ADHD and delay aversion: the influence of non-temporal stimulation on choice for delayed rewards

Authors


Inge Antrop, Ghent University Hospital, Department of Child and Adolescent Psychiatry, De Pintelaan 185, B-9000 Gent, Belgium; Tel: ++32/9/240.65.25; Fax: ++32/9/264.64.89; Email: Inge.Antrop@UGent.be

Abstract

Background:  Delay aversion, the motivation to escape or avoid delay, results in preference for small immediate over large delayed rewards. Delay aversion has been proposed as one distinctive psychological process that may underlie the behavioural symptoms and cognitive deficits of attention deficit/hyperactivity disorder (ADHD). Furthermore, the delay aversion hypothesis predicts that ADHD children's preference for immediate small over large delayed rewards will be reduced when stimulation, which makes time appear to pass more quickly, is added to the delay interval. The current paper tests these predictions.

Methods:  A group of children with a diagnosis of ADHD (with or without oppositional defiant disorder (ODD)), a group with a diagnosis of high-functioning autism (HFA), and a normal control group were compared on an experimental paradigm giving repeated choices between small immediate and large delayed rewards (Maudsley Index of Delay Aversion–MIDA) under two conditions (stimulation and no stimulation).

Results:  As predicted, ADHD children displayed a stronger preference than the HFA and control children for the small immediate rewards under the no-stimulation condition. The ADHD children preferences were normalised under the stimulation condition with no differences between the groups. This pattern of results was the same whether the ADHD children had comorbid ODD or not.

Discussion:  The findings from the MIDA are consistent with the delay aversion hypothesis of ADHD in showing that preference for small immediate rewards over large delayed rewards is a specific feature of ADHD and that this preference can be reduced by the addition of stimulation. Further research is required to better understand the emotional and motivational mechanisms underpinning delay aversion.

The delay aversion hypothesis (Sonuga-Barke, 2002) challenges the notion of a cognitive deficit in executive control processes as the major underlying cause of ADHD. According to this model, children with ADHD experience a greater sensitivity to delay than their peers, which leads to decisions to choose smaller-sooner rewards over larger long-term rewards on laboratory tasks designed to measure the relationship between impulsivity and delay aversion (Kuntsi, Stevenson, Oosterlaan, & Sonuga-Barke, 2001b; Sonuga-Barke, Taylor, Sembi, & Smith, 1992; Sonuga-Barke, Williams, Hall, & Saxton, 1996). Sonuga-Barke and colleagues (Sonuga-Barke, 1994; Sonuga-Barke, De Houwer, De Ruiter, Ajzenstzen, & Holland, 2004) further describe the delay aversion as a core motivational characteristic of the disorder that could also account for the problems in the domain of attention and hyperactive behaviour that may reflect an attempt to either attend to or create so-called non-temporal stimulation in a way that will alter the subjective experience of delay. Sonuga-Barke et al. (2004) have argued that in this way attention in ADHD is related to both the controlled/strategic ‘top down’ as well as the automatic ‘bottom-up’ attentional processes. At the ‘top down’ level, children with ADHD will allocate their attention to aspects of the environment that speed up the perceived passage of time and so avoid or escape the subjective experience of delay (Sonuga-Barke, 1994). Initial evidence has indeed shown improvements in attentional performance of children with ADHD, when incorrect responses were followed by extra delay (Sonuga-Barke et al., 1996). At the ‘bottom up’ level, delay aversion may cause attentional bias to delay-related cues, which are then interpreted as emotionally or motivationally significant events. Using the dot-probe conditioning paradigm as a test of motivational influence of attention, Sonuga-Barke et al. (2004) recently provided the first evidence of qualitative differences in the attentional style of children with ADHD towards delay-related stimuli.

Alternative ‘non-executive’ theories of ADHD also focus on the stimulation-seeking behaviour of children with ADHD. They link ADHD to an inappropriate arousal level (Zentall, 1975; Zentall & Zentall, 1976) or a deficit in effort required to optimally modulate sensory arousal and motor activation (e.g., Sergeant & van der Meere, 1990; Sergeant, 2000). In most studies of stimulation seeking in ADHD, children are randomly assigned to different conditions with a well-defined amount of stimulation. To our knowledge, only one study (Leung, Leung, & Tang, 2000) examined whether children with ADHD actively seek higher levels of stimulation when performing a laboratory task (an auditory CPT). In this study the event rate was varied (to vary the stimulation level of the task), as was the opportunity to elicit visual stimulation during the auditory task (to increase non-task stimulation). Control children and children with ADHD elicited stimulation to the same extent – a finding that appears to contradict predictions based on the stimulation-seeking theories. Clearly further research is required to compare the predictions made by these different accounts of the role of stimulation-seeking behaviour in ADHD. The first major aim of the present study is therefore to test the delay aversion hypothesis further by examining the performance of children with ADHD on a traditional delay aversion task giving choices between small immediate and large delayed rewards (Kuntsi et al., 2001b) in the presence and absence of the availability of additional non-task stimulation.

The second aim is to examine diagnostic specificity of delay aversion (e.g., Kuntsi et al., 2001b; Oosterlaan, Logan, & Sergeant, 1998). Some investigators have suggested that comorbid externalising disorders, oppositional defiant disorders (ODD) and conduct disorders (CD) may account for the underlying delay aversion in ADHD. If this is so, the subgroups of ADHD children with and without ODD should differ both on their levels of choice for immediate over delayed rewards and on the impact of non-task stimulation during delay. Other investigators have suggested that ADHD and high-functioning autism (HFA) share many clinical and neuropsychological features. Although the DSM-IV-TR (APA, 2000) does not allow an additional diagnosis of ADHD in children with HFA, a behavioural overlap exists between both groups (Geurts et al., 2004). Children with HFA often show cardinal features of ADHD (e.g., Volkmar, Klin, & Cohen, 1997) and typical autism features, such as qualitative problems in social interactions and communication frequently being observed in children with ADHD (e.g., Clark, Feehan, Tinline, & Vostanis, 1999; Roeyers, Keymeulen, & Buysse, 1998). Recent studies have attempted to discriminate between ADHD and HFA with respect to their problems in executive functioning (EF). For example, Geurts et al. (2004) found that children with HFA exhibit similar but more generalised and profound EF-problems compared to children with ADHD. Also, Raymaekers and colleagues (2004) showed that adults with HFA only have problems with inhibitory control in a state of overarousal, in contrast to similar deficits in ADHD associated with underarousal. Thus, theories based on underlying deficits of EF and arousal make different predictions about these two disorders.

In summary, the present study tested the following predictions derived from the delay aversion hypothesis: It is predicted that children with ADHD will choose the small immediate option more often than controls in a no-stimulation condition (prediction 1). It is expected that when additional stimulation is available during delay this will make the delay more tolerable for children with ADHD and lead to an increase in delayed choices for larger rewards (prediction 2). It is further hypothesised that unlike measures of EF and state regulation, task performance should be worse for children with ADHD than for children with HFA, when no stimulation is available, but not when stimulation is available (prediction 3). In addition, if delay aversion is a specific feature for ADHD, no confounding effects of ODD are expected (prediction 4).

Method

Subjects

Twenty-five children with ADHD, 23 children with HFA and 25 normal controls within the age range of 6 to 14 were recruited for this study. Inclusion criteria included an estimated IQ of at least 80 measured by a short version of the Wechsler Intelligence Scale for Children, third edition (Wechsler, 1991–WISC-III) based on four subtests (Vocabulary, Similarities, Block Design, and Picture Arrangement; Grégoire, 2001). The groups did not differ for age (F(2,70) = 2.32, ns) and estimated IQ (F(2,64) = 1.79, ns). No gender differences were found across groups (χ2 (2) = 3.58, ns). Frequencies, sex distribution and mean and standard deviation for age and estimated IQ are reported in Table 1a, 1b.

Table 1a.   Mean and standard deviation for age and IQ for each group
 ADHD M (SD)HFA M (SD)Controls M (SD)Total M (SD)
Age9.85 (1.93)10.88 (2.23)9.75 (1.83)10.14 (2.03)
Estimated IQ100.79 (6.12)105.74 (12.94)105.36 (7.01)104.19 (9.41)
Table 1b.   Sex distribution for each group
GenderADHD NHFA NControls NTotal N
Male16201753
Female93820

To assess symptoms of ADHD and HFA, the Disruptive Behaviour Disorder Rating Scale (DBDRS; Pelham, Gnagy, Greenslade, & Milich, 1992; Dutch translation: Oosterlaan, Scheres, Antrop, Roeyers, & Sergeant, 2000), the Social Communication Questionnaire (SCQ; Rutter, Bailey, & Lord, 2003; Dutch translation: Warreyn, Raymaekers, & Roeyers, 2004), the Diagnostic Interview Schedule for Children for DSM-IV, parent version (PDISC-IV; Shaffer, Fisher, Lucas, Dulcan, & Schwab-stone, 2000; Dutch translation: Ferdinand, Van der Ende, & Mesman, 1998) and the Revised Autism Diagnostic Interview (ADI-R; e.g., Lord, Rutter, & Le Couteur, 1994; Lord, Storoschuk, Rutter, & Pickles, 1993) were administered by a trained psychologist.

Only children who were already diagnosed with ADHD respectively HFA were included in the research. Confirmation of the diagnoses of the clinical groups was further based on the PDISC-IV and the ADI-R for children with ADHD or HFA, respectively. Children with HFA who also met the criteria for ADHD–combined type on the PDISC-IV were excluded from the study.

Normal control children were required to be rated significantly below the age-appropriate clinical cut-off score (95th percentile) on the hyperactivity/impulsivity and inattention scales of the DBDRS by both parents and teachers and below the clinical cut-off of 15 on the SCQ by their parents. Three normal control children were, however, rated clinically for ODD.

All children were free of medication for 24 hours prior to participation in the research.

Task materials

The Maudsley Index of Delay Aversion (MIDA; Kuntsi et al., 2001b).  The MIDA (Kuntsi et al., 2001b) is based on performance on a computer task in which children have to make a choice (20 times) between a small immediate reward (1 point involving a 2-second pre-reward delay) and a large delayed reward (2 points involving a 30-second pre-reward delay) by pressing on a computer mouse. If the reward is chosen, the next trial starts immediately afterwards. The total number of trials required for the task is displayed beside the screen. Measure of delay aversion is the percentage of choices for the 2 points delayed reward. Satisfactory validity and reliability of this measure have been demonstrated by Kuntsi and colleagues (Kuntsi, Oosterlaan, & Stevenson, 2001a; Kuntsi et al., 2001b).

Extra non-temporal stimulation.  In correspondence with Leung and colleagues (2000), extra stimulation was offered by visual stimuli. The stimuli consisted of simple coloured cartoons without text, presented on a 25 by 35 cm screen placed beside the computer screen. In the condition that provided an opportunity for extra stimulation, the child could press the button to self-administer one picture as extra stimulation during the waiting period, but only when the larger delayed reward was chosen. Across these trials, the number of slides requested was taken as a measure of attention towards additional stimulation during waiting. In the condition without an opportunity for extra stimulation, a green slide was continuously presented on the screen.

Procedure

Prior to participation in the experiment, parents were informed about the aims of the study, received a full description of the experiment, and provided written consent. Children were tested individually in a neutral room, seated in front of a notebook computer on a desk. The experimenter explained the task and read the instructions out loud, and then the child practised using the mouse and choosing each of the rewards. To ensure correct understanding of the instructions, the experimenter also asked the child questions about the game.

Each child experienced two conditions within one testing session on the MIDA. Only in one condition was extra stimulation allowed. The order of the conditions was counterbalanced across subjects. At the end of the first session, the child was given a 5-minute rest during which the experimenter set up the computer for the next condition. After each experimental condition, children were asked to report on how they had felt. The session ended with a reward for participation.

Analyses

The dependent variable was the delay aversion measure: percentage of choices for the large delayed reward on the MIDA. The independent variable was Diagnostic Group (ADHD, HFA and controls). After checking whether groups differed in the amount of extra non-temporal stimulation they elicited, an univariate analysis of variance was performed for the dependent variable with one within-subject factor, Stimulation (With or Without extra Stimulation) and one between-subject factor, Diagnostic Group (ADHD, HFA, Controls). Finally the effect of ODD was analysed. Since high ratings of ODD on the DBDRS (parent and teacher version) were present in approximately half of the children with ADHD, and only in a few children of the control group (n = 1) and the HFA group (n = 4), the use of ODD as a covariate was not appropriate. Instead, the ADHD group was classified based on the presence or absence of ODD. Children were assigned to the combined ADHD+ODD group when they had been rated above the clinical cut-off score (95th percentile) on the ODD scale of the DBDRS by both parents and teacher. An additional univariate analysis of variance was performed to compare both groups. For the effects obtained by the analyses of variance, partial ETA squares were additionally calculated as a measure of effect size. In agreement with Cohen's guidelines (Cohen, 1988) inline image of .01, .10 and .25 were used as thresholds to define small, medium and large effects, respectively.

Results

Factorial analyses of variances

Groups did not differ in the amount to which they requested additional stimulation, F(2,72) < 1, ns.

As shown in Table 2, the main effect of Stimulation was significant for the delay aversion measure; i.e., percentage of choices for large delayed reward (inline image = .332). Children chose the large delayed reward more often when additional stimulation was available. The Group × Stimulation interaction was also significant for percentage of choices for the large delayed reward (inline image = .116).

Table 2.   F-ratios, means and standard deviations of delay aversion measure and emotional dimensions for the effects of diagnostic group and the availability of extra stimulation
 EffectsGroups
Group F(2,70)Stimulation F(1,70)Group ×  Stimulation F(2,70)ADHDHFAControls
Without stimulation M (SD)With stimulation M (SD)Without stimulation M (SD)With stimulation M (SD)Without stimulation M (SD)With stimulation M (SD)
  1. Note.*p < .05; **p < .01; ADHD = attention deficit hyperactivity disorder; HFA = high-functioning autism.

Percentage of choices for large delayed reward2.2434.78**4.62*52.80 (33.01)85.80 (12.13)69.57 (33.03)86.09 (12.61)75.60 (27.09)84.80 (13.58)

The presence of this significant interaction effect requires simple effects analyses to interpret. One-way analyses of variance of Group were performed within each level of Stimulation. The Groups differed for the Without Stimulation condition (F(2,70) = 3.59, P < .05), but not for the With Stimulation condition (F(2,70) < 1, ns). Multiple comparisons of the 3 groups within the Without Stimulation condition revealed that the ADHD group differed from HFA (LSD, P < .10) and Control groups (LSD, P < .05), but the HFA and Control groups did not differ from each other (LSD, ns). The interaction between Group and Stimulation is depicted in Figure 1.

Figure 1.

 Interaction effect of diagnostic group and stimulation level for percentage of choices for the large delayed reward

Finally, the ADHD group was divided into two groups, ADHD+ODD (n = 10) and ADHD without ODD (n = 12), based on the clinical cut-off scores of parent and teacher ratings on the DBDRS. Three children were excluded because of missing data. As reported in Table 3, no differences were found between these groups.

Table 3.   F-ratios, means and standard deviations of delay aversion measure and emotional dimensions for the effects of both ADHD groups and the availability of extra stimulation
 EffectsGroups
Group F(1,20)Stimulation F(1,20)Group ×  Stimulation F(1,20)ADHD without ODDADHD with ODD
Without stimulation M (SD)With stimulation M (SD)Without stimulation M (SD)With stimulation M (SD)
  1. Note. **p < .01; ADHD = attention deficit hyperactivity disorder; ODD = opposite defiant disorder.

Percentage of choices for large delayed reward< 123.94**< 147.92 (33.06)82.92 (15.29)55.50 (36.24)88.00 (8.88)

Discussion

The results of this study confirmed the 4 predictions made in the introduction on the basis of the delay aversion hypothesis and are compatible with those obtained from previous studies that used similar delay aversion tasks (Kuntsi et al., 2001a; Solanto et al., 2001; Sonuga-Barke et al., 1992). First, children with ADHD chose the small immediate option more often than controls in a no-stimulation condition. Second, with extra stimulation the children with ADHD were more willing to choose the large delayed reward. Although children in all three groups requested the same amount of additional stimulation, the ADHD group was most sensitive to this manipulation, showing a larger increase than the HFA and Control groups. This provides evidence that children with ADHD may try to compensate for their delay aversion by attending to additional stimulation. Third, when the HFA group was added as a contrast group in the present study, since delay-averse behaviours were previously attributed to children with HFA (Clark et al., 1999), the current findings revealed a similar pattern for children with HFA as for the normal control children on delay aversion measures. Therefore, it can be concluded that children with HFA do not have increased delay aversion.

Finally, the current study aimed to examine the specificity of the delay aversion hypothesis with regard to ODD comorbidity. The current results were in contrast to those of Kuntsi et al. (2001a) who found that co-occurring conduct problems explain most of the association between ADHD and delay aversion. After dividing the ADHD group according to the presence or absence of clinical ODD ratings by both parent and teacher, no group or interaction effects were revealed. In the Kuntsi et al. (2001a) study, group differences on the delay aversion measure did not remain significant after controlling for conduct disorder. Analysing ODD as a categorical variable might produce different results than using ODD as a dimensional variable. This is particularly the case for ODD and CD which are considered as a complication of ADHD (e.g., Taylor, Chadwick, Heptinstall, & Danckaerts, 1996). The issue of comorbidity with ODD relates strongly to the issue of heterogeneity and it could be argued that a delay-averse type of ADHD is likely to be comorbid with ODD. Although there was no evidence in the present study that the presence of elevated levels of ODD altered the reward choice, the power to detect such effect was limited by the small sample size. Future investigations urgently need to compare ADHD children with children with a clinical diagnosis of ODD/CD.

The current study had a number of limitations that are worthy of note. One potentially important factor that is worth while to investigate concerns defining the amount of stimulation needed to overcome delay aversion. This reasoning can be equated with studies looking at delay–reward trade-offs (e.g., Sonuga-Barke et al., 1992) in which the total amount of reward is considered a crucial factor in determining the degree of the motivation to wait in children with ADHD (Solanto et al., 2001). The additional stimulation used in the present study may play an important motivational role in order to overcome delay. Leung and colleagues (2000) previously underlined the importance of examining the distinguishing characteristics related to the nature of the additional stimulation and the way they are presented on the performance and behaviour of children with ADHD. In the present study, children had, once for each trial, the opportunity to request a meaningful visual stimulus. Research in which the presentation probability of stimuli, stimulus salience and stimulus modality is varied is needed to determine more clearly what features define a stimulus as a catalyst for ADHD.

With prudence, two clinical implications can be made. First, the argument that ADHD children are not truly impulsive, but delay averse (i.e., their preference for immediate rewards is conditional on their environment), once again leads to considering the functional meaning of the ADHD symptoms in relation to its environment. Moreover, the present findings show how children with ADHD themselves find strategies to cope with these variations. A good alternative for examining the excess and frequent fluctuations in symptomatology may then derive from explorative studies into the functional analysis of ADHD behaviours (e.g. Ervin, DuPaul, Kern, & Freeman, 1998; Reid & Maag, 1998), which are frequently used in intervention studies. Second, although the MIDA is not a diagnostic tool, delay aversion seem to be a distinguishing feature between ADHD and HFA, which remains a strong impasse in the executive account of both disorders (e.g., Geurts et al., 2004). More research is needed to replicate the present findings to check the predictive power of such a task.

In summary, the current result supported the delay aversion hypothesis by showing that ADHD children choose delay rewards less than normal controls and HFA children but show normal levels of choice when stimulation is added during delay. Further research is, however, warranted to investigate the specificity of this theory with respect to ODD and should incorporate clinically diagnosed children with ODD as well. Similarly, the HFA group contained only children who did not have elevated scores for hyperactivity/impulsivity. However, owing to the overlap of symptoms of ADHD and HFA, the necessity to incorporate combined groups as well needs to be underlined (Tannock, 1998).

Acknowledgements

We would like to thank James M. Swanson for his constructive feedback on earlier versions of this manuscript.

Ancillary