Published in final edited form as:
Addictions are commonly presaged by problems in childhood and adolescence. For many individuals this starts with the early expression of impulsive risk-taking, social gregariousness and oppositional behaviors. We propose here that these early diverse manifestations reflect a heightened ability of emotionally salient stimuli to activate dopamine pathways that foster behavioral approach. If substance use is initiated, these at-risk youth can also develop heightened responses to drug-paired cues. Through conditioning and drug-induced sensitization, these effects strengthen and accumulate, leading to responses that exceed those elicited by other rewards. At the same time, cues not paired with drug become associated with comparatively lower dopamine release, accentuating further the difference between drug and non-drug rewards. Together, these enhancing and inhibiting processes steer a pre-existing vulnerability toward a disproportionate concern for drugs and drug-related stimuli. Implications for prevention and treatment are discussed.
An integrative neurodevelopmental model of substance use disorders
Drug addiction is the most prevalent neuropsychiatric disorder affecting society today. The social, medical and economic costs are enormous, with drug use contributing to 12% of deaths worldwide  and costing the U.S. government alone an estimated $400 billion per year [2–3].
Because only a minority of people who try drugs of abuse develop a substance use disorder (SUD), attempts have been made to identify predisposing neurobiological features. One long considered hypothesis is that increased susceptibility reflects preexisting perturbations in the mesolimbic dopamine system . Still debated though is whether this perturbation ultimately expresses itself as a decrease in dopamine activity, as in opponent-process and reward deficiency models [5–6], or heightened dopamine activity, as in incentive sensitization models [7–8]. The present neurodevelopmental model integrates each of these features. It recognizes a role for both hypo- and hyper-activity in mesolimbic dopamine systems, and outlines how each might become particularly pronounced in individuals at risk.
As summarized below, converging evidence from studies in human adolescents, young adults, and laboratory animals suggests that youth exhibiting heightened dopamine responses to emotionally intense stimuli are at increased susceptibility to engage in a wide range of impulsive, reward-seeking behaviors. Although these behaviors may initially target diverse non-drug stimuli, the initiation of drug use steers the heightened dopamine reactivity toward drug related cues, leading to drug conditioning and sensitization. These effects further enhance brain dopamine responses to the drugs and drug-paired cues, thereby augmenting the attentional focus of at-risk individuals on these stimuli and obtaining the drug. Because non-drug paired cues simultaneously become associated with comparatively lower dopamine responses, the overall result is a narrowed behavioral repertoire, setting the stage for progressively more frequent drug taking and a SUD.
This model represents a departure from single factor theories of drug abuse (Table 1). By incorporating both hypo- and hyper-dopamine activations, and combining this with identifiable predisposing factors, the present neurodevelopmental model provides a more comprehensive accounting of the addiction process. It is also, we propose, better positioned to inform the development of more effective therapeutic strategies.
Increased impulsive reward-seeking and dopamine responsivity prior to drug use
A recent series of adoption, twin, and longitudinal follow-up studies have supported a strikingly consistent conclusion: many SUDs reflect the outcome of an ‘externalizing’ trajectory characterized by risky thrill-seeking, social gregariousness, and oppositional tendencies in childhood and adolescence [9–19]. The core processes underlying these predispositions are thought to include over- and under-sensitivity to reward and punishment related cues, respectively [20–22]. For example, adolescents with high externalizing traits make risky choices, preferring high frequency rewards even when the losses are higher [23–25].
Marked individual differences in substance use are also seen in laboratory animals, and not all readily develop drug self-administration behaviors . One of the best-described predictors of susceptibility to acquire drug self-administration is a greater tendency to explore novel environments [26–29]. Among those animals that acquire drug self-administration, only a subset will transition to compulsive use, as defined by willingness to work more for the drug, endure aversive events to obtain it, and persist in drug seeking for much longer than average [30–31]. These “compulsive” drug-using rats are distinguished by high novelty preference and forms of impulsivity, such as premature responding to cues .
The behavioral traits that predict drug use behaviors co-vary with the tendency to engage with other rewarding stimuli and individual differences in dopamine cell responsiveness. In rats, high dopamine cell firing at baseline and release in response to diverse challenges predict greater novelty exploration [29,33], greater sugar feeding [29,34], more incentive learning , and the more rapid acquisition of drug self-administration [4,29,36–38]. The evidence is more than just correlational. Dopamine agonists increase premature responses during tests of impulsivity and a wide range of situation dependent reward-seeking behaviors including drug seeking (Box 1).
Dopamine and reward
Animal studies indicate that risky, reward-seeking behaviors are potently influenced by dopamine. Different components of these behaviors can be anatomically dissected. The best studied is the willingness to approach and sustain effort to obtain a reward, behaviors that are closely influenced by dopamine transmission in the ventral striatum, amygdala, and anterior cingulate [7–8,39–44]. Dopamine also affects the tendency to prematurely respond to reward cues , reflecting effects in the striatum , the willingness to tolerate delay for a larger reward, reflecting effects in the amygdala and orbitofrontal cortex [42–43,47], and executive control engagement with the task, reflecting effects in the orbitofrontal cortex . The weight of evidence suggests that dopamine is not closely related to pleasure [7,48].
In humans too individual differences in externalizing behaviors may be related to differences in dopamine responsiveness. In young healthy adults, greater striatal dopamine responsiveness co-varies with novelty seeking [49–50] and other impulsivity related traits [50–52]. In fMRI studies, similar results are seen. The greater the striatal responses to monetary reward, the greater the tendency to risky behavior [53–55]. The greater the striatal response to monetary reward anticipation, the higher the positive affective response scores . The greater the striatal response to cues paired with erotic images, the more likely these cues will be chosen two months later . And the greater the striatal responses to images of food and sex, the greater the weight gain and sexual activity at follow-up six months later .
The above associations in humans are thought to reflect causal effects since manipulating dopamine transmission alters many of the same processes [59–61]. Lowered dopamine transmission disrupts corticostriatal functional connectivity , top-down regulation by the cortex and the ability of reward related cues to activate the striatum [63–64]. These neurophysiological effects are associated with a decreased behavioral tendency to preferentially respond to rewards [65–67], and a decreased willingness to sustain effort to obtain rewards, including alcohol , tobacco  and money . Elevated dopamine function, in comparison, increases the ability of reward related cues to guide behavioral choices , diminishes the ability to differentiate between high and low value rewards , and induces steeper temporal discounting, a form of impulsivity defined by preference for immediately available small rewards over larger, more distal ones . In clinical populations, patients with schizophrenia – considered a hyper-dopamine disease – have very high rates of substance use problems  while those with Parkinson’s disease exhibit, if anything, decreased rates of substance abuse . Indeed, administering Parkinson’s patients dopamine agonist medications can induce a dysregulation syndrome characterized by various impulse-control problems, including pathological gambling, hyper-sexuality, and substance abuse .
Hyper- and hypo-dopamine activity following the initiation of drug use
Once drug use begins, some of the effects can become sensitized; i.e., previously ineffective low doses can now produce a response and previously effective doses elicit larger responses. In laboratory animals, repeated drug administration regimens can lead to progressive increases in drug-induced behavioral activation, greater willingness to sustain effort to obtain drug reward, and greater drug-induced dopamine release [7–8].
The conditions most likely to produce sensitization resemble early drug use patterns in humans: multiple exposures to moderate to high doses taken days apart in the presence of the same environmental stimuli. When these conditions have been simulated in human research, drug-induced sensitization has been demonstrated including greater drug-induced dopamine release and greater energizing effects [74–76]. This noted, even under these conditions, not all subjects exhibit the augmented responses. In rats, sensitization is more likely to develop in those that exhibit high reactivity to novel environments [27,33]. In humans, dopamine sensitization was greater in those with high novelty seeking scores .
Repeated drug administration can also lead to conditioned effects; i.e., environmental stimuli paired with the drug can come to elicit many of the same effects as the drug itself, including behavioral activation, dopamine release and reward-seeking [77–81]. The optimal conditions for producing these conditioned effects are the same as those for eliciting sensitization. Moreover, individual differences are also apparent . Finally, high novelty exploring rats engage more actively with cocaine cues, and are more susceptible to the cue-induced reinstatement of drug-seeking following an extinction procedure .
In humans too, cues paired with drug use can come to elicit many of the same effects as the drugs, including increased reward-seeking , conditioned place preferences [85–86], greater drug-induced drug craving , and dopamine pathway activation [88–89]. Individual differences in cue-induced dopamine  and craving responses are seen , and some evidence suggests that this could reflect a trait .
The cue-induced effects appear to be particularly marked in subjects at risk for addictions. In heavy drinkers at risk for alcohol use disorders, alcohol related cues induce a heightened electroencephalogram (EEG) P300 signal, an index of motivational salience . In fMRI studies, high externalizing adolescents show greater responses to monetary reward notification than control subjects in the ventral striatum . Similarly, compared to healthy controls, subjects with a family history of alcohol use disorders exhibit larger responses to alcohol associated cues in the nucleus accumbens and other aspects of the mesocorticolimbic circuit [91–93]. Indeed, in a large study of heavy drinkers (n=326), the greater the severity of alcohol use problems, the greater the alcohol cue-induced striatal activation [94–95]. Finally, intriguing preliminary evidence suggests that a subpharmacological taste of beer leads to significant striatal dopamine responses in participants with a family history of alcohol use disorders, but not in low risk drinkers .
The presence vs. absence of drug related cues and contexts can modify the readiness to respond to other events [76,97–99]. If a natural reward is presented in a place previously paired with drug, the animal will exhibit invigorated engagement with this natural reward [82,100]. If, more typically, drug cues are presented in association with the possibility of receiving drug, drug-seeking behaviors are fostered [77,81,101]; if the drug is administered, the expression of dopamine  and behavioral sensitization is enabled [102–103]. Conversely, cues explicitly paired with the absence of drug reward can have potent inhibitory effects, actively decreasing dopamine release , behavioral activation [97,102–103,105–106] as well as drug-taking and reinstatement [107–108].
The effects of stimuli explicitly paired with the absence of drug reward are less well studied in humans. However, recent evidence suggests that inhibitory processes can be engaged. For example, when non-dependent smokers were presented with cigarette cues, craving scores increased significantly above baseline; presentation of cues explicitly paired with the absence of cigarettes, in comparison, significantly decreased craving below baseline . Evidence of these diminished effects can be seen in brain also. High-risk subjects who have begun substance use exhibit smaller EEG P300 responses to positive non-substance related cues such as erotica than drug related cues . fMRI studies support the same conclusion: compared to healthy controls, at-risk subjects exhibit smaller striatal-limbic responses to various low non-drug cues, perhaps particularly those with low immediate salience [110–112; cf, 55].
The presence vs. absence of drug related cues might also affect the readiness of dopamine cells to respond in humans. For example, when non-dependent stimulant drug users ingested cocaine in the presence of drug related cues (immersed in the familiar microenvironment of preparing and inhaling cocaine powder) , the greater the lifetime history of stimulant drug use, the greater the drug-induced striatal dopamine response. In comparison, in non-dependent stimulant users tested in the absence of drug-related stimuli, greater lifetime histories of substance use were associated with smaller drug-induced striatal dopamine responses  (Figure 1). One interpretation of these results is that the absence of drug related cues dampens dopamine cell reactivity (Figure 2).
Together, the above studies suggest that low dopamine transmission in the absence of drug-related cues can result from two processes. The first is a passive process in which dopamine transmission is low as compared to responses seen when drug cues are present. The second is an active process, reflecting conditioned inhibition (Box 2). Moreover, not only can these non-drug cues usher in a period of low dopamine activity and motivation, their lack of attractiveness cannot compete with the pull of drug-paired cues. These effects may also have implications for behavior during withdrawal, and, indeed, the heightened susceptibility to seek and use drugs when in drug withdrawal may well reflect the same processes. Just as deprivation states can enhance the incentive value of natural reward cues, such as food , compelling evidence suggests that drug seeking observed during drug withdrawal may also reflect the heightened incentive salience of drug cues rather than avoidance of withdrawal [117–119]. Thus, drug use during withdrawal may reflect elements of positive rather than negative reinforcement processes. In these ways, cues unpaired with drug may be critical for the development of two overarching features of SUD: the progressive narrowing of interests toward drug related cues and drug taking and a diminished interest in pursuing the non-drug related goals necessary to thrive.
Environmental cues and reward
Imagine you are walking up a steep hill. If past experience has taught you that an enticing reward is at the top, your motivation to continue will be high, and cues indicating that the reward is forthcoming will augment and sustain your drive. These motivational states are closely related to changes in dopamine transmission; i.e., reward-paired contexts increase the readiness of dopamine cells to burst fire in response to discrete reward-paired cues [44,98,115]. In comparison, environments explicitly paired with the absence of reward can acquire the properties of a conditioned inhibitor  and the ability to actively inhibit dopamine readiness and the ability to respond to rewards and reward-related cues [76,104]. Together, this combination of effects produces strong preferences for drug-paired environments and cues, steering individuals away from non-drug related activities and events.
Two very recent studies suggest that subjects at high risk for SUDs might be particularly susceptible to these effects (Figure 3). First, a distinctively high dopamine response was seen in impulsive substance users at elevated risk for addictions, as compared to low risk users, when they were tested with drug cues present (alcohol ingested with the beverage’s sight, smell, taste and touch) . Second, and in striking contrast, exceptionally low dopamine release was observed in impulsive substance users at elevated risk for addictions when they were tested without drug cues present (d-amphetamine tablets hidden in nondescript gelcaps) . In both of these studies, the group differences persisted after controlling for lifetime substance use. Indeed, in these high-risk users, the dopamine responses in the absence of drug-related cues were significantly lower than those seen in low risk subjects matched for personal drug use histories . Such observations raise the possibility that, in these high-risk populations, conditioned control over the response to rewards is developing faster or more extensively. Together, the findings reviewed here suggest that the combination of drug-induced sensitization, conditioning, and individual differences in susceptibility to these effects could come to steer at-risk youth toward progressively more frequent drug use, setting the stage for a SUD.
Implications for prevention and treatment
Unlike single factor views of addiction that focus on either hyper- or hypo-mesolimbic dopamine activations, the integrative model proposed here combines both features, thus providing a novel neurobiological starting point for intervention strategies, including prevention (Box 3). Recent work gives reason for optimism. For example, externalizing adolescents given impulse-control training exhibit fewer substance use problems at two-year follow-up .
Dopamine and impulsive behavior
The relation between impulsive behaviors, heightened dopamine release, and greater susceptibility to substance abuse can propagate across generations. In addition to propagation via heritable traits, impulsive rodents exhibit less maternal care , leading to greater impulsivity, reward cue sensitivity, dopamine release, and drug self-administration in their offspring [122–124]. In a natural environment, these animals may also be more likely to come in contact with adverse events. These stressors also induce dopamine release, and can lead to long lasting behavioral and dopaminergic cross-sensitization to drugs of abuse [125–127], aggravating further the pre-existing tendencies. The same effects may be occurring in humans too. Indeed, children growing up in families characterized by externalizing behaviors are at elevated risk for stress, trauma and neglect, putting them at even higher risk for SUDs .
It remains speculative whether the processes described above (externalizing traits, alternating hyper- and hypo-dopamine function) are relevant once a severe addiction has developed. On the one hand, drug related cues consistently induce striatal activations in people with current addictions, these activations are larger than those seen in healthy controls, and individual differences in the magnitude of drug cue-induced dopamine responses correlate with craving . Based on these observations, we propose that it is premature to reject elevated dopamine transmission as a target for treatment.
At the same time, individuals with current SUDs are also consistently reported to have decreased striatal dopamine release, compared to healthy controls, when challenged with amphetamine . Two points are of interest here. First, in all but one of these studies , amphetamine was administered without drug related cues being present (Box 4). Second, not all individuals with current SUDs exhibit diminished amphetamine-induced dopamine release when tested in the absence of drug-paired cues. This differential response appears to have clinical significance: the roughly 50% of subjects who exhibit a normal dopamine response under these conditions are also better responders to monetary reinforcement-based behavioral therapies, raising the intriguing possibility that patients who can express a dopamine response in the absence of drug related cues are also better able to learn new reward related behaviors [138–139]. It remains unclear whether the low dopamine release seen in the other substance dependent patients reflects the absence of drug-related cues, differential vulnerability to neurotoxic effects of extensive substance abuse, a pre-existing trait, dopamine D2 pre- and post-synaptic receptor super-sensitivity, or some combination of these factors. Irrespective, Martinez and colleagues  intriguingly noted that these individuals might display a biomarker indicating that they would benefit better from behavioral therapies if they were pre-treated with agents that increase presynaptic dopamine function, such as L-DOPA .
Dopamine and “behavioral addictions”
Evidence of augmented dopamine responses in the presence of addiction related cues has been seen consistently in people with ‘behavioral addictions’. Compared to healthy controls, people with non-substance related ‘behavioral addictions’ (Pathological Gambling, Binge Eating Disorder) exhibit evidence of exaggerated striatal dopamine responses to food, monetary rewards and undisguised amphetamine tablets [131–134; cf, 135]. The greater the elicited dopamine release, the more severe the clinical problems [132,134,136–137]. Low dopamine release has not been reported in these populations. However, the fMRI pathological gambling literature reports both increases and decreases in striatal activations, and these differential responses appear to reflect in substantial part the presence vs. absence of explicit gambling related cues .
Other dopamine based treatment strategies are also under development. Dopamine D1 and D2 receptor ligands have shown little efficacy but D3 receptor antagonists have tentatively shown potential . Other receptor subtypes (D4, D5) have yet to be examined. Finally, since addicts appear to experience dopamine spikes in response to drug cues and dips when the cues are absent, dopamine modulators may provide a novel treatment consistent with the present model. The proposition is that these compounds will diminish the increases in dopamine that reinstate drug seeking without negating all dopamine transmission and producing a pervasive loss of interest .
The present model combines a neurodevelopmental perspective with evidence that the presence vs. absence of drug-related cues can come to regulate dopamine reactivity, directing motivational processes and setting the stage for progressively more frequent drug use and a SUD. This integrated perspective shows promise for guiding early intervention preventative strategies, and suggests that a fruitful direction for novel pharmacotherapeutic approaches could be to develop compounds that foster the ability to sustain interest in non-drug related activities. Strengthening the appeal of these goals may help those with SUDs steer away from drug related cues and attend better to ones necessary for healthy living.
- Addictions are commonly presaged by problem behaviors in childhood
- Susceptibility might reflect increased dopamine responses to salient events
- Drugs hijack dopamine responses, directing behavior preferentially toward drugs
- Non-drug events become less salient, and less able to activate dopamine
- Narrowed interests develop, setting the stage for frequent drug use and addictions
This review was made possible by grants from the Canadian Institutes for Health Research (MOP-36429 and MOP-64426, ML) and the National Institutes of Health (DA09397, PV).
Publisher’s Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.