Contrary to conventional wisdom, adults with ADHD manifest clear signs of hyperactivity on objective assessment. Interestingly, effect sizes for ADHD versus control differences in individual activity measures were, on average, 2-fold greater than effect sizes for differences for No-4’s attention measures and 3-fold greater than CPT-II attention measures. These results are similar to findings in children with ADHD
[5, 13, 28]. Activity measures were superior to attention measures even when distinguishing predominantly inattentive subjects with ADHD from control subjects. Random forest classification provided additional confirmation that activity measures were considerably more important than attention measures in discriminating adults with ADHD from healthy controls in this sample. The superiority of activity measures over attention measures emerged when examining the ROC-AUC curves for the composite index scores and in all of the cross-validated predictive mathematical models. In short, objective data provides no support for the hypothesis that hyperactivity fades with age and becomes a less discriminatory feature of the disorder.
This observation was not likely an artifact of recruiting an abnormally hyperactive sample of adults with ADHD. Clinical ratings of these subjects on the WRAAS showed that activity differences were significantly less robust than ratings of attention, or disorganization, as has been observed in previous adult samples. Further, activity measures remained superior to attention measures in discriminating even predominantly inattentive subtype ADHD adults from controls. Objective measures of activity may differ from clinical impressions as subjective ratings of hyperactivity are skewed by the valence of the behavior, such that aggressive individuals are rated as hyperactive regardless of their actual activity levels
. The clinical impression that hyperactivity fades with age may actually be a reflection of fading levels of aggression rather than a normalization of capacity to sit still or to inhibit activity to low levels. Further, the signs or symptoms of hyperactivity used in clinical ratings and diagnostic criteria may simply be age inappropriate for adults. Draft criteria for DSM-5 include four additional symptoms of hyperactivity-impulsivity to better capture the phenomenon in adults.
There are a number of possible interpretations for the predominance of objective activity over attention measures. First, activity may be a more powerful discriminator because it can be measured more accurately. Activity is a physical property and motion analysis systems can track overt movements with precision. Attention is an internal state and is not directly assessed but rather inferred from a subject’s performance on a task.
Second, we may have underestimated the importance of attention by selecting suboptimal tasks, though we are not aware of any task that would have provided superior results. For example, the most robust differences between children with ADHD and controls on the stop signal delay task and Stroop have smaller mean effect sizes (i.e., 0.73
 and 0.58
, respectively) than what we observed in the present study. Meta analyses show that effect size differences for measures of executive function are of intermediate magnitude (d’ = 0.46 – 0.69)
, with the largest reported differences in adults occurring on the Trails Making Test B (d’=0.73)
. Schweiger et al.
 evaluated a computerized test battery in 28 male undergraduates with ADHD and 49 controls. The battery sampled a wide range of cognitive domains including verbal and non-verbal memory, executive function, visual spatial processing, information processing, motor speed and problem solving ability. Significant effect size measures ranged from 0.50 – 0.87. The greatest between-group differences emerged during an extended CPT that was very similar to the CPT-II and No-4’s test
. In short, computer measures of attention or executive function reported in all the studies we examined had effect size findings that were at best equivalent to, and in almost all instances inferior to, effect size differences we observed on the No-4’s test.
Third, attention deficits may simply not be as fundamental to the disorder as the name implies. Individuals with ADHD appear to have problems ‘paying attention’ when tasks are boring or unrewarding, but research shows that they can usually perform as well as controls on tasks that are engaging or provide sufficient incentives
[35–39]. The evidence suggests that their attentional abilities are relatively intact, and their performance difficulties are more related to problems with motivation
[35, 36, 39], reward processing
 or inhibitory control
. Interestingly, ADHD is one of the few psychiatric disorders whose closest equivalent in the International Classification of Disease has a completely different name; i.e., Hyperkinetic Disorder – emphasizing a very different feature of the disorder.
The fourth possibility, which follows logically, is that the ability to inhibit motor activity to low levels during a sustained attention task may be a particularly good index of the primary problem present in individuals with ADHD. Barkley’s theoretical model postulates that the essential impairment in ADHD is a deficit involving response inhibition
. This deficit leads, in turn, to an array of problems with executive functions that depend on behavioral inhibition for their execution
. Neurobiologically, the capacity to inhibit both voluntary and spontaneous motor activity depends on prefrontal corticostriatal circuits
[41–48]. Hence, it may be the case that the capacity to inhibit voluntary responses and the capacity to inhibit spontaneous movements are closely allied. If so, measuring the latter may provide a meaningful index of behavioral inhibition. This possibility is supported by the observation (consonant with Barkley’s theory) that the activity severity composite correlated strongly with multiple domains of impaired executive function on the Brown ADD Scale, and explained about four-fold more of the variance in executive function ratings than either of the attention composites. In short, the strong association between impairments in executive function and objective measures of activity is consistent with the hypothesis that these phenomena are interrelated neurobiologically.
In this regard we have reported a highly significant association between objective measures of hyperactivity and T2-relaxation time (an indirect index of diminished resting cerebral blood volume) in the putamen
, which was further strengthened when T2-relaxation time in the right dorsolateral prefrontal cortex was also taken into account
. Jucaite et al.
 also reported a strong correlation between D2 receptor binding in the right caudate nucleus and motion analysis measures of head movements. Conversely, recent studies have also reported strong associations between measures of caudate and putamen volume, functional activity, connectivity or dopamine release and executive functions in typically developing children
, healthy adults
[51–53], elderly adults
 and patients with schizophrenia and bipolar disorder
. There is, in short, increasing recognition that tonic release of dopamine into the caudate and putamen, in addition to regulating motor activity, also acts to focus and filter non-motor activities such as working memory, implicit learning, decision making, and planning
. Hence, there are compelling reasons to link inhibitory motor control and executive functions, and to envision that a single abnormality may frequently be responsible for deficits in both domains.
These present findings show that tests of attention by themselves perform suboptimally as diagnostic aids for ADHD. This evidence is consonant with previous reports in adults and children. For example, Epstein, Conners’ and colleagues observed that the CPT-II discriminated adults with ADHD from controls with only 55% sensitivity and 76.4% specificity
. Similarly, Edwards et al.
 reported that the CPT-II confidence index had a maximal ROC-AUC of only 0.64 in classifying children with ADHD. We observed an ROC-AUC of 0.63 for classifying adults with ADHD.
There are a number of design differences between the CPT-II and the No-4’s cognitive control task, such as the use of relatively large geometric shapes presented at random screen positions versus the presentation of smaller letters at the same fixed center point. The CPT-II emphasized signal detection theory and regression measures of performance change across blocks or interstimulus intervals. The No-4’s task emphasized an analysis based on fluctuations in attention state
. Random forest regression indicated that 5 or 6 of the No-4’s measures were more important discriminators than any of the CPT-II measures based on mean accuracy or Gini criteria. However, in the end the differences were minor. Neither the CPT-II nor the No-4’s test (by itself) could discriminate between adults with ADHD and healthy controls with acceptable levels of accuracy.
In contrast, measurements of motor activity alone in the present study had the predictive capacity to differentiate a mixed group of ADHD adults from a control group with up to 83% sensitivity and 87% specificity (Table
5), and in a recently published study discriminated 62 combined subtype children with ADHD from 62 controls with up to 100% accuracy
. Combining activity and attention measures in the present study provided excellent discrimination of subjects with ADHD versus controls (ROC-AUC = 0.96). While we do not believe that these measures can substitute for a comprehensive clinical assessment, we do believe that they may aid in diagnosis by providing objective information, and precisely measured target symptoms for gauging response to treatment.
The main limitations of this study are moderate sample size and the selection of adults with ADHD without comorbidity. Further studies will need to assess whether these findings can be replicated, extended to subjects with comorbid disorders, and tested for their capacity to differentiate adults with ADHD from subjects with other disorders that alter activity or impair attention.