High Impulsivity Predicts the Switch to Compulsive Cocaine-Taking

See allHide authors and affiliations

Science  06 Jun 2008:
Vol. 320, Issue 5881, pp. 1352-1355
DOI: 10.1126/science.1158136


Both impulsivity and novelty-seeking have been suggested to be behavioral markers of the propensity to take addictive drugs. However, their relevance for the vulnerability to compulsively seek and take drugs, which is a hallmark feature of addiction, is unknown. We report here that, whereas high reactivity to novelty predicts the propensity to initiate cocaine self-administration, high impulsivity predicts the development of addiction-like behavior in rats, including persistent or compulsive drug-taking in the face of aversive outcomes. This study shows experimental evidence that a shift from impulsivity to compulsivity occurs during the development of addictive behavior, which provides insights into the genesis and neural mechanisms of drug addiction.

Compulsive cocaine use has been hypothesized to result from a failure in top-down executive control over maladaptive habit learning (1, 2). In neural terms, this may reflect the diminishing influence of prefrontal cortical function, as behavioral control devolves from ventral to dorsal striatum (1). In behavioral terms, we predict that the development of addiction reflects a shift from impulsivity to compulsivity (3).

Human studies have implicated individual differences in different forms of impulsivity and sensation-seeking in vulnerability to drug use and abuse (46). However, whether the enhanced impulsivity observed in drug addicts (7, 8) predates the onset of compulsive drug use or is a consequence of protracted exposure to drugs has not been fully established. In addressing this issue experimentally, we operationalized these human traits in experimental animals as an inability to wait before performing an appropriate response, one phenotype of impulsivity (9) measured as premature responses in a five-choice serial reaction-time task (5-CSRTT) of sustained visual attention (10), as distinct from locomotor reactivity to a novel environment, a sensation-seeking phenotype (11). These animal models support the existence of a “vulnerable phenotype” that predisposes to drug addiction. Thus, outbred rats exhibiting high levels of novelty-induced locomotor activity, called high responder (HR), show increased sensitivity to the reinforcing effects of addictive drugs and self-administer lower doses of psychostimulants than low-responder (LR) littermates (11). Impulsivity, on the other hand, correlates with ethanol intake (12) and predicts instead the escalation of cocaine self-administration (10, 13), which may be more indicative of a necessary stage in the transition to compulsive drug-seeking. Although these studies have addressed the initiation of drug-taking, they have not captured the essential feature of addiction, namely, the persistence of drug-seeking in the face of negative consequences, a characteristic incorporated into recent animal models based on the DSM-IV criteria for substance dependence (14, 15). Therefore, we have used a model of addiction based on individual differences in compulsive cocaine use (14) to investigate the contrasting contribution of high impulsivity (HI) and high reactivity to novelty (HR) to the development of compulsive drug-taking.

In this model, we have operationally defined three addiction-like criteria in rats that correspond to those of the DSM-IV description of substance dependence (16), namely, (i) increased motivation to take the drug, (ii) an inability to refrain from drug-seeking, and (iii) maintained drug use despite aversive consequences [see (17) for details]. Thus, rats positive for none of the three criteria (zero-criteria rats) are resistant to addiction, whereas rats that have three addiction-like criteria (three-criteria rats) are considered “addicted,” and represent 15 to 20% of the population initially exposed to cocaine (14), a proportion that is similar to that observed in human populations (18).

To compare the propensity of HI and HR rats both to acquire cocaine self-administration and to make the transition to compulsive cocaine-taking, we first identified HI and LI rats in the 5-CSRTT (10), then HR and LR rats in a novelty-induced locomotor activity procedure (11). Subsequently, we compared the propensity of these different groups to acquire cocaine self-administration and to develop the three addiction-like criteria following protracted self-administration (17). HI and LI rats did not differ in their novelty-induced locomotor activity; conversely, HR and LR rats were not impulsive (Fig. 1, A and B). As predicted (11), HR rats were more prone to acquire cocaine self-administration than LR rats, showing an upward shift in the cocaine dose-response curve (Fig. 2). However, HI rats did not differ from LI rats in their acquisition of cocaine self-administration.

Fig. 1.

Impulsivity and novelty-induced locomotor activity: two distinct phenotypes. On two baseline days (b), premature responses in the 5-CSRTT were measured. (A and B) During long intertrial intervals (LITIs), HI rats showed more premature responses than LI rats (Group: F3,36 = 14.4, P < 0.01; Schedule: F8,288 = 130.22, P < 0.01; Schedule × Group: F24, 288 = 7.01, P < 0.01) (***P < 0.001) (A) and HR (P < 0.01) or LR rats (P < 0.05) (B). HR rats did not differ from LR rats or from LI subjects (B). (C and D) HR rats were more reactive to novelty than LR rats for the first 30 min (left histogram, inset) or the total duration of the session (right histogram) (Group: F3,35 = 12.17, P < 0.01 and F3,35 = 17.63, P < 0.01, respectively; Group × Time: F6,70 = 1.26, not significant, and F30,350 < 1, P < 0.001, respectively). (Inset) HI and LI subjects differed from both HR (P < 0.01) and LR rats (#, P < 0.01) but never from each other. *Comparison with HR: *P < 0.05, **P < 0.01, ***P < 0.001. (D) Black and gray dotted lines represent the average premature responses during the last two long intertrial intervals for HI and LI rats, respectively.

Fig. 2.

Novelty-induced locomotor activity predicts the propensity to acquire cocaine self-administration. (A) HR rats showed an upward shift of the cocaine dose-response curve compared with LR littermates (Group: F1, 16 = 4.9, P < 0.05; Dose: F6, 96 = 11.73, P < 0.01; and Group × Dose: F6, 96 = 4.39, P < 0.01). HR rats infused more cocaine at the lowest three doses than vehicle (P < 0.01). (B) HI and LI subjects did not differ in their number of self-administered cocaine infusions (Group: F1,16 < 1; Dose: F6, 96 = 10.79, P < 0.01; Group × Dose: F6, 96 < 1).

After 40 days of cocaine self-administration, we measured the three addiction-like behaviors in a cohort of 23 rats (17), so that each rat was defined as showing none or one, two, or three of these behaviors (table S2), as well as an addiction score, calculated as the sum of the standardized scores of each of the addiction-like criteria (17). Thus, rats with zero, one, two, or three criteria were linearly distributed along an addiction scale, corresponding operationally to the Addiction Severity Index (ASI) in humans (17, 19). On this scale, three-criteria rats had scores higher than all the other groups (Fig. 3A), especially when compared with zero-criteria animals, from which they differed for each of the addiction-like behaviors (fig. S1). Only the zero-criteria rats had highly negative addiction scores.

Fig. 3.

HI rats closely resemble three-criteria rats. After protracted self-administration, rats with zero, one, two, or three criteria were identified. (A) When ranked on a linear addiction scale (R2 = 0.99, Group: F3,19 = 34.43, P < 0.01), three-criteria rats had addiction scores (2.8 ± 0.6) above the standard deviation (2.1), and higher than all the other groups (versus zero- and one-criteria rats: ♦, P < 0.01, versus 2 criteria rats: #, P ≤ 0.05). (B) HI rats displayed higher addiction scores than LI rats (F1,9 = 7.55, *P < 0.05), whereas HR rats did not differ from LR rats. Only HI rats did not differ from three-criteria rats for their addiction score (F5,30 = 10.13, P < 0.01) and displayed higher addiction scores than zero-criteria (P < 0.01) and HR rats (P < 0.05). HR, LR, and LI rats did not differ from zero-criteria rats.

Although reactivity to novelty predicts the vulnerability to acquire cocaine self-administration, it is high impulsivity that predicts the transition from controlled to compulsive cocaine-taking. HI rats displayed higher addiction scores than LI rats whereas, in marked contrast, HR rats did not differ from LR rats (Fig. 3B). LI, HR, and LR rats were represented mainly in the zero- and one-criteria populations; however, HI rats were largely represented in the two- and three-criteria populations. Additionally, only HI rats were more frequently represented in the three-criteria group than in the zero-criteria group (table S1).

A factor analysis revealed that impulsivity and addiction-like behavior are explained by the same factor that was itself orthogonal to reactivity to novelty, thereby identifying an impulsivity-addiction construct (fig. S2). Thus, HI rats did not differ from three-criteria rats in any of their addiction-like behaviors (Fig. 4A and fig. S3).

Fig. 4.

Impulsivity predicts the transition to compulsivity. (A) HI rats receiving punishment (17) continued compulsive cocaine self-administration (SA). HI rats (n = 5) displayed higher resistance to punishment than LI rats (n = 6) (F1, 9 = 12.79, P < 0.01), whereas HR (n = 5) rats did not differ from LR rats (n = 5). When compared with zero- and three-criteria rats for their resistance to punishment (Group: F5,30 = 10.13, P < 0.01), only HI rats were similar to three-criteria rats; they showed greater resistance to punishment than zero-criteria, LI, and HR rats (P < 0.05). LI and HR rats differed from three-criteria, but not from zero-criteria, rats (A). **P < 0.01. (B) Impulsivity predicts compulsive cocaine self-administration (R = 0.42, P < 0.05). Gray and black shadings represent LI and HI rats, respectively.

More specifically, the high addiction score of HI rats derived from the development of compulsive cocaine self-administration. HI rats displayed greater resistance to punishment of the drug-taking response than did LI, HR, and LR rats (Fig. 4A), and at the population level, correlational analysis revealed that impulsivity predicts compulsivity (Fig. 4B). However, HI rats did not differ from LI, HR, or LR rats in their total intake of cocaine (fig. S3); therefore, the development of compulsive cocaine-taking observed specifically in the highly impulsive rats cannot be attributed to differential exposure to cocaine. Because clinical investigations generally compare addicted subjects with drug-naïve controls, we analyzed whether animals vulnerable and resistant to addiction differed in their impulsivity and locomotor reactivity to novelty before any exposure to cocaine. This analysis showed that three-criteria rats were more impulsive, but not more reactive to novelty, than zero-criteria rats before cocaine self-administration (fig. S4).

These data allow us to identify one variety of impulsivity, measured as an inability to wait and sample predictive stimuli before responding (20), as a key behavioral marker specific for the vulnerability to progress to compulsive cocaine use, the hallmark of addiction. Our results are in accord with observations that (i) highly impulsive humans are overrepresented in drug-addicted populations (21), (ii) impulsivity or sensation-seeking may predate compulsive drug use (22, 23), and (iii) there is a high comorbidity between drug addiction and disorders characterized by impulsive behavior, such as attention deficit–hyperactivity disorder (ADHD) (21).

The results indicate that the relation between high impulsivity and addiction-like behavior is completely independent of the initial propensity to acquire cocaine self-administration (fig. S2) (17), an observation consistent with the demonstration that impulsivity is unrelated to the subjective effects of oral amphetamine administration (24). Instead, this early vulnerability to cocaine's reinforcing effects was predicted by high locomotor reactivity to novelty. Our observations further suggest that the subjective and behavioral responses to cocaine during initial exposure to the drug do not determine the subsequent progression to addiction, as might perhaps have been previously suspected (11).

Our study also provides experimental evidence that high levels of impulsivity can antedate the onset of compulsive drug use and, thereby, emphasizes the importance of preexisting impulsivity observed in addicts (2, 4, 7). Moreover, by demonstrating the link between impulsivity and compulsivity in the development of addiction, these data provide a major impetus for investigating the neurobiological mechanisms underlying this transition. One candidate is the apparent devolution of control over drug-seeking behavior from the ventral to the dorsal striatum (25), which has been shown to depend on the cascading, serial ascending circuitry that links these striatal domains via its regulatory dopaminergic innervation arising in the midbrain (26, 27). This hypothesis is further supported by the observation that the early vulnerability to escalate cocaine intake shown by highly impulsive rats is predicted by low D2 and/or D3 dopamine receptor levels in the ventral, but not the dorsal, striatum (10). In contrast, chronic exposure to cocaine in monkeys (28) and drug-abusers (29) is associated with low D2 and/or D3 dopamine receptor availability, predominantly in the dorsal striatum.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4

Tables S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article