|Home | About | Journals | Submit | Contact Us | Français|
We argue that impulsiveness is characterized by compromised timing functions such as premature motor timing, decreased tolerance to delays, poor temporal foresight and steeper temporal discounting. A model illustration for the association between impulsiveness and timing deficits is the impulsiveness disorder of attention-deficit hyperactivity disorder (ADHD). Children with ADHD have deficits in timing processes of several temporal domains and the neural substrates of these compromised timing functions are strikingly similar to the neuropathology of ADHD. We review our published and present novel functional magnetic resonance imaging data to demonstrate that ADHD children show dysfunctions in key timing regions of prefrontal, cingulate, striatal and cerebellar location during temporal processes of several time domains including time discrimination of milliseconds, motor timing to seconds and temporal discounting of longer time intervals. Given that impulsiveness, timing abnormalities and more specifically ADHD have been related to dopamine dysregulation, we tested for and demonstrated a normalization effect of all brain dysfunctions in ADHD children during time discrimination with the dopamine agonist and treatment of choice, methylphenidate. This review together with the new empirical findings demonstrates that neurocognitive dysfunctions in temporal processes are crucial to the impulsiveness disorder of ADHD and provides first evidence for normalization with a dopamine reuptake inhibitor.
Dictionary definitions of impulsiveness include terms such as ‘acting without thinking’, ‘on the spur of the moment’ and ‘in brief intervals of time’. Psychologists define impulsiveness as a lack of persistence, reduced decision time and increased threshold for boredom (Buss & Plomin 1975), lack of patience, risk, sensation and seeking (Eysenck 1993), lack of ‘futuring’/temporal foresight (Barrat 1994) and resistance to delayed rewards (Logue 1995). We have defined impulsiveness as a poorly controlled and inappropriately timed, usually premature, non-reflected, immediateness-bound and delay-aversed response style where actions are executed before all available information and the future consequences are being considered (Rubia 2002). It becomes obvious from these definitions of impulsiveness that abnormalities in functions of timing are an essential component of it.
‘Prematurity’, ‘acting in brief intervals of time’, ‘reduced decision time’ and ‘acting on the spur of the moment, without thinking’, for example, refer to the execution of acts at an inappropriately early moment in time, reflecting poor motor timing and underuse of reflection time. ‘Present boundedness’ or ‘poor futuring’ indicates insufficient use of inter-temporal bridging/temporal foresight. ‘Risk taking and sensation seeking’ reflect temporal myopia or poor consideration of the negative future consequences of such behaviours. ‘Lack of persistence and resistance to delayed rewards’ refer to steeper temporal discounting, i.e. the subjective devaluation of reward in proportion to its delay in time. Steeper temporal discounting is presumed to reflect shorter tolerance of temporal delays and enlarged subjective perception of time. A reduced tolerance to the passage of time is also reflected in other definitions such as ‘lack of patience’ and ‘increased thresholds of boredom’.
In this paper, we argue that abnormalities in timing functions are fundamental to impulsiveness. Accordingly, impulsiveness manifests in abnormalities in different timing functions within different temporal domains that can be classified into the following neuropsychologically measurable subcomponents:
We do not claim that all impulsiveness features can be accounted for by poor timing functions. We have previously argued that besides timing functions, problems with self-control functions, manifesting in poor inhibitory and attention functions are also key ingredients to impulsiveness (Rubia 2002). Timing, attention and inhibitory functions are in fact closely interrelated in behavioural as well as neuropsychological datasets (Olson et al. 2007; Rubia et al. 2007a), and appear to be different aspects of a multifaceted construct of impulsiveness (Reynolds et al. 2008). Most timing functions furthermore co-measure other cognitive basis functions depending on the temporal domains they cover. For example, attention to time is co-measured in many timing tasks, in particular estimation of longer intervals. Time reproduction relies on working memory functions and the ability to delay a response (inhibitory control), whereas psychophysical time discrimination tasks also tap into sensory processes.
Problems with reward and motivation have also been related to impulsiveness (Gray 1987). However, the most consistent association is with reduced sensitivity to reward delays, i.e. the timing of the reward, not the reward itself, reflecting hypersensitivity to the passage of time, not reward. It is not within the scope of this paper to discuss the inter-relationship between timing and these other functions that contribute to the complex construct of impulsiveness. The aim of this paper is to shape out—in our opinion—the underrated and underinvestigated timing aspects of impulsiveness. However, this does not reflect an exclusively reductionistic viewpoint and we acknowledge the importance of other aspects to impulsiveness, in particular inhibitory and attention functions.
In support of the hypothesis that impulsiveness is essentially an abnormality of temporal processes, impulsiveness features have been associated with impairment in these different timing functions both in the normal and pathological populations. Thus, motor timing, time estimation and temporal foresight processes have been shown to be impaired in impulsive personalities (Van den Broek et al. 1992; Ostaszewski 1996; Reynolds & Schiffbauer 2004) and in adult impulsiveness disorders such as substance abuse (Reynolds 2006; Wittmann et al. 2007a,b), borderline personality disorder (Bazanis et al. 2002; Berlin et al. 2005) and mania (Bschor et al. 2004; Christodouiou et al. 2006). They are also impaired in patients with fronto-striatal brain lesions and secondary impulsiveness (Rubia et al. 1997; Bechara & Van der Linden 2005).
Attention-deficit hyperactivity disorder (ADHD) is the archetypal neurodevelopmental impulsiveness disorder. ADHD develops in childhood and is defined by age-inappropriate problems with hyperactivity, inattention and impulsiveness, with impulsiveness being the key feature (APA, DSM IV). All of the above given definitions of impulsiveness are met by the behavioural and cognitive manifestation of ADHD.
Only relatively recently, timing functions have been investigated in children with ADHD. These by and large support the hypothesis that timing problems in several temporal domains are characteristic of ADHD and correlate with their behavioural impulsiveness (Toplak et al. 2006).
One of the most consistent findings are deficits in the fine-temporal discrimination of intervals that differ by milliseconds and seconds (Smith et al. 2002; Toplak et al. 2003; Rubia et al. 2003b, 2007a). Within a large battery of executive function tasks, time discrimination deficits were the most sensitive group discriminator, correctly classifying over 70 per cent of cases and controls (Rubia et al. 2007a).
Although findings have been less consistent, overestimation and under(re)production of time intervals have also been observed, suggestive of a faster internal time sense, which correlated with behavioural impulsiveness in several studies (Toplak et al. 2006).
ADHD children are more inconsistent or erratic in synchronizing or anticipating their motor response to sensory stimulation (Rubia et al. 1999a, 2003b; Ben-Pazi et al. 2006). They also make consistently premature responses across tasks, reflecting a task-independent and ubiquitous deficit that has been considered an indirect indicator of poor motor timing (Rubia et al. 2001, 2007a); after time discrimination errors, with which it was closely inter-correlated, it was one of the best discriminators between ADHD and controls among a range of neuropsychological measures (Rubia et al. 2007a).
Recent evidence also has pointed at problems with temporal foresight and temporal discounting of longer intervals. ADHD children show poor inter-temporal choice in gambling tasks, in correlation to their behavioural impulsiveness (Garon et al. 2006; Toplak et al. 2008) and have steeper temporal discounting (Luman et al. 2005).
It cannot be excluded, however, that the timing deficits observed in ADHD are related to other cognitive deficits. Problems with attention to time, working memory and inhibition of immediate responding, for example, rather than abnormalities with temporal processes per se, could well affect performance on timing tasks.
Interestingly, the neural correlates of those timing functions that are impaired in ADHD children are strikingly similar to the neural correlates of ADHD, further supporting the link between the two.
Several brain regions have been implicated in timing functions, including prefrontal and parietal cortices, the basal ganglia, the cerebellum, anterior cingulate gyrus (ACG) and supplementary motor area (SMA). Evidence from the literature suggests that each of these brain regions may contribute to specific aspects of timing functions (Rubia & Smith 2004; Rubia 2006; Wittmann 2009).
The cerebellum and the basal ganglia seem to be the closest in mediating pure timing mechanisms as they have been associated with motor and perceptive timing of both short and long temporal domains. The cerebellum is important for event timing and temporal prediction (Ivry et al. 2004). The basal ganglia show sustained activity during temporal intervals and have been suggested to play a role in sustained attention to time and inter-temporal bridging (Buhusi & Meck 2005; Koch et al. 2009).
Lateral prefrontal regions are particularly involved in timing processes that require the temporal bridging of longer temporal intervals, which has been linked to other functions that are known to be mediated by these regions such as sustained attention to time, delay of immediate responses, magnitude comparation and working memory (Koch et al. 2009; Lewis & Miall 2006).
The SMA and the ACG have traditionally been related to motor timing, although more recent evidence has showed they have a role in pure perceptual time estimation, necessary for fine-temporal adjustment of movement. The ACG and SMA via their close connections to fronto-striatal pathways may have a more generic role in attentional components necessary for both motor timing and time estimation in an evaluative comparator role (Macar et al. 2006; Rubia 2006).
The parietal lobes have been related to perceptual timing and temporal foresight of longer time domains, and are thought to support timing functions through the allocation of sustaining attention to time and their function as a number and magnitude comparator (Rubia 2006).
It thus appears that complex fronto-parieto-striato-cerebellar neural networks are involved in timing functions concerting efforts from these different brain regions that mediate basis functions that together compose complex temporal processes (Rubia 2006).
All these brain regions that are crucial to temporal processes are impaired structurally and functionally in ADHD. Structural imaging studies in children with ADHD have demonstrated abnormal brain volumes, a delayed development and reduced white matter integrity in the cerebellum, caudate, frontal, temporal and parietal lobes (Ashtari et al. 2005; Shaw et al. 2007; Valera et al. 2007).
Functional imaging studies show dysfunctions in the same brain regions during tasks of cognitive control, most consistently in the inferior prefrontal cortex, caudate and anterior and posterior cingulate (Rubia et al. 1999b, 2001, 2005a, 2007b, 2008, 2009; Smith et al. 2006) but also in temporo-parietal and cerebellar areas during attention tasks (Smith et al. 2006; Rubia et al. 2007b, 2009).
To investigate our hypothesis that ADHD is associated with neurofunctional abnormalities of brain regions that mediate timing functions, we scanned adolescents with ADHD while performing several functional magnetic resonance imaging (fMRI)-adapted timing tasks of different temporal domains. In our first studies in 1999 and 2001, we investigated the neural correlates of sensorimotor synchronization to time intervals that either lasted 600ms—measuring pure sensorimotor synchronization—or 5s—requiring additional frontal lobe-mediated time estimation. Adolescents with ADHD showed no performance differences in the task and no brain dysfunctions during sensorimotor tapping to the short 600ms interval (Rubia et al. 2001); however, they showed significant underfunctioning of the anterior and posterior cingulate during motor timing to the 5s interval (Rubia et al. 1999b, 2001; figure 1a). In a subsequent study, we investigated the neural correlates of time discrimination of hundreds of millisecond differences from a base interval of 1s in 21 medication-naive adolescents with ADHD compared with 17 control children. We found significantly reduced activation in the ADHD adolescents in brain regions, which correlated significantly with time discrimination performance in controls in right inferior and dorsolateral prefrontal cortices, the ACG and SMA (Smith et al. 2008; figure 1b).
In order to investigate the neural substrates of the tolerance of time and the use of temporal foresight in ADHD children, we conducted a third fMRI experiment using a task of hypothetical temporal discounting of longer intervals of weeks to years in 10 adolescents with and without ADHD. The choice of delayed versus immediate rewards during delay discounting (DD) tasks has been associated with the activation of lateral prefrontal and parietal cortices (McClure et al. 2004), but also subcortical regions including the insula, basal ganglia and cerebellum, suggestive of a fronto-parieto-striato-cerebellar network for temporal discounting (Hariri et al. 2006; Wittmann et al. 2007a,b).
Given the evidence for steeper temporal discounting in ADHD children (Luman et al. 2005) and abnormalities in brain regions that mediate these functions (§3), we hypothesized that ADHD adolescents would show underactivation in task-relevant brain regions of lateral prefrontal and parietal cortices, basal ganglia and cerebellum (McClure et al. 2004; Hariri et al. 2006).
Ten male right-handed adolescent boys in the age range of 11–16 years (mean age=14, s.d.=2) with a clinical diagnosis of the combined inattentive–hyperactive subtype of ADHD were recruited through parent groups and clinics. Exclusion criteria were neurological abnormalities, epilepsy and substance abuse, co-morbidity with other major psychiatric disorders including language or learning problems, except for conduct disorder, met by one patient. Four patients were medication naive and the other six were taken off methylphenidate (MPH) for 36 hours prior to scanning. Patients scored above the cut-off for hyperactive symptoms on the strengths and difficulties questionnaire (SDQ; Goodman & Scott 2005).
Ten male right-handed adolescent boys between 11 and 17 years (mean age=15, s.d.=4) were recruited through advertisements; they scored below the cut-off for behavioural problems on the SDQ and had no history of psychiatric disorder.
All participants achieved a score above 80 on the Wechsler abbreviated scale of intelligence (Wechsler 1999) (mean IQ controls=119, s.d.=7; patients=109, s.d.=9) and were paid £30 for participation. Parental and child informed consent/assent and approval from the local ethical committee were obtained.
Univariate ANOVAs showed no group differences in age (F1,19=2, p=n.s.), but IQ (F1,19=7; p<0.02). IQ and age had no effect on the performance variables. Nevertheless, because controls differed in IQ and were on average a year older, all subsequent analyses were co-varied for both IQ and age.
In the 12min task, subjects choose by pressing a left or right button (with the right index or right middle finger) between a smaller amount of money (between £0 and £100), immediately, and a larger amount (always £100) after 1 week, 1 month or 1 year; choices were randomly displayed (20 trials for each delay) to the right and left sides of the screen for 4s, followed by a blank screen of at least 8s (inter-trial interval: 12s). The immediate reward is adjusted in an algorithm based on previous choices for each of the three different delays to narrow the range of values converging into an indifference value that is considered by the subject as equivalent to the delayed reward (Richards et al. 1999). The algorithm ensures that equal numbers of immediate and delayed reward choices are obtained at the end of the task which are then contrasted in the fMRI analysis.
Reward is typically discounted in a hyperbolic function that depends on amount, delay and a free impulsiveness indicator k, the main dependent task variable that can be calculated by fitting a hyperbolic function to the indifference values for every delay. Larger k values are associated with greater reward devaluation (Richards et al. 1999).
fMRI images were acquired on a 3T General Electric (GE) MRI scanner. In each of 22 non-contiguous planes parallel to the anterior–posterior commissure, 480 T2-weighted MR images were acquired depicting blood oxygen level-dependent (BOLD) contrast with echo time (TE)=30s, repetition time (TR)=1.5s, 22 functional slices and functional slice thickness=5.5mm (superimposed on a high-resolution (3mm) image of the whole brain).
Time-series analysis for each individual subject was based on a wavelet-based data resampling method for functional MRI data (Bullmore et al. 1999). Images were registered into Talairach standard space. A group brain activation map was then produced for each experimental condition and hypothesis testing was carried out at the cluster level, shown to give excellent cluster-wise type I error control in structural and functional fMRI analyses (Bullmore et al. 1999). For each task, less than 1 false-positive-activated cluster was expected at a p-value of <0.05 at the voxel and <0.02 at the cluster level. ANCOVA analysis for between-group differences was conducted using randomization-based test for voxel or cluster-wise differences (Bullmore et al. 1999). Less than 1 false-activated cluster was expected at a p-value of p<0.05 for voxel and p<0.006 for cluster comparisons.
Discounting followed the typical hyperbolic function. Univariate ANCOVAs showed no group differences for the main impulsiveness factor k (controls mean k=0.015, s.d.=0.012; ADHD means k=0.034, s.d.=0.048, F1,19=0.1, p=n.s.); larger k was positively correlated with higher SDQ hyperactivity scores (r=0.4, p<0.04). However, there was a larger reaction time effect for delayed choices compared with immediate choices for controls (i.e. delayed choice RT effect (RT delayed–RT immediate): controls mean=178ms, s.d.=159ms; ADHD mean=11ms, s.d.=373ms, F1,19=5, p<0.036), suggesting that controls, but not ADHD, deliberated longer when choosing delayed over immediate rewards.
Controls activated bilateral cerebellum, posterior cingulate/precuneus and parietal lobe. Boys with ADHD activated insula, SMA, premotor cortex and parietal lobes.
ANCOVA showed increased activation for controls compared with ADHD boys in two clusters; one comprising left orbital and inferior prefrontal cortices, putamen, thalamus, inferior parietal lobe, posterior cingulate/precuneus and cerebellum; and the other in the right inferior prefrontal cortex (table 1; figure 1). Activation in both group difference clusters correlated significantly negatively with the hyperactivity measures on the SDQ (r=0.6, p<0.005). The activation in cerebellum correlated positively with the delayed choice RT effect (r=0.5, p<0.013). ADHD compared with control boys showed no increased activation.
Although ADHD children had steeper discounting values that correlated with behavioural symptoms, this did not reach significance. This is probably due to the small number of subjects for neuropsychological data and the older age group tested in this fMRI study compared with the younger populations that have previously shown to have deficits in the task (Luman et al. 2005). However, ADHD children deliberated significantly less than control boys over delayed compared with immediate reward choices, suggesting a decreased use of inter-temporal bridging time and thus presumably less foresighted delayed reward choices.
In the fMRI data, ADHD boys showed significantly reduced activation in left and right inferior prefrontal and left lateral orbitofrontal cortices, as well as in left thalamus, putamen, cerebellum, inferior parietal lobe and posterior cingulated/precuneus. As mentioned above, ventrolateral fronto-parieto-striato-cerebellar brain regions are typical areas of temporal discounting (McClure et al. 2004; Hariri et al. 2006; Wittmann et al. 2007a,b) as well as key areas of dysfunction in ADHD children during tasks of inhibition and attention (Rubia et al. 1999b, 2005a, 2007b, 2008, 2009). It is likely that these ventrolateral fronto-parieto-striato-cerebellar networks that are compromised in ADHD are mediating several component functions that are necessary for temporal foresight. The orbitofrontal cortex, for example, is thought to be important for holding information in representational memory as well as incentive motivation, and is thus crucial for comparator operations of future and current rewards (Schoenbaum et al. 2006). Inferior prefrontal activation is inversely associated with the steepness of DD in adults (Wittmann et al. 2007a,b) but also mediates closely related functions of sustained attention, inhibition and working memory (Rubia et al. 2003a, 2006, 2007c, 2009); these functions may contribute to delayed gratification, which involves the inhibition of the immediate reward as well as attention to and holding online of the future reward perspective. The inferior prefrontal dysfunction is a key finding in ADHD in the context of millisecond time discrimination (Smith et al. 2008) as well as non-timing cognitive control functions (Rubia et al. 1999b, 2005a, 2008, 2009). Furthermore, both the right inferior prefrontal as well as the left fronto-striato-cerebellar dysfunctions correlated with the behavioural hyperactivity measures. The inferior parietal regions are important for allocating attention to time, imagery and quantity representations, and may thus contribute to inter-temporal choice in their role as magnitude comparators of both time and reward (Sandrini et al. 2004). The posterior cingulate mediates visual–spatial attention to reward (Small et al. 2003) and delay gratification (Wittmann et al. 2007a,b), and was also reduced in ADHD children during rewarded attention trials (Rubia et al. 2009) and may thus reflect reduced attentional representation of the delayed reward option. The putamen is important for attention to time, in particular when comparing long with short intervals (Rubia et al. 1998; Coull et al. 2004). As mentioned, the cerebellum is a key area for both long and short interval timings and was the only region that correlated positively with the delayed RT choice effect, suggesting that this region may be instrumental in the temporal perspective of the expected delayed gratification. The functional interplay between these different brain regions may thus provide the combination of skills that are necessary to compare and then make an informed decision for delayed gratification. The reduced recruitment of this ventrolateral fronto-striato-cerebellar network in ADHD adolescents during delayed gratification choices combined with the reduced use of reflection time—a typical impulsiveness feature—may be the neural substrate for the typically observed reduced temporal foresight mechanisms in ADHD.
To summarize, these three fMRI studies demonstrate that ADHD children have task-relevant brain dysfunctions during different temporal processes, motor timing, time discrimination and temporal discounting, covering three different temporal domains. Brain dysfunctions were in anterior cingulate during motor timing of seconds and millisecond time discrimination, presumably reflecting attention to time; in right lateral prefrontal cortex during time discrimination and temporal discounting of longer intervals, an area thought to mediate time estimation; and in a left ventrolateral fronto-parieto-striato-thalamo-cerebellar network during temporal discounting, presumably mediating temporal foresight. These brain regions are key areas of functional and structural brain abnormalities in ADHD, also during inhibition and attention, which may interact with timing abnormalities to provide the complex construct of impulsiveness.
Dopamine is the neurotransmitter that has most consistently been implicated in timing functions. Dopaminergic agonists/antagonists are known to improve/deteriorate time estimation, motor timing and temporal discounting in humans (Rammsayer 1993; Takahashi 2007) and animals (Buhusi & Meck 2005).
Given that timing processes of longer intervals of several seconds co-measure other than pure timing functions, such as working memory, attention to time, magnitude estimations and the ability to inhibit immediate responses, several neurotransmitters besides dopamine including noradrenergic, serotoninergic, GABAergic and glutamatergic systems have also been involved in these functions (Rammsayer et al. 2001; Rammsayer 2003). Shorter subsecond time-interval timing, however, appears to be more specifically modulated by dopaminergic activity (Rammsayer & Stahl 2006).
ADHD is thought to be mediated by a dopamine dysfunction; there is consistent evidence for elevated striatal dopamine transporter levels and reduced DA availability (Krause 2008). A dopamine dysfunction is likely to account for their deficits in timing functions.
The psychostimulant MPH is the most effective, first choice treatment for the behavioural symptoms of ADHD (Arnsten 2006). However, little is known on its mechanisms of action. MPH is a catecholamine reuptake inhibitor with stronger dopaminergic effects subcortically (Arnsten 2006).
Neuropsychological studies show that MPH improves abnormal timing functions in ADHD children. We found that chronic MPH administration over a month improved speed, errors and intra-subject variability of sensorimotor synchronization and anticipation (Rubia et al. 2003a). Acute doses of MPH have been shown to increase precision of subsecond synchronization (Ben-Pazi et al. 2006), of time estimation of several seconds (Baldwin et al. 2004) and to improve inter-temporal choice in gambling tasks in patients with ADHD (DeVito et al. 2008).
Relatively, few fMRI studies have studied the effects of MPH on brain activation in ADHD, finding enhanced activation in the caudate and ACG (Vaidya et al. 1998; Shafritz et al. 2004). However, no fMRI study so far has investigated temporal processes.
In order to investigate the effects of MPH on timing functions in ADHD, we conducted a double-blind, randomized, placebo-controlled fMRI experiment in 12 medication-naive boys with and without ADHD while they performed a task of time discrimination. We investigated (i) the effects of MPH compared with placebo within patients and (ii) potential amelioration or normalization effects of MPH on brain dysfunctions in comparisons between controls and patients under either placebo or medication.
Given the evidence for dopamine involvement in subsecond time discrimination (Rammsayer 1993; Rammsayer & Stahl 2006), we used a discrimination task of temporal differences of hundreds of milliseconds which has shown to elicit deficits in ADHD children (Smith et al. 2002; Rubia et al. 2007a). The task activates key timing regions of right dorsolateral and inferior prefrontal cortices, the cerebellum, ACG and the SMA in adults and children (Smith et al. 2003, 2008), and elicits underactivation in ADHD adolescents in right lateral prefrontal cortex, ACG and SMA (§3; figure 1b; Smith et al. 2008). We hypothesized that we would replicate our previous findings of fronto-cingulate underactivation in ADHD adolescents during the placebo condition and, furthermore, that MPH would ameliorate or normalize the activation differences between patients and controls.
Twelve male right-handed adolescent boys in the age range of 10–15 years (mean age=13, s.d.=1), who met clinical diagnostic criteria for the combined, inattentive–hyperactive subtype ADHD (DSM-IV), were recruited through clinics. All the patients were medication naive. Exclusion criteria were co-morbidity with other major psychiatric disorders except for conduct disorder (one patient), with language and learning disabilities, neurological abnormalities, drug or substance abuse. Patients with ADHD scored above cut-off for hyperactive symptoms on the SDQ (Goodman & Scott 1999).
The patients were scanned twice, in a randomized fashion, one week apart, 1 hour after either 0.3mgkg−1 of MPH administration or placebo (vitamin C, 100mg).
Twelve male right-handed adolescent boys in the age range of 11–16 years (mean age=13, s.d.=1) were recruited through advertisements. They scored below the cut-off for the SDQ, and had no history of psychiatric disorder.
All the participants were above the fifth percentile on the Raven progressive matrices performance IQ (Raven 1960; IQ mean estimate controls=100, s.d.=14; ADHD=91, s.d.=9) and were paid £30 for participation. Parental and child informed consent/assent and approval from the local Ethical Committee were obtained.
Univariate ANOVAs showed no group differences between children with ADHD and controls for IQ (F1,23=0.8, p=n.s.) or age (F1,23=4, d.f.=1, p=n.s.).
The block design time discrimination task consisted of 5×30s alternated blocks for each condition (six trials each). In the time discrimination blocks, starting with the appearance of a cue letter ‘L’ for 3s, a pair of differently coloured circles was presented consecutively left and right from each other. One of these circles was randomly presented for a standard duration of 1000ms, and the comparison circle for 1300, 1400 or 1500ms. Subjects had to decide which of the two circles had the longer duration by pressing either the left or right button (left/right thumb). The temporal order judgement (control) condition was identical to the time discrimination condition except that these blocks began with the presentation of the cue number ‘2’ and subjects had to indicate which circle came second. For the fMRI analysis, activation during the control condition was subtracted from activation during the time discrimination condition (Smith et al. 2003, 2008).
fMRI images were acquired on a 1.5T GE MR scanner. In each of 16 non-contiguous planes parallel to the anterior–posterior commissure, 100 T2-weighted MR images depicting BOLD contrast were acquired with TE=40ms; TR=3s; flip angle 90°; 64×64 matrix; in-plane voxel size 3.75×3.75mm; slice thickness=7mm, slice skip=0.7mm. A high-resolution structural scan of 128 axial slices was acquired after the functional series.
Individual and group analyses were as described in §4. For group activations, less than 1 false-activated cluster was expected at a p-value of <0.05 for voxel and <0.01 for cluster comparisons. Three ANOVA analyses were conducted as described in §4. To test for a within-group effect of MPH versus placebo, a within-group repeated-measures ANOVA was conducted. To test for normalization effects, two ANOVAs were conducted: one between controls and patients under the placebo condition and the second one between controls and patients under the MPH condition. Less than 1 false-activated cluster was expected at a p-value of <0.05 for voxel and <0.01 for cluster comparisons. To adjust for multiple testing, we used a more lenient threshold of p<0.006. A more lenient p-value of <0.05 was used to test for hypothesized subthreshold medication effects.
Although the error rates in ADHD children were smaller during the MPH condition, this did not reach significance in the within-subject repeated measures ANOVA analysis (F1,11=0.3, p=n.s.). Between-group ANOVAs showed no group differences between controls and ADHD patients under the placebo (F1,23=1, p=n.s.) or MPH conditions (F1,23=1.4, p=n.s.; error rates: controls: mean=18%, s.d.=17%; ADHD placebo: mean=24%, s.d.=17%; ADHD MPH: mean=16%, s.d.=22%).
Within-group activations for the contrast of time discrimination–order judgement are shown in figure 2 for controls (figure 2a), and for ADHD patients under the placebo (figure 2b) or the MPH condition (figure 2c). In all subjects activation was observed in dorsolateral and inferior prefrontal cortices, ACG and SMA, left caudate and insula.
Repeated-measures ANOVA within patients showed that MPH compared with placebo increased brain activation in a left hemispheric cluster in orbital and inferior prefrontal cortices. At a more lenient threshold (p<0.05), MPH also increased right dorsolateral prefrontal cortex, right cerebellum and ACG (table 2a; figure 3a).
Placebo compared with the MPH condition enhanced activation in patients in right hemispheric medial and superior frontal gyri, medial temporal lobe, hippocampus and lentiform nucleus (table 2b; figure 3b).
ANOVA comparison between controls and patients under the placebo condition showed increased activation in controls compared with patients for the contrast of time discrimination–order judgement in bilateral orbitofrontal cortex, caudate and ACG. At a more lenient threshold (p<0.05), there was also increased activation in right cerebellum, right dorsolateral prefrontal cortex and a more dorsal part of ACG (table 3a; figure 4a). Activation in dorsal ACG furthermore correlated with time estimation errors in all subjects (r=−0.4, p<0.03).
Patients compared with controls showed enhanced activation in two large clusters, comprising left temporal and occipital regions, and right superior temporal and inferior occipital regions reaching into dorsomedial frontal cortex (table 3a; figure 4a).
ANOVA comparison between controls and patients under the MPH condition revealed no significant differences, even at a more lenient cluster p-value of <0.01 (figure 4b).
Despite non-significant differences in task performance, ADHD compared with control boys showed decreased activation during time discrimination in bilateral orbital, inferior and mesial prefrontal cortices, and caudate and enhanced activation in predominantly posterior regions of temporal lobes, thalamus and putamen. Within patients, MPH did not significantly increase performance, but enhanced activation in left inferior and right dorsolateral prefrontal cortices, ACG and cerebellum. MPH completely normalized all group activation differences that were observed under the placebo condition.
Despite the fact that the group differences and changes in performance elicited with MPH were in the right direction (reduced error rates in ADHD patients at baseline which were increased with MPH), they did not reach significance. This may be due to the relatively low statistical power for neuropsychological data and the use of an older adolescent age group compared with the childhood age groups, previously shown to have time- discrimination deficits (Smith et al. 2002; Rubia et al. 2007a). The findings show that brain activation is more sensitive than performance to detect abnormalities and pharmacological effects. We have previously shown that adolescents with ADHD show marked brain dysfunctions despite not being significantly impaired in task performance (Rubia et al. 1999b, 2001, 2005a, 2008, 2009), even during the same time discrimination task (Smith et al. 2008; §4). We have also shown that the brain is more sensitive than behaviour to show pharmacological effects (Rubia et al. 2005b). However, some of the reduced activation clusters in ADHD patients which were normalized with stimulants correlated negatively with error rates, thus reinforcing the relationship between reduced brain activation and (albeit non-significantly) lower performance.
The underactivation in right prefrontal cortex and ACG under a more lenient threshold replicates our previous findings in a different group of 21 ADHD patients during the same task (§4; figure 1b; Smith et al. 2008). ACG was the only region that correlated with time estimation errors. The ACG underactivation in ADHD children was furthermore observed during motor timing to a 5s interval (figure 1a). As discussed in §3, the ACG has consistently been implicated in timing functions of several temporal domains and is likely to mediate important basis functions that are necessary for timing processes such as attending, monitoring and comparing time intervals (Rubia et al. 1998; Rubia 2006).
The statistically more powerful dysfunction in bilateral ventromedial orbitofrontal cortex and the caudate as well as the subthreshold underactivation in the cerebellum was not observed in our previous study (figure 1b). As reviewed in §3, however, inferior/orbitofrontal cortex, ACG, caudate and cerebellum are key areas of structural and functional abnormalities in ADHD as well as key mediators of time discrimination (Smith et al. 2003, 2008). Furthermore, we also observed orbitofrontal dysfunction in ADHD adolescents during the temporal discounting task, albeit more laterally, together with reduced cerebellum and putamen activation (figure 1c). Lesion studies have attributed an important role to orbitofrontal cortex and its striatal connections in both temporal foresight of longer intervals as well as time estimation and discrimination of shorter time spans (Rubia et al. 1997; Bechara & Van der Linden 2005), possibly linked to its role in holding information in representational memory (Schoenbaum et al. 2006), important for the comparison of both long and short time intervals. The cerebellum plays an important role in millisecond discrimination (Smith et al. 2003, 2008; Rubia 2006). It is interesting that a similar location of the lateral cerebellum as well as the vermis was reduced during both temporal discounting and time discrimination, in line with the role of the cerebellum in the timing of both shorter and longer intervals (Rubia & Smith 2004; Rubia 2006). The cerebellum is the most compromised brain structure in ADHD (Valera et al. 2007) and a functional abnormality during short and long interval timing processes could be the underlying neural substrate for several impulsiveness-related timing deficits.
Within patients, MPH primarily enhanced left orbital/inferior prefrontal activation, but at a more lenient threshold also the ACG and right lateral prefrontal cortex, which were dysfunctional during placebo. As mentioned before, inferior prefrontal dysfunction is the most consistent fMRI finding in ADHD during tasks of inhibition and attention (§4), and even appears to be a specific biomarker for ADHD when compared with related pathologies such as conduct disorder (Rubia et al. 2008, 2009). The fact that inferior prefrontal dysfunction was also observed in other ADHD samples during the same task and during temporal discounting (§5; Smith et al. 2008; figure 1b,c) suggests that this brain dysfunction in ADHD may not only account for their poor attention and inhibitory control, but also be instrumental to their timing deficits. The inferior prefrontal cortex depending on its functional interconnectivities is important for many higher cognitive functions, and a localized dysfunction in this region in ADHD could account for the different deficits in the interrelated functions of attention, inhibition and timing. The MPH-induced enhancement and normalization of this consistently reported and potentially disorder-specific brain dysfunction in the context of timing could be associated with the disorder-specific effectiveness of MPH on ADHD-impulsive behaviours.
To our knowledge, this is the first functional imaging study that demonstrated (i) increased activation with MPH in ADHD patients in the most consistent dysfunction area, the inferior prefrontal cortex, and (ii) complete normalization of all brain activation abnormalities. Few functional imaging studies have included healthy controls to test for normalization effects of MPH. Only one of these studies found normalization of activation, but only in the caudate (Shafritz et al. 2004), while another study showed enhanced, but not normalized, activation in caudate and frontal regions (Vaidya et al. 1998). Not even chronic MPH administration over four weeks could show complete normalization of prefrontal and subcortical functions during attention tasks (Konrad et al. 2007; Bush et al. 2008). The normalization findings in ACG, inferior prefrontal cortex, caudate and cerebellum—key regions of both time discrimination and ADHD dysfunction (§4)—reinforce the association between dopaminergic neurotransmission abnormalities, ADHD and timing. While the effects on the caudate were probably mediated by the dopaminergic system (Krause 2008), the effects on frontal, cingulate and cerebellar brain regions, however, could also have been influenced by noradrenaline reuptake inhibition (Arnsten 2006).
In conclusion, this study shows that MPH has a complete normalization effect on brain dysfunctions in ADHD children in key areas of fine-temporal perception such as orbital and inferior prefrontal cortices, the caudate, ACG and cerebellum. Complete normalization with MPH on brain dysfunction has not been observed in any of the other disorder-relevant cognitive tasks of inhibition and attention (Vaidya et al. 1998; Shafritz et al. 2004), which suggests that MPH may impact more prominently on neurocognitive abnormalities of timing functions, in line with the implication of DA in both timing and ADHD.
In this paper, we claim that impulsiveness is essentially a problem of poor timing functions. We illustrate this in the classical neurodevelopmental impulsiveness disorder of ADHD. We review evidence that children with ADHD have deficits in timing processes of several temporal domains and point out the similarities between brain correlates of timing and ADHD brain pathology. Furthermore, we review and provide new direct fMRI evidence that ADHD children show dysfunctions in timing–mediating prefrontal, cingulate, striatal and cerebellar brain regions during cognitive time management across several temporal domains such as time discrimination of milliseconds, motor timing to seconds and temporal discounting of longer intervals of weeks to years. Given that impulsiveness, timing deficits and more specifically ADHD have been related to dopamine dysregulation, we tested the effect of the dopamine agonist MPH on brain dysfunction during a time-discrimination task. Despite no significant performance effects, MPH normalized all brain dysfunctions in ADHD patients. The findings show the superior sensitivity of the brain compared with behaviour to medication effects in ADHD and reinforce the importance of the role of dopamine-mediated timing dysfunctions in the impulsive pathology of ADHD. Future studies should investigate whether other pharmacological agents such as the noradrenaline agonist atomoxetine or behavioural interventions can normalize timing-related brain dysfunction in ADHD. We believe that the role of a disruption of temporal processes underlying the impulsive pathology of ADHD is relatively underinvestigated and hope this review motivates more research into the timing aspects of ADHD.
This research was funded by two grants from the Medical Research Council UK to E.T. (G9900839) and K.R. (GO300155).
One contribution of 14 to a Theme Issue ‘The experience of time: neural mechanisms and the interplay of emotion, cognition and embodiment’.