a model of DMR
(Figure 3) that links our finding of delay aversion to cause problems with risk adjustment in
children with ADHD, and not risk proneness a previously suggested (Groen et al., 2013). Our
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
20
findings suggest, however, that despite children with ADHD being significantly more delay
aversive than typically developing children, the effect of delay aversion on the risk adjustment
outcome is not specific to ADHD. This means that delay aversion may also explain poorer
risk adjustment – i.e., DMR – in typically developing children
Third, the problems with adjusting risk taking in ADHD may relate to other factors
than the three suggested in our study: Delay aversion, risk proneness, and reflection
impulsivity. Problems with DMR in ADHD may reflect a general inattention to changes in the
environment that leads to these being ignored (Barkley, 1997). It could also be linked to a
more general insensitivity to changes in reinforcement schedules that have been previously
observed in ADHD (Kollins, Newland, & Critchfield, 1997). And lastly, it could be due to
problems comparing multiple outcomes of different value at the same time (Sonuga-Barke &
Fairchild, 2012). Future studies of DMR in ADHD could test the model depicted in Figure 3
against other possible causal factors of DMR”.
___________________________________
Please insert Figure 3 about here
___________________________________
Conclusion
The children with ADHD’s performance could be differentiated from that of typically
developing children in terms of their ability to adjust responses to optimize outcomes in the
face of changing probabilities and their level of delay aversion: With these two effects being
correlated. Children with ADHD were, however, not risk takers in a general sense. Thus,
motivational factors such as delay aversion may be important in understanding why
individuals with ADHD have impaired DMR.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
21
References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions.
Newbury Park: Sage.
Ambrosini, P.J. (2000). Historical development and present status of the schedule for
affective disorders and schizophrenia for school-age children (K-SADS). Journal of
American Academy of Child and Adolescent Psychiatry, 39, 49-58.
doi:10.1097/00004583-200001000-00016
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders (4th ed.). Washington, DC: Author.
Aron, A. R., Dowson, J. H., Sahakian, B. J., & Robbins, T. W. (2003). Methylphenidate
improves response inhibition in adults with attention-deficit/hyperactivity disorder.
Biological Psychiatry, 54(12), 1465-1468. doi:10.1016/S0006-3223(03)00609-7
Baddeley, A. D. (1986). Working memory. Oxford: Clarendon Press.
Barkley, R. A. (1997). Behavioral inhibition, sustained attention, and executive functions:
constructing a unifying theory of ADHD. Psychological Bulletin, 121(1), 65-94.
doi:10.1037/0033-2909.121.1.65
Barkley, R.A., Edwards, G., Laneri, M., Fletcher, K., & Metevia, L. (2001). Executive,
discounting, and sense of time in adolescents with attention deficit hyperactivity
disorder (ADHD) and oppositional defiant disorder (ODD). Journal of Abnormal
Child Psychology, 29, 541-556. doi:10.1023/A:1012233310098
Bechara, A. (2007). Iowa gambling task professional manual. Lutz: Psychological Asessment
Resources.
Bechara, A., Damasio, A.R., Damasio, H., & Anderson, S.W. (1994). Insensitivity to future
consequences following damage to human prefrontal cortex. Cognition, 50(1-3), 7-15.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
22
doi:10.1016/0010-0277(94)90018-3
Beitz, K.M., Salthouse, T.A., & Davis, H.P. (2014). Performance on the Iowa Gambling
Task: From 5 to 82 years of age. Journal of Experimental Psychology: General,
143(4), 1677-1689. doi:10.1037/a0035823
Brand, M., Recknor, E.C., Grabenhorst, F., & Bechara, A. (2007). Decisions under ambiguity
and decisions under risk: correlations with executive functions and comparisons of
two different gambling tasks with implicit and explicit rules. Journal of Clinical and
Experimental Neuropsycholgy, 29, 86-99. doi:10.1080/13803390500507196
Buelow, M.T. & Suhr, J.A. (2009). Construct validity of the iowa gambling task.
Neuropscyhology Review, 19, 102-114. doi:10.1007/s11065-009-9083-4
Chamberlain, S. R., Robbins, T. W., Winder-Rhodes, S., Müller, U., Sahakian, B. J., Blackwell, A. D., &
Barnett, J. H. (2011). Translational approaches to frontostriatal dysfunction in attentiondeficit/hyperactivity disorder using a computerized neuropsychological battery. Biological
Psychiatry, 69(12), 1192-1203. doi:10.1016/j.biopsych.2010.08.019
Clark, L., Bechara, A., Damasio, H., Aitken, M.R.F., Sahakian, B.J., & Robbins, T.W. (2008).
Differential effects of insular and ventromedial prefrontal cortex on risky decisionmaking. Brain, 131, 1311-1322. doi:10.1093/brain/awn066
Coghill, D.R., Seth, S., & Matthews, K. (2014). A comprehensive assessment of memory,
delay aversion, timing, inhibition, decision making and variability in attention deficit
hyperactivity disorder: advancing beyond the three-pathway models. Psychological
Medicine, 44(9), 1989-2001. doi:10.1017/S0033291713002547
Deakin, J., Aitken, M., Robbins, T., & Sahakian, B.J. (2004). Risk taking during decision
making in normal vonlunteers changes with age. Journal of International
Neuropsychological Society, 10, 590-598. doi:10.1017/S1355617704104
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
23
doi:10.1017/S1355617709090481
DeVito, E.E., Blackwell, A.D., Kent, L., Ersche, K.D., Clark, L., Salmond, C.H., …
Sahakian, B.J. (2008). The effects of methylphenidate on decision making in
attention-deficit/hyperactivity disorder. Biological Psychiatry, 64, 636-639. doi:
10.1016/j.biopsych.2008.04.017
Faraone, S.V., Asherson, P., Banaschewski, T.,Biederman J, Buitelaar, J.K., Ramos-Quiroga,
J.A.,… Franke, B. (2015). Attention-Deficit/Hyperactivity Disorder. Nature Reviews:
Disease Primers. 15020
Faregh, N. & Derevensky, J. (2011). Gambling behavior among adolescents with attentiondeficit/hyperactivity disorder. Journal of Gambling Studies, 27(2), 243-256.
doi:10.1007/s10899-010-9211-3.
Field, A. P. (2005). Discovering statistics using SPSS (2nd ed.). London ; Thousand Oaks,
Calif.: Sage Publications.
Garon, N., Moore, C., & Waschbusch, D.A. (2006). Decision making in children with ADHD
only, ADHD-anxious/depressed, and control children using a child version of the iowa
gambling task. Journal of Attention Disorders, 9(4), 607-619.
doi:10.1177/1087054705284501
Geurts, H.M., van der Oord, S., & Crone, E.A. (2006). Hot and cool aspects of cognitive
control in children with ADHD: decision-making and inhibition. Journal of Abnormal
Child Psychology, 34, 813-824. doi:10.1007/s10802-006-9059-2
Groen, Y., Gaastra, G.F., Evans, B.L., & Tucha, O. (2013). Risky behavior in gambling tasks
in individuals with adhd - a systematic literature review. PLOS ONE, 8(9).
doi:10.1371/journal.pone.0074909
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
24
Hobson, W.H., Scott, S., & Rubia, K. (2011). Investigation of cool and hot executive function
in ODD/CD independently of ADHD. Journal of Child Psychology and Psychiatry,
52(10), 1035-1043. doi:10.1111/j.1469-7610.2011.02454.x
Kaplan, E.; Fein, D.; Kramer, J.; Morris, RD., Delis, D., & Maerlender (2004). Wechsler
intelligence scale for children, fourth edition – integrated. San Antonio, TX: Harcourt
Assessment, Inc.
Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P., … Ryan, N. (1997).
Schedule for affective disorders and schizophrenia for school-age children-present
and lifetime version (K-SADS-PL): initial reliability and validity data. Journal of
American Academy of Child and Adolescent Psychiatry, 36(7), 980-988.
doi.org/10.1097/00004583-199707000-00021
Kollins, S.H., Newland, M.C., & Critchfield, T.S. (1997). Human sensitivity to reinforcement
in operant choice: How much do consequences matter? Psychonomic Bulleting and
Review, 4, 208-220. doi:10.3758/BF03209395
Leckman, J.F., Sholomskas, D., Thompson, D., Belanger, A., & Weissman, M.M. (1982).
Best estimate of lifetime psychiatric diagnosis a methodological study. Archives of
General Psychiatry, 39(8), 879-883. doi:10.1001/archpsyc.1982.04290080001001
Lemiere, J., Danckaerts, M., Van Hencke, W., Mehta, M.A., Peeters, R., Sunaert, S., &
Sonuga-Barke, E. (2012). Brain activation to cues predicting inescapable delay in
adolescent attention deficit/hyperactivity disorder: an fMRI pilot study. Brain
Research, 1450, 57-66. doi:10.1016/j.brainres.2012.02.027
Marco, R., Miranda, A., Schlotz, W., Melia, A., Mulligan, A., Muller, U., … Sonuga-Barke,
E.J. (2009). Delay and reward choice in ADHD: An experimental test of the role of
delay aversion. Neuropsychology, 23, 367-380. doi:10.1037/a0014914
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
25
Mayers, A. (2013). Introduction to statistics and spss in psychology. Harlow: Pearson
Education.
Miyake, A., Friedman, N.P., Emerson, M.J., Witzki, A.H., Howerter, A., & Wager, T.D.
(2000). The unity and diversity of executive functions and their contributions to
complex “frontal lobes” tasks: A latent variable analysis. Cognitive Psychology, 41,
49-100. doi:10.1006/cogp.1999.0734
Ochsner, K.N. & Gross, J.J. (2014). The neural bases of emotion and emotion regulation: A
valuation perspective. In J.J. Gross (Ed), Handbook of emotion regulation, 2nd edition.
New York: The Guilford Press.
Paloyelis, Y., Asherson, P., Mehta, M.A., Faraone, S.V., & Kuntsi, J. (2010). DAT1 and
COMT effects on delay discounting and trait impulsivity in male adolescents with
attention deficit/hyperactivity disorder and healthy controls.
Neuropsychopharmacology, 35, 2414-2426. doi:10.1038/npp.2010.124
Patros, C.H.G., Alderson, R.M., Lea, S.E., Tarle, S.J., Kasper, L.J., & Hudec, K.L. (2015).
Visuospatial working memory underlies choice-impulsivity in boys with attentiondeficit/hyperactivity disorder. Research in Developmental Disabilities, 38, 134-144.
doi:10.1016(j.ridd.2014.12.016
Rogers, R.D., Everitt, B.J., Baldacchino, A., Blackshaw, A.J., Swanson, R., Wynne, K., …
Robbins, T.W. (1999). Dissociable deficits in decision-making cognition of chronic
amphetamine abusers, opiate abusers, patients with a focal damage to prefrontal
cortex, and
tryptophan-depleted normal volunteers; Evidence for monoaminergic
mechanisms. Neuropsychopharmacology, 20, 322-339.
doi:10.1016/S0893-133X(98)00091-8
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
26
Rooney, M., Chronis-Tuscano, A., & Yoon, Y. (2012). Substance use in college students with
ADHD. Journal of Attention Disorders, 16, 221-234. doi:10.1177/1087054710392536
Sarver, D.E., McCart, M.R.., Sheidow, A.J., & Letourneau, E.J. (2014). ADHD and risky
sexual behavior in adolescents: Conduct problems and substance use as mediators of
risk. Journal of Child Psychology and Psychiatry, 55(12), 1345–1353.
doi: 10.1111/jcpp.12249
Solanto, M.V., Abikoff, H., Sonuga-Barke, E.J., Schachar, R., Logan, G.D., Wigal, T., …
Turkel, E. (2001). The ecological validity of delay aversion and response inhibition as
measures of impulsivity in AD/HD: A supplement to the NIMH multimodal treatment
study of AD/HD. Journal of Abnormal Child Psychology, 29, 215-228.
doi:10.1023/A:1010329714819
Solanto, M.V., Gilbert, S.N., Raj, A., Zhu, J., Pope-Boyd, S., Stepak, B., Vail, L., &
Newcorn, J.H. (2007). Neurocognitive functioning in AD/HD, predominantly
inattentive and combined subtypes. Journal of Abnormal Child Psychology, 35, 729744. doi:10.1007/s10802-007-9123-6
Sonuga-Barke, E.J., 1994. On dysfunction and function in psychological theories of childhood
disorder. Journal of Child Psychology and Psychiatry, 35, 801–815.
doi:10.1111/j.1469-7610.1994.tb02296.x
Sonuga-Barke, E.J. (2003). The dual pathway model of AD/HD: An elaboration of neuro-d
evelopmental characteristics. Neuroscience and Biobehavioral Reviews, 27, 593-604.
doi:10.1016/j.neubiorev.2003.08.005
Sonuga-Barke, E.J. (2005). Causal models of attention-deficit/hyperactivity disorder. From
common simple deficits to multiple developmental pathways. Biological Psychiatry,
57, 1231-1238. doi:10.1016/j.biopsych.2004.09.008
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
27
Sonuga-Barke E.J.S., Cortese, S., Fairchild, G., & Stringaris, A. (in press). Trans-diagnostic
neuroscience of child and adolescent mental disorders: Differentiating decisionmaking in attention-deficit/hyperactivity disorder, conduct disorder, depression and
anxiety. Journal of Child Psychology & Psychiatry,
Sonuga-Barke, E.J.S. & Fairchild, G. (2012). Neuroeconomics of attention deficit/
hyperactivity dirorder: differential influences of medial, dorsal, and ventral
prefrontal brain networks on suboptimal decision making? Biological Psychiatry, 72,
126-133. doi:10.1016/j.biopsych.2012.04.004
Sonuga-Barke, E.J.S., Taylor, E., Sembi, S., & Smith, J. (1992). Hyperactivity and delay
aversion – I. The effect of delay on choice. Journal of Child Psychology and
Psychiatry, 33(2), 387-398. doi:10.1111/j.1469-7610.1992.tb00874.x
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston:
Pearson/Allyn & Bacon.
Thompson, A.L., Molina, B.B., Peham, W.Jr., & Gnagy, E.M. (2007). Risky driving in
adolescents and young adults with childhood ADHD. Journal of Pediatric
Psychology, 32(7), 745-759. doi:10.1093/jpepsy/jsm002
Toplak, M.E., Jain, U., & Tannock, R. (2005). Executive and motivational processes in
adolescents with attention-deficit/hyperactivity disorder (ADHD). Behavioral and
Brain Functions, 1(8). doi:10.1186/1744-9081-1-8
Young, S., Morris, R., Toone, B., & Tyson, C (2007). Planning ability in adults with
attention-deficit/hyperactivity disorder. Neuropsychology, 21, 581-589.
doi:10.1037/0894-4105.21.5.581
Waluk, O.R., Youssef, G.J., & Dowling, N.A. (in press). The relationship between problem
gambling and attention deficit hyperactivity disorder. Journal of Gambling Studies.
doi:10.1007/s10899-015-9564-8
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
28
Wechsler, D. (2003). Wechsler intelligence scale for children, 4th edition (WISC-IV). San
Antonio, TX: Psychological Corporation.
Wilbertz, G., Trueg, A., Sonuga-Barke, E.J., Blechert, J., Philipsen, A:, & Tebartz van Elst, L.
(2013). Neural and psychophysiologial markers of delay aversion in attentiondeficit/hyperactivity disorder. Journal of Abnormal Psychology, 122(2), 566-572.
doi:10.1037/a0031924
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
29
a)
b)
Figure 1. a) Screen shot of the CGT and b) description of the CGT variables.
Notes. DMR = decision making in the face of risks. In b) the 10 red or blue boxes hiding the
yellow token are presented at the top of the screen. The children push either the square
indicating either red or blue at the bottom of the screen to guess which colored box the token
is hidden behind. They make a bet by touching the number presented in the box at the righthand side, numbers representing percentages of the points displayed at the left-hand side that
are presented either in ascending or descending sequences.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
30
0,60
0,40
**
**
z scores
0,20
0,00
ADHD
-0,20
Controls
-0,40
-0,60
Delay
Aversion
Risk
Proneness
Reflec on
Time
Risk
Adjustment
The CGT Scores
Figure 2. The estimated marginal means from the between-group analyses on the CGT scores
of risk proneness, reflection time, delay aversion, and risk adjustment.
Notes. The means was adjusted for the effects of age and sex in the between-group analyses.
The error bars represent the standard errors. ** = p < .013. To adjust for multiple analyses, a
Bonferronni correction of alpha level is applied; α level of 0.05 / 4 univariate analyses =
0.013.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
31
Risk Proneness
ADHD
Delay Aversion
Risk adjustment
Figure 3. A descriptive model describing the relationship between ADHD and suboptimal
decision making in the face of risk (DMR). The findings from the current study indicate that
delay aversion, and not risk proneness, is the predominant cause for children with ADHD to
engage in impulsive decisions. This was evidenced by a significant effect of ADHD diagnosis
on delay aversion and risk adjustment; effects not otherwise explained by age, gender, or
other neuropsychological deficits of IQ, working memory, or inhibitory control. Further,
delay aversion predicted the outcome of risk adjustment when taking into account risk
proneness and reflection impulsivity (not specific to children with ADHD).
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
Table 1. Descriptive information for each diagnostic subgroup.
Demographic and background information:
Variables
ADHD
Gender (n, %)#
Controls
T
value
Male
25 (69.4%)
19 (55.9%)
Female
11 (30.6%)
15 (44.1%)
Age (M, SD)
10.20 (1.33)
9.99 (1.00)
-.74
IQ (M, SD)
92.61 (6.53)
106 (11.03)
6.22 **
Letter-number seq.
12.86 (3.70)
15.85 (4.05)
3.23 **
5.89 (1.28)
7.38 (2.02)
3.72 **
259.62 (115.61)
222.66 (71.16)
Delay Aversion
0.54 (0.22)
0.42 (0.21)
2.36 *
Risk proneness
0.59 (0.12)
0.59 (0.16)
0.4
Reflection time
3317.16 (1173.55)
3351.02 (811.40)
0.14
Risk adjustment
0.61 (0.65)
1.20 (0.85)
Spatial span
SSRT
-1.60
CGT raw scores:
Notes Letter-number seq. = letter-number sequencing; SSRT = stop
signal reaction time. # Chi square analysis shows no significant
difference in distribution of gender between the groups (x2 = 1.38).
** p < .01; * p < .05.
3.27 **
32
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
33
Table 2. The prediction of the CGT process scores on the outcome of risk adjustment.
Predictors
R2
∆R2
df
∆F
p
β step 2
0.02
0.02
4/65
0.40
0.81
-.01
Model 1:
Step 1 Age
Gender
-.01
Risk proneness
-0.33 *
Reflection time
-0.11
Step 2 Delay aversion
0.22
0.19
1/64
15.68
< .001 **
-0.52 **
Notes. In the hierarchical linear regression model, R2 represents the amount of variance in the
outcome variable risk adjustment accounted for by the predictors entered into the analyses at
each step.
** p < .01; * p < .05.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
Supplemental Table 1. Correlation matrix of the bivariate relationship between the CGT
scores, age, gender, and FSIQ.
2
3
4
-.36 ** -.08
-.07
-.47 ** -.05
-.02
1 Risk adjustment
2 Delay aversion
3 Risk proneness
4 Reflection time
Note. ** p < .01; * p < .05.
FSIQ
.27 *
-.31 *
-.03
-.02
Age
Gender
.08
.03
-.34 **
.07
.28 *
-.24 *
-.05
.06
34
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
35
Supplemental Table 2. Estimated marginal means from the between-group analyses of
comorbid status in the group of children with ADHD on the CGT scores of risk proneness,
reflection time, delay aversion, and risk adjustment.
CGT scores:
ADHD with externalizing
ADHD without externalizing
com:. ODD/CD n = 17
com.: n = 19
Est. Marg. M
St. Error
Est. Marg. M
St. Error
F
P
Delay aversion
0.40
0.23
0.14
0.22
0.60
0.25
Risk taking
0.18
0.21
-0.17
0.20
1.35
0.44
Reflection time
Risk
adjustment
-0.28
0.26
0.12
0.25
1.09
0.31
-0.11
0.20
-0.57
0.19
2.52
0.12
CGT scores:
ADHD with internalizing
ADHD without internalizing
com.: anxiety n = 11
com.: n = 25
Est. Marg. M
St. Error
Est. Marg. M
St. Error
F
P
Delay aversion
-0.09
0.27
0.42
0.18
2.39
0.13
Risk taking
0.20
0.25
-0.10
0.16
0.96
0.33
Reflection time
Risk
adjustment
0.25
0.31
-0.21
0.21
1.50
0.23
-0.32
0.25
-0.37
0.16
0.03
0.87
Notes. N = 36 with ADHD; com. = comorbid disorders. Est.Marg.M = estimated marginal
means; St.Error = standard error. Age and gender were included as covariates in the
between-group analyses (ANCOVAs), and only age showed to covary with delay aversion
(F = 5.75, p < .02) in the ANCOVAs on externalizing comorbidities, and with delay
aversion (F = 5.35, p < .03) and risk proneness (F = 6.39, p < .02) in the ANCOVAs on
internalizing comorbidities. df = 1/32.
SUBOPTIMAL DECISION MAKING IN ADHD IN THE FACE OF RISK
36
Supplemental Table 3. The prediction of the CGT process scores on the outcome of risk
adjustment: Testing the specificity of ADHD on the effect of delay aversion.
Predictors
R2
∆R2
df
∆F
p
β step 2
0.02
0.02
4/65
0.40
0.81
.03
Model 1:
Step 1 Age
Gender
-.02
Risk proneness
-.32 *
Reflection time
-.13
Step 2 Delay aversion
.30
.28
3/62
8.27
< .001 **
-.26
ADHD
-.28 *
Delay aversion*ADHD
-.27
Notes. In the hierarchical linear regression model, R2 represents the amount of variance in the
outcome variable risk adjustment accounted for by the predictors entered in to the analyses at
each step. The F score represents the significance of an increment in R2. In the second step, the
main effect of delay aversion, ADHD, and of the interaction between delay aversion and ADHD were
tested.
** p < .01; * p < .05.