The high heritability of attention deficit hyperactivity disorder (ADHD), averaging at 76 % (Faraone et al. 2005
), has led to studies aiming to identify cognitive endophenotypes that underlie the genetic liability for the disorder (Castellanos & Tannock, 2002
; Kuntsi et al. 2006a
). One of the cognitive indices most strongly associated with ADHD at the phenotypic level is reaction time (RT) variability (Kuntsi et al. 2001
; Klein et al. 2006
), which has made it a potential target for genetic studies.
Although RT variability may not have the instant appeal of neater cognitive neuroscientific concepts - on the surface it could be viewed as the ‘noise’ that we typically aim to reduce in our data - its greater sensitivity to ADHD compared to traditional measures of task performance (Kuntsi et al. 2001
; Klein et al. 2006
) call for a rigorous investigation of this measure. RT variability may reflect lapses in attention and hence relates to the role of sustained attention in ADHD. In aiming to explain RT variability and poor sustained attention, models of ADHD emphasize alternatively the top-down cortical control of executive attention (Bellgrove et al. 2004
) or the role of low arousal and alertness in leading to poor attentional control (van der Meere, 1996
; O’Connell et al. 2008
) or both (Johnson et al. 2007
). Two recent studies that directly measured arousal support the association between decreased arousal and increased RT variability in ADHD: Loo & Smalley (2008)
reported an association between RT variability and electro-encephalogram (EEG)-indexed underarousal, whereas O’Connell et al. (2008)
established a similar association using skin conductance measures.
The proposal that increased RT variability in ADHD relates to underarousal under low stimulation conditions links to theoretical models that view the regulation of state as the key difficulty in ADHD (van der Meere, 2002
; Sergeant, 2005
). According to this view, cognitive performance deficits in ADHD, such as RT variability, disappear under conditions that successfully optimize the child’s arousal, activation or effort state. We recently demonstrated with the present sample that RT variability, and also the slow overall speed, normalized to group means in children with high ADHD symptoms on an RT task called the ‘fast task’ under a condition that combined a fast event rate and incentives (Kuntsi et al. 2009
). These findings extend our previous finding with a clinically diagnosed ADHD group, where RT variability on the same task improved more among the ADHD group than the control group but did not completely normalize (Andreou et al. 2007
). Overall, the findings suggest that RT variability in ADHD can be modulated, at least in part, by energetic or motivational factors.
RT performance is under at least a moderate degree of genetic influence, as indicated by our univariate twin model fitting analyses on approximately 60 % of the present sample (Kuntsi et al. 2006b
). For both mean RT and RT variability, we obtained the highest heritability estimates when using composite scores, based on two tasks. These heritability estimates increased further, from 60% to 73 % for mean RT and from 48 % to 68 % for RT variability, when corrected for measured test-retest unreliability (Kuntsi et al. 2006b
). This suggests a similar degree of underlying heritability for the RT variables as for ADHD.
The evidence for shared genetic effects on ADHD and RT variability first emerged in a small-scale twin study (Kuntsi & Stevenson, 2001
) and further support has since emerged from twin and family studies by independent groups (Nigg et al. 2004
; Bidwell et al. 2007
). For the first time reporting such data for the ‘fast task’, we recently estimated within a sibling design that 60 % (or 100% for a male-only subsample) of the phenotypic association between ADHD and RT variability was due to shared familial influences (Andreou et al. 2007
Despite these advances in our understanding of the association between ADHD and RT variability, we know little about whether this aetiological pathway is shared with other affected cognitive processes in ADHD. Furthermore, the shared versus
unique aetiological association with ADHD has not been investigated across the two RT constructs of mean RT and RT variability. Our sibling study on clinically diagnosed ADHD indicated shared familial effects on mean RT and ADHD and RT variability and ADHD (Andreou et al. 2007
) but a multivariate approach between the two RT constructs has as yet been lacking. ADHD is also associated with lower IQs, which we and others showed to be due to shared genetic influences, for both ADHD diagnosis and continuous ADHD symptom scores in the general population (Kuntsi et al. 2004
; Polderman et al. 2006
). Although one study reports a limited shared genetic aetiology between measures of mean choice RT and IQ in adolescence (with both shared and measure-specific genetic effects; Luciano et al. 2004
), this finding would benefit from replication in school-age children when ADHD may be first diagnosed. It is also not clear whether any aetiological pathway between IQ and RT (Luciano et al. 2004
) contributes to the covariation between ADHD and RT (Andreou et al. 2007
). Here, our main aim was to address the novel question of whether there is a shared set of genes that influence RT variability, mean RT, IQ and ADHD symptom scores or whether separate aetiological pathways exist. In addition, focusing on RT variability, we aimed to investigate how we can maximally increase its detectable genetic association with ADHD symptom scores, to maximize its usefulness for molecular genetic studies. These analyses further incorporate the aim of replicating the findings of shared genetic influences on ADHD and RTV, and on ADHD and IQ, using a general population sample.