We propose that in individuals with drug addiction, PFC activity in response to non-drug related rewards is opposite to PFC activity changes that characterize drug-related processing (). Specifically, in addicted individuals who are in a state of craving, intoxication, withdrawal or early abstinence, sensitivity of the PFC to non-drug related rewards will be markedly attenuated compared with that in healthy non-addicted subjects. Indeed, decreased sensitivity to non-drug related rewards is a challenge in the therapeutic rehabilitation of patients with substance use disorders. Therefore, it is important to study how drug-addicted individuals respond to non-drug related reinforcers.
Such decreased sensitivity to non-drug related reward has been explained as an allostatic adaptation59
. In this interpretation, frequent and high-dose drug use leads to compensatory brain changes that limit appetitive hedonic and motivational processes (‘reward’), instead strengthening aversive (opponent or ‘anti-reward’) systems60
. This process is similar to tolerance, in which sensitivity to reward is decreased. It is also captured by the opponent-process hypothesis set forth by Slomon and Corbit61,62
, which describes the temporal dynamics of opposing emotional responses; here, negative reinforcement (for example, withdrawal) prevails over positive reinforcement (for example, drug-induced high) in the transition from occasional drug use to addiction. This process is relevant to emotional reactivity and emotion regulation, which, insofar as emotions are defined as ‘states elicited by reinforcers’63
, are bound to be impaired in drug addiction, especially during drug-biased processing such as craving and bingeing.
Anhedonia is a defining characteristic of drug dependence64
, and criteria for major depressive disorder — which includes anhedonia as a core symptom — are met by many drug-addicted individuals (for example, 50% of cocaine-addicted individuals65
). The strong association between mood and substance use disorders is not limited to depression66
; for example, emotional distress is a risk factor for drug relapse67
. However, research on how altered emotion processing is implicated in substance use disorders is in its infancy68,69
, as discussed below (Supplementary information S5 (table)
Money is an effective abstract, secondary and generalizable reinforcer that acquires its value by social interaction, and it is used in emotional learning in everyday human experience; compromised processing of this reward may therefore point to a socially disadvantageous emotional learning mechanism in addiction. Such a deficit, all the more distinct given the strong motivational and arousal value that is normally associated with this reward, would corroborate the idea that in addiction, brain reward circuits are ‘hijacked’ by drugs, although the possibility for a pre-existing deficit in reward processing also cannot be ruled out.
One fMRI study investigated how cocaine-addicted individuals and controls responded to receiving monetary reward for correct performance on a sustained attention and forced-choice task70
. In controls, sustained monetary reward (gain that did not vary within task blocks and that was fully predictable) was associated with a trend for the left lateral OFC to respond in a graded fashion (activity monotonically increased with amount: high gain > low gain > no gain), whereas the DLPFC and rostral ACC responded equally to any monetary amount (high or low gain > no gain). This pattern is consistent with the OFC’s role in processing relative reward, as documented in non-human71
and human subjects72–76
, and with the DLPFC’s role in attention77
. Cocaine-addicted subjects showed reduced fMRI signals in left OFC for high gain compared to controls and were less sensitive to differences between monetary rewards in left OFC and in DLPFC. Remarkably, more than half of the cocaine-addicted subjects rated the value of all monetary amounts equally (that is, US$10 = US$1000)78
. Eighty-five percent of the variance in these ratings could be attributed to the lateral OFC and medial frontal gyrus (and amygdala) responses to monetary reward in the addicted subjects. Although these findings need to be replicated in a larger sample size and with more sensitive tasks, they nonetheless suggest that some cocaine-addicted individuals may have reduced sensitivity to relative differences in the value of rewards. Such ‘flattening’ of the perceived reinforcer gradient may underlie over-valuation or bias towards immediate rewards (such as an available drug)79
and the discounting of greater but delayed rewards80,81
, therefore reducing sustained motivational drive. These results may be therapeutically relevant as monetary reinforcement in well-supervised environments has been shown to enhance drug abstinence82
, and may also be relevant in predicting clinical outcomes. In line with this idea, in a similar population of subjects, the degree of dACC hypoactivation in a task in which correct performance was monetarily remunerated correlated with frequency of cocaine use, whereas degree of rostroventral ACC (extending to mOFC) hypoactivation correlated with task-induced craving suppression83
. There was an inverse association of these PFC ROIs with cue reactivity in the midbrain in cocaine-addicted subjects but not in control subjects, which implicates these ACC subdivisions in the regulation of automatic drug responses84
It should be noted that in the studies described above, subjects were not asked to choose between monetary rewards. We predict that choice would similarly follow a linear function (choice of higher over lower reward) in healthy controls more so than in addicted individuals, who we expect to show less flexibility in choice (choosing drug over other reinforcers), particularly during craving and bingeing. Studies that allow subjects to choose between reinforcers have mostly been conducted in laboratory animals. These studies have shown that, when given the choice, previously drug-exposed animals choose the drug over novelty85
, adequate maternal behaviour86
and even food87–89
, indicating that drug exposure can decrease the perceived value of natural rewards, even those that are needed for survival. In a recent human neuroimaging study in which subjects could win cigarettes or money, occasional smokers were more motivated to obtain money than cigarettes, whereas dependent smokers made similar efforts to win money or cigarettes90
. A similar group by reward interaction was observed in the right OFC, bilateral DLPFC and left ACC, such that in the occasional smokers these regions showed higher activity to stimuli predicting an increasing monetary reward than to stimuli predicting a cigarette reward, whereas the dependent smokers showed no significant differences in such anticipatory brain activity. These regions also showed higher activation to money in the occasional than in dependent smokers90
These results, together with behavioural results on neuropsychological tests in cocaine-addicted individuals91,92
(see also BOX 2
), contribute to our understanding of how relative reward preferences may change in addiction such that preference for the drug competes with (and sometimes exceeds) preference for other reinforcers, with a concomitant decrease in the ability to assign relative values to non drug-related rewards.
Several studies that are reviewed above compared PFC responses to non-concern-specific yet emotionally arousing stimuli with responses to concern-related (for example, drug-related) cues25,26,28,46,47
(Supplementary information S3 (table)
). The PFC was hyperactive in response to images from all emotional categories in alcohol-addicted subjects28
, the anterior PFC was hypoactive in response to pleasant pictures in heroin-addicted individuals26
, and in patients with eating disorders PFC responses to aversive pictures were normal46,47
. Thus, in contrast to our model’s predictions (), there were no differences in the PFC response between drug-related and affective yet non-drug related cues in any of these studies. This result, and the variability in the pattern of results, could be attributed to — among other factors — the small number of studies, differences between studies (such as sample sizes, the primary drug of abuse and duration of abstinence) and sensitivity of the measures used. Future studies would benefit from using event-related potential recordings or electroencephalography, which have much higher temporal resolution than fMRI or PET.
A clearer picture emerges when studies incorporate emotional processing into cognitive–behavioural tasks (Supplementary information S5 (table)
). For example, when required to empathize with a protagonist in a series of cartoons, each depicting a short story, methamphetamine-addicted individuals provided fewer correct answers than controls to the question “what will make the main character feel better?”93
. Compared to control subjects, the addicted individuals also showed hypoactivation in OFC (and hyperactivation in DLPFC) when answering this question. With the exception of one study in abstinent heroin-addicted individuals94
, other similar studies also reported differences between addicted and control groups in PFC responses to tasks requiring processing of emotional stimuli such as faces, words or complex scenes. For example, when men with alcohol addiction judged the intensity of five facial expressions, negative expressions were associated with lower activations in the left ACC but higher activations in the left DLPFC and right dACC compared to controls95
. In addition, compared to healthy controls, cocaine users showed ACC and dorsomedial PFC hypoactivations while performing a letter discrimination task during the presentation of a set of pleasant (versus neutral) pictures and hyperactivations in the bilateral DLPFC during the presentation of unpleasant (versus pleasant) pictures96
. Similarly, compared to healthy controls, marijuana smokers showed left ACC hypoactivations, and right DLPFC and inferior frontal gyrus hyperactivations in response to presentation of masked angry faces (versus neutral faces); right ACC responses positively correlated with frequency of drug use and bilateral ACC responses correlated with urinary cannabinoid levels and alcohol use97
. By contrast, the left dACC was hyperactive in methamphetamine-dependent subjects compared to controls when judging emotional expression on faces in an affect matching task (versus judging the shape of abstract figures) and this was associated with more self-reported hostility and interpersonal sensitivity in the addicted subjects98
Taken together, these studies indicate that the DLPFC is mostly hyperactive during emotion processing in addicted individuals compared to control subjects, especially for negative emotions. The ACC shows mixed results, although with more studies showing hypoactivity than hyperactivity. It is possible that the DLPFC hyperactivity may be compensating for the ACC hypoactivity, which would explain the lack of difference in task performance between drug abusers and healthy controls in most of these studies. Disadvantageous and/ or impulsive behaviours may be observed during greater emotional arousal challenges such as stress, craving or more difficult tasks. Clearly, the roles of these regions in relation to the proposed model () need to be better understood. It is possible that, by prematurely recruiting higher-order PFC executive function (mediated by the DLPFC), negative emotional arousal enhances risk for drug use in addicted individuals, particularly in situations that place additional strain on the limited cognitive control resources. This interpretation is consistent with the competition between drug and non drug-related processes and between ‘cold’ and ‘hot’ processes in the model ().
Although several of the above studies used negatively valenced stimuli, a lingering question is whether altered sensitivity to non-drug reinforcers in addicted individuals also applies to negative reinforcers such as money loss. Studies in animals show that ‘addicted’ subjects manifest persistent drug seeking even if the drug is associated with receiving an electric shock99
. In humans, hypoactivation in the right ventrolateral PFC in smokers during monetary loss, and in gamblers during monetary gain, have been reported100
(Supplementary information S5 (table)
). Although more studies are clearly needed, the implication of reduced sensitivity to negative reinforcers in addiction has practical implications as, in addition to positive reinforcers (such as vouchers and privileges), negative reinforcers (such as incarceration) are increasingly being used in the management of drug abusers. Interventions could be optimized by selecting the most effective type and dose of reinforcer. Future studies could also help to ascertain whether addicted individuals may resort to taking drugs because they are easily bored, frustrated, angry or fearful, perhaps as a result of altered PFC functioning. Low threshold for experiencing any of these emotions, or the inability to sustain goal-directed behaviour (for example, completing a boring task) when experiencing these emotions, may be associated with impaired inhibitory control (that is, enhanced impulsivity) as reviewed below. In cocaine-addicted individuals, PFC activity habituates prematurely to repeated presentation of an incentive sustained attention task101
, which could be a measure of compromised sustainability of effort and result in inadequate engagement in treatment activities.