Despite these conceptual uncertainties, there is evidence of addiction-related impairment in many of these different aspects of impulsivity suggesting that common (disrupted) mechanisms may be at play. For example, cocaine users show higher delayed discounting rates (Coffey et al. 2003; Simon et al. 2007), are slower or poorer at motor inhibition (Fillmore et al. 2002; Hester & Garavan 2004; Colzato et al. 2007), make riskier decisions on various gambling tasks (Bartzokis et al. 2000; Monterosso et al. 2001; Bolla et al. 2003; Fishbein et al. 2005; Verdejo-García et al. 2007) and amphetamine users sample less of the available information prior to making decisions (Clark et al. 2006). A growing literature suggests both anatomical changes and functional dysregulation in lateral, ventral and medial prefrontal cortex in chronic cocaine users (Volkow et al. 1993; Franklin et al. 2002). Anterior cingulate cortex (ACC) hypoactivity for performance monitoring processes has been demonstrated in chronic cocaine users (not currently under the influence of cocaine) when compared with non-using control participants (Kaufman et al. 2003); a similar effect has also been reported for opiate-dependent individuals (Forman et al. 2004) and cannabis users (Eldreth et al. 2004). Poor motor response inhibition in cocaine users has been associated with reduced activity in dorsolateral prefrontal cortex, insula and the ACC, and increased activity in the cerebellum (Kaufman et al. 2003; Hester & Garavan 2004; Li et al. 2007). Stroop task performance has been associated with changes in orbitofrontal cortex in cocaine users (Goldstein et al. 2001).

The availability of data from a previous control participant study allowed us to observe that an acute cocaine administration rendered activation levels in users largely indistinguishable with that of controls, seemingly ‘normalizing’ the cortical hypoactivity associated with chronic drug abuse. This normalization of function was observed in midline cingulate areas previously shown to be hypoactive for errors in cocaine users (Kaufman et al. 2003) and in right hemisphere parietal and insular regions. By contrast, the right dorsolateral prefrontal cortex region active for STOPS remained hypoactive relative to controls in both the IV cocaine and saline conditions, despite this region increasing in activity in the users following cocaine relative to the saline condition. Although the comparison between users and previously tested controls was imperfect experimentally and should be interpreted with some caution, it is also important to note that stable group differences between users and controls such as sex or education levels cannot account for the different patterns of results observed when comparing controls first to users in the IV saline condition and then to cocaine condition. Furthermore, although the user and control groups did differ in sex composition, we have previously shown that neither performance on the task nor error-related midline activity differs between males and females (Hester et al. 2004).

Much evidence exists for the capacity of stimulant drugs to enhance cognitive performance. This holds true not only for populations with known dysfunction in brain regions targeted by the mesolimbic dopamine system, such as ADHD (Vaidya et al. 1998; Aron et al. 2003; Bedard et al. 2003), but also for normal healthy control populations for whom no pre-existing cognitive deficits are identified (Sostek et al. 1980; Koelega 1993; Wiegmann et al. 1996). As stimulant medications are used in these populations to enhance cognitive performance, it is possible that one aspect of the reinforcing nature of chronic cocaine use is the drug's capacity to improve cognitive function through its action on cortical structures involved in cognitive control. A related possibility is that the cocaine administration alleviated a withdrawal or craving state in the users. While this or other motivational differences may have existed between the cocaine and saline conditions, it is important to note that a similar effect, an increase in electrophysiological error-related signal following d-amphetamine (de Bruijn et al. 2003), has been observed in drug-naive controls for whom withdrawal would not have applied. Furthermore, we found no relationships between the time since the users' last use, which may indirectly index their craving or withdrawal levels, and their performance or activation levels on the task or, most critically, on the change in activation between the IV cocaine and saline conditions.

The phasic modulation of activity levels in specific cortical regions is consistent with cocaine having either a direct or indirect long-term detrimental effect on those same cortical structures. For example, if drug use produces a phasic increase in activity in a brain region and the brain's homeostatic response is to downregulate receptors in that region (Volkow et al. 2002) then, relative to a control condition, one may identify regions of possible downregulation by their increased phasic activity following a drug administration. Tonic downregulation of medial or lateral prefrontal regions may result from repeated exposure to a drug-induced hyperdopaminergic state, which has been suggested to account for decreased dopamine receptor levels in users, and consequently, decreased metabolism in response to stimuli other than the drug itself (Volkow et al. 1999). In this regard, the cognitive tests can serve as functional probes of cocaine's effects. The cognitive tests can also identify the profile of deficits, linked to specific brain structures, likely to accompany drug abuse. Although there is much evidence of impaired cognitive abilities in cocaine users (Fillmore & Rush 2002; Goldstein et al. 2004), the relationship between these behavioural impairments and their underlying neurobiology is not yet very well understood. In this regard, the present results nicely complement previous investigations that have demonstrated diminished inhibitory control in cocaine users (Fillmore & Rush 2002; Colzato et al. 2007). Identifying the neurocognitive profile of this group should inform therapeutic interventions and may also provide an assay of the efficacy of these interventions.

In the human model, it is unclear whether observed deficits reflect a consequence of drug abuse or a pre-existing difference. Neurofunctional deficits such as those observed may render an individual susceptible to the development of addiction (i.e. the transition from recreational to uncontrolled use; Tarter et al. 2003; Dalley et al. 2007; Verdejo-García et al. 2008) and may be related to the psychiatric comorbidities observed in drug-dependent users, or such deficits may result from the effects that prolonged cocaine use may have on the brain or a combination of both these factors. These uncertainties notwithstanding the present results demonstrate that an acute cocaine administration affects the functioning of brain areas critical for cognitive control, thereby showing a direct relationship between cocaine and impulse control, as mediated by right prefrontal cortex, and performance monitoring as mediated by the ACC. Curiously, the results run counter to a hypothesis that acute cocaine would disrupt these control processes. Instead, the observation of improved performance and increased activation levels are consistent with similar ameliorative effects on inhibitory control that have been observed with methylphenidate in patients with ADHD (Scheres et al. 2003). By contrast, alcohol has been shown to impair error monitoring (Ridderinkhof et al. 2002), but this effect was observed in non-alcoholics and is thus in keeping with the neurofunctional effects of drugs of abuse being determined by a history of use and its associated brain changes (discussed further below). That said, as noted above, d-amphetamine increased an electrophysiological marker of error monitoring (but not performance) in drug-naive controls (de Bruijn et al. 2003); d-amphetamine improved information processing but had no effect on inhibitory control in drug-naive controls (Fillmore et al. 2005a,b), while d-amphetamine and cocaine administrations have also been observed to impair inhibitory control in users (Fillmore et al. 2002, 2003).

One important consideration in attempting to reconcile these findings is the role of dose on the observed patterns of cortical activation and behavioural performance. While the beneficial effects of stimulant medications such as methylphenidate to enhance cognitive performance in both children and adults have been demonstrated (Chelonis et al. 2002; Aron et al. 2003; Bedard et al. 2003, 2004), these benefits appear to vary by dose, with more unfavourable behaviours appearing at higher doses (Stein et al. 2003). Similar inverted U-shaped function curves for behaviour are observed in studies of chronic cocaine users (Johnson et al. 1998). A relevant series of studies by Fillmore and colleagues have demonstrated the importance of drug dose insofar as inhibitory control performance on a Go/No-Go task was found to be disimproved following oral cocaine administration in the 50–150mg dose range (Fillmore et al. 2002), but improved following administration in the 100–300mg dose range (Fillmore et al. 2005a,b; but see also Fillmore et al. (2006), in which performance was shown to increase with dose but with different dose–response effects on two different tests of motor response inhibition). Although full dose–response studies are difficult for both methodological and safety reasons, the present results, having established a drug-related enhancement effect, may warrant further study of dose-related effects in future neuroimaging studies.

A further consideration when interpreting a drug's beneficial or deleterious effects on cognitive performance is the drug use history of one's participants. The present study did not include a drug-naive control group. Even if one might surmount the significant ethical and safety issues involved in administering cocaine to drug-naive controls, it is likely the case that the response of controls, given, for example, their baseline levels of dopaminergic activity, might be quite different from those of experienced cocaine users. Individual variation of this kind can be observed in drug-naive controls who, based on working memory capacity measures thought to reflect tonic dopaminergic activity, can show widely divergent performance and brain activation responses following a dopaminergic challenge (Gibbs & D'Esposito 2005). Fillmore and colleagues note that earlier findings of improved inhibitory control following d-amphetamine in drug-naive controls (de Wit et al. 2000) were limited to those subjects who displayed poor inhibitory abilities (Fillmore et al. 2005a,b). These observations suggest that the effects of a drug administration will be modulated by where the recipient falls on that drug's dose–response function curve.