ARCHIVAL REPORT

Increased Impulsivity Retards the Transitionto Dorsolateral Striatal Dopamine Control ofCocaine Seeking

Jennifer E. Murray, Ruth Dilleen, Yann Pelloux, Daina Economidou, Jeffrey W. Dalley,David Belin, and Barry J. Everitt

Background: Development of maladaptive drug-seeking habits occurs in conjunction with a ventral-to-dorsal striatal shift indopaminergic control over behavior. Although these habits readily develop as drug use continues, high impulsivity predicts loss ofcontrol over drug seeking and taking. However, whether impulsivity facilitates the transition to dorsolateral striatum (DLS) dopamine-dependent cocaine-seeking habits or whether impulsivity and cocaine-induced intrastriatal shifts are additive processes is unknown.

Methods: High- and low-impulsive rats identified in the five-choice serial reaction-time task were trained to self-administer cocaine(.25 mg/infusion) with infusions occurring in the presence of a cue-light conditioned stimulus. Dopamine transmission was blocked inthe DLS after three stages of training: early, transition, and late-stage, by bilateral intracranial infusions of α-flupenthixol (0, 5, 10, or 15μg/side) during 15-min cocaine-seeking test sessions in which each response was reinforced by a cocaine-associated conditionedstimulus presentation.

Results: In early-stage tests, neither group was affected by DLS dopamine receptor blockade. In transition-stage tests, low-impulsive ratsshowed a significant dose-dependent reduction in cocaine seeking, whereas high-impulsive rats were still unaffected by α-flupenthixolinfusions. In the final, late-stage seeking test, both groups showed dose-dependent sensitivity to dopamine receptor blockade.

Conclusions: The results demonstrate that high impulsivity is associated with a delayed transition to DLS-dopamine-dependent controlover cocaine seeking. This suggests that, if impulsivity confers an increased propensity to addiction, it is not simply through a more rapiddevelopment of habits but instead through interacting corticostriatal and striato-striatal processes that result ultimately in maladaptivedrug-seeking habits.

Key Words: Cocaine, dopamine, drug addiction, goal-directed,habitual, striatum

I ncreasing evidence suggests that addiction results from theconvergence of various neurobiological adaptations in vulner-able subjects, eventually resulting in the loss of control over

maladaptive drug seeking (1–3). Exposure to addictive drugs,such as cocaine, not only impairs executive processes, resulting inimpulse control deficits and behavioral inflexibility (4), but it alsofacilitates the development of drug-seeking habits (3,5,6), therebyrendering instrumental actions that are resistant to their imme-diate consequences and motivational significance (6,7). Addictivedrugs trigger adaptations within corticostriatal circuitry, includingreductions in metabolic activity and D2 dopamine receptors, thatare initially restricted to the ventral limbic areas of the striatumand prefrontal cortex but eventually encompass the more dorso-lateral, associative and cognitive, territories of these structures (8–

Authors DB and BJE contributed equally to this work.

From the Department of Psychology (JEM, RD, DE, JWD, BJE); Behaviouraland Clinical Neuroscience Institute (JEM, RD, DE, JWD, BJE); Depart-ment of Psychiatry (JWD), Addenbrooke’s Hospital, University ofCambridge, Cambridge, United Kingdom; INSERM U1084-LNEC teamPsychobiology of Compulsive Disorders (DB), Universtié de Poitiers;INSERM European Associated Laboratory (JEM, DB, BJE), Poitiers; andthe Institute of Neuroscience de la Timone (YP), University of Aix-Marseille, Marseille, France.

Address correspondence to Jennifer E. Murray, Ph.D., Department ofPsychology, University of Cambridge, Downing Street, Cambridge, UK,CB2 3EB; E-mail: jem98@cam.ac.uk.

Received May 21, 2013; revised and accepted Sep 18, 2013.

0006-3223/$36.00http://dx.doi.org/10.1016/j.biopsych.2013.09.011

10). This progressive shift from limbic to cognitive corticostriatalnetworks that occurs over the course of addiction (11) takes placealongside a transition from the nucleus accumbens to thedorsolateral striatum (DLS) in the locus of control over drugseeking and taking (12) and the associated imbalance in fronto-striatal and striato-striatal functional coupling (13) displayed byformer and current addicted individuals.

Studies in animals have further demonstrated that this ventralto DLS shift in the control over drug seeking (14,15) is not onlyassociated with the development of habitual responding for thedrug as assessed by devaluation procedures (3,6) but also reflectsthe emergence of compulsive cocaine seeking (16). The latter, ahallmark feature of addiction (17), is predicted by the behavioraltrait of high impulsivity (18), which is associated with low D2/3dopamine receptor availability in the ventral striatum (19). Thishas led to hypotheses suggesting that impulsivity and habits,with their striatal dopaminergic substrates, interact during thedevelopment of cocaine addiction, but the neurobiological basisof this interaction is unknown. Neurocomputational learningtheory-based, actor-critic models of basal ganglia function (20)suggest that high impulsivity and its associated low D2 dopaminereceptor availability in the ventral striatum facilitate the transitionto DLS control over drug self-administration. However, we andothers have suggested that compulsive drug seeking in addictionmight instead result from weak inhibitory control over a ratherindependently established, drug-influenced, maladaptive incen-tive habit (4,21).

We therefore directly investigated whether high impulsivityinteracts with the recruitment of dopamine-dependent DLScontrol over cocaine-seeking behavior over an extended periodof cocaine self-administration. To do this, we investigated theeffects of bilateral infusions of the dopamine receptor antagonist

BIOL PSYCHIATRY 2013;]:]]]–]]]& 2013 Society of Biological Psychiatry

2 BIOL PSYCHIATRY 2013;]:]]]–]]] J.E. Murray et al.

α-flupenthixol into the DLS of rats identified as high (HI) and lowimpulsive (LI) in the 5-choice serial reaction-time task (5-CSRTT),on cue-controlled cocaine-seeking behavior at early, transitional,and late stages of training under a second-order schedule ofreinforcement for cocaine (22). Under these conditions we havepreviously shown that cocaine seeking becomes dependent upondopamine transmission in the DLS (14,18,23), and the functionalrecruitment of this dopaminergic mechanism is a neurobiologicalmarker of the emergence of drug-seeking habits (3,6).

Methods and Materials

SubjectsForty male Lister Hooded rats (Charles River Laboratories, Kent,

United Kingdom) weighing approximately 300 g on arrival werehoused as described previously (23). Experiments were conductedin accordance with the United Kingdom 1986 Animals (ScientificProcedures) Act.

5-CSRTTApparatus and Procedure. The 5-CSRTT apparatus has been

described in detail elsewhere (24,25) (Supplement 1). The trainingprocedure was identical to that previously described (18). Eachtraining session began with illumination of the operant chamberby a house light and the delivery of a food pellet in the magazine.Pushing open the magazine panel and collecting this pelletinitiated the first trial. After a fixed intertrial interval (ITI), a lightat the rear of one of the response apertures was brieflyilluminated. Responses in this aperture within a limited-holdperiod (5 sec) were reinforced by the delivery of a food pelletin the magazine (correct responses). Responses in a nonillumi-nated aperture were recorded as incorrect responses and werepunished by a 5-sec time-out period. Failure to respond withinthe limited-hold period counted as an omission and was likewisepunished. Additional responses in any aperture before foodcollection (perseverative responses) were recorded but notpunished. Responses made in any aperture before the onset ofthe target stimulus, or premature responses, were punished by a5-sec time-out period. Across training sessions, the ITI wasgradually increased, and the stimulus duration was graduallydecreased (25). Subjects were considered to have acquired thetask when accuracy was � 75% and omissions were fewer than20% while the stimulus duration was .5 sec with a 5-sec ITI.

After 2 weeks of stable responding, rats underwent three60-min challenge 7-sec ITI (long intertrial interval [LITI]) sessions,separated by baseline 5-sec ITI sessions (18,26). The LITIsmarkedly increase premature responding, thereby facilitatingthe identification of interindividual differences in impulsivity.The number of premature responses during LITI sessions providesan index of impulse control (18,19,24–26), which is used toidentify HI or LI rats. Subjects were ranked according to themean number of premature responses during the last two LITIsessions (10,18). Those with �20 or �50 premature responses

Figure 1. The timeline of self-administration experimentation. Subjects undebeginning behavioral training. There were five sessions of fixed-ratio 1 (FR1response requirement was increased across sessions to the mid-stage trainingbefore entering mid-stage testing. The response requirement was again incre(FR10:S). Rats were again tested after 15 training sessions from Days 32 to 46 od, day; FI, fixed-interval.

www.sobp.org/journal

were selected as LI and HI rats, respectively (n = 8/group) (FigureS1 in Supplement 1).

In addition, premature responses, magazine panel pushes, correctand incorrect responses, omitted trials, and collection latency (milli-seconds to collect the food pellet) were averaged across the baselinesessions preceding each of the last two LITI sessions to comparebaseline behavioral performance in LI and HI rats.

SurgeryRats then underwent standard intravenous and intrastriatal

surgeries under general anesthesia (Supplement 1). Cannulaewere implanted bilaterally 2 mm above the dorsolateral striatum(anterior/posterior�1.2, medial/lateral�3, dorsal/ventral-3 [15];AP and ML coordinates measured from bregma, DV coordinatesfrom the skull surface, incisor bar at �3.3 mm [27]).

DrugsCocaine hydrochloride (Macfarlan-Smith, Edinburgh, United

Kingdom) was dissolved in sterile .9% saline. α-Flupenthixol(Sigma Aldrich, Poole, United Kingdom) was dissolved indouble-distilled water. Drug doses are reported in the salt form.

Cocaine Self-AdministrationApparatus. Twelve standard operant conditioning chambers

described in detail elsewhere (15) were used (Methods inSupplement 1).

Procedure. The timeline of self-administration procedures isshown in Figure 1. Briefly, cocaine self-administration trainingsessions began 7 days after surgery. Cocaine (.25 mg/infusion;.1 mL/5 sec) was available under a fixed-ratio 1 (FR1) (continuousreinforcement) schedule of reinforcement in which one active leverpress resulted in an infusion and initiated a 20-sec timeout. Duringthat 20 sec, the cue-light (conditioned stimulus [CS]) above the activelever was illuminated, the house light was extinguished, and bothlevers were retracted. Pressing on the inactive lever was recorded toprovide an index of general activity but had no programmedconsequence. A maximum of 30 cocaine infusions was available atthis stage. Active and inactive lever assignment was counterbalanced.

After five training sessions under the FR1 schedule of reinforce-ment, the dose-dependent effects of striatal dopamine receptorblockade on early-stage cocaine seeking were tested. Bilateralinfusions of α-flupenthixol were made into the DLS. These 15-mintest sessions [FI15(FR10:S)] instituted a change in contingency inthat every active lever press resulted in a 1-sec light CS presenta-tion, and cocaine was only delivered on the first lever press afterthe 15-min interval (23). Thus, the early performance tests wereconducted before and were thus unaffected by self-administeredcocaine on these sessions, because they were explicitly assessed forcocaine seeking within the fixed interval rather than a fixed ratio.Each test session was immediately followed by a FR1 cocaine self-administration training session (30 reinforcers over 2 hours), andrats were given a training session between test days so as toconfirm and maintain a stable cocaine-taking baseline.

rwent intravenous catheter and central cannulae surgery a week before) training followed by early-acquisition testing. From Days 13 to 17, theschedule of FR10(FR4:S). Rats remained on that schedule for five sessionsased on Days 30 and 31 to the final second-order training schedule, FI15n the final schedule of reinforcement. Late-stage testing began on Day 37.

J.E. Murray et al. BIOL PSYCHIATRY 2013;]:]]]–]]] 3

After the tests evaluating the early performance of cocaineseeking, the response requirement was increased across the dailytraining sessions through the following schedules of reinforce-ment: FR1; FR3; FR5(FR2:S); FR10(FR2:S); then to FR10(FR4:S).Under each intermediate second-order schedule, completion ofthe unit schedule (given within parentheses) resulted in a 1-secCS light presentation; cocaine infusions and the 20-sec timeoutwere given only upon completion of the overall schedule.Therefore, for the transition-stage assessments, rats had beentrained under conditions that promote the association betweeninstrumental responding and conditioned reinforcers: contingentpresentations of the cocaine-associated CS occurred after4 responses (FR4:S); and cocaine was delivered on completionof the 10th set of four lever presses. Rats remained on thisschedule for five training sessions before beginning thetransition-stage cocaine-seeking tests. During each 15-min testsession with α-flupenthixol infusions in the DLS, every four activelever presses continued to result in a 1-sec light CS presentation,and cocaine was only delivered on the fourth lever press after the15-min interval [i.e., FI15(FR4:S)]. Thus, the transition-stage per-formance tests were again conducted before and were unaf-fected by daily self-administered cocaine. Each test session wasimmediately followed by an FR10(FR4:S) cocaine self-administration training session (30 reinforcers over 2 hours),and rats were given a training session between test days so asto confirm and maintain a stable cocaine-taking baseline.

After completing the tests evaluating cocaine seeking atthe transition stage, the response requirements were againincreased through daily training sessions across the followingschedules of reinforcement: FR10(FR6:S); FR10(FR10:S); and finallyto an overall fixed interval (fixed ratio) schedule of FI15(FR10:S)used in previous studies (23,28). During the final FI15(FR10:S)schedule, responding was maintained by contingent presentationof the cocaine-associated CS after 10 responses (FR10:S); cocainewas delivered on completion of the first 10 lever presses after theexpiration of each 15-min fixed interval. At this final stage, therewas a limit of five available cocaine infusions. Rats were trainedunder this FI15(FR10:S) schedule of reinforcement for 15 sessionsbefore the well-established, or late-stage, tests were conducted,in which the effects of α-flupenthixol infusions in the DLS wereagain assessed. The first interval (FI15) of the second-orderschedule provides a time period in which no cocaine has beenadministered, yet rats are actively seeking the drug. Two rats wereremoved before the final tests, due to faulty catheters. Rats weregiven at least one session of training under FI15(FR10:S) con-ditions between each α-flupenthixol infusion test to ensurebaseline stable baseline levels of responding.

Intrastriatal InfusionsFor all three testing stages, intrastriatal infusions (.5 μL/side) of

α-flupenthixol (0, 5, 10, and 15 μg/infusion in a counterbalanced,Latin-square order of treatment) were made with 28-gauge steelhypodermic injectors (Plastics One, Roanoke, Virginia) lowered tothe injection sites 2 mm ventral to the end of the guide cannulae(i.e., DV-5 mm). Bilateral infusions were made over 90 sec with asyringe pump (Harvard Apparatus, Holliston, Massachusetts) andwere followed by a 60-sec diffusion period before injectors wereremoved and obturators were replaced. Test sessions began5 min later.

HistologyAt the end of the experiment, histology was conducted as

described previously (23) (Supplement 1).

Statistical AnalysesPremature responses in the 5-CSRTT were analyzed with 2-way

analyses of variance (ANOVAs) with Session as the within-subjectfactor, and Group (HI or LI) as the between-subjects factor.Premature responses were then correlated with selected trainingmeasures from the 5-CSRTT, and significant correlations wereconfirmed with between-subject t tests.

Recruitment of DLS dopaminergic involvement in cocaineseeking was confirmed with a three-way ANOVA with Stage(early, transition, and well-established), Dose (0, 5, 10, and 15 μg),and Lever (active and inactive) as within-subject factors. Thedifferential recruitment of DLS dopaminergic involvement incocaine seeking between HI and LI rats was investigated with athree-way ANOVA with planned contrasts (29) with Session(weights on session 2 vs. session 1) and Dose (weights on dosesof 10 and 15 mg/side vs. vehicle) as the within-subject factors andGroup (HI or LI) as the between-subjects factor. Differencesbetween HI and LI rats for each stage were then investigatedwith ANOVA with Dose and Lever as within-subject factors.Significant interactions were analyzed further with Tukey’s hon-estly significant difference (HSD) tests. Significance was set atα ¼ .05.

Results

5-CSRTTRats selected as HI (n ¼ 8) in the 5-CSRTT displayed greater

sensitivity to increased ITI duration than the LI (n ¼ 8) rats assupported by increases in premature responses for the three LITItrials for the HI compared with the LI rats (Figure 2) (main effectsof Group: F1,14 ¼ 65.20, p � .001, Session: F14,196 ¼ 59.34,p � .001, and Group � Session interaction: F14,196 ¼ 25.44,p � .001). Post hoc analysis revealed that group differencesemerged as a result of lengthening the ITI (HSD ¼ 14.477).

Higher impulsivity (measured as the level of prematureresponses during the last two LITI sessions) was related to greateramounts of goal tracking (measured as panel pushes into themagazine) and latency to collect earned pellets as revealed by apositive relationship between premature responses and panelpushing during training (τ ¼ .481, p ¼ .010) (Figure 3A); this wasfurther confirmed by a follow-up t test comparing the number ofpanel pushes in HI and LI rats (t14 ¼ 2.36, p ¼ .033). Impulsivitywas not, however, related to motivation for the reinforce, asrevealed by both the lack of relationship between the number ofpremature responses and the latency to collect pellets after acorrect trial (τ ¼ �.211, p ¼ .259) (Figure 3B) and the absenceof a difference in this latter measure between HI and LI rats(t14 ¼ 1.14, p ¼ .273). Baseline behavioral measures recordedduring the training sessions immediately preceding LITI 2 and 3are shown in Table S1 in Supplement 1.

Histological AssessmentsAll rats had cannulae located bilaterally within the DLS

(Figure 4) (27).

Recruitment of DLS Dopamine Control over Cocaine SeekingProgressive recruitment of dopamine-dependent DLS proc-

esses in the control over well-established, habitual, cue-controlledcocaine-seeking behavior was observed from early to late stagetests as illustrated by the progressive increase in the effect ofbilateral intra-DLS α-flupenthixol infusions on active lever pressesduring the 15-min drug-free cocaine-seeking interval (Stage �

www.sobp.org/journal

Figure 3. Premature responses during the long inter-trial interval (LITI)sessions were correlated with magazine panel pushes (goal-tracking) (A)and latency to collect reinforcers (motivation) (B) during training sessions.High-impulsive rats showed higher levels of interaction with the maga-zine, but not more motivated to obtain the reward, than low-impulsiverats.

Figure 2. High-impulsive rats are characterized by a high number ofpremature responses made before the onset of the target stimulus duringthe long inter-trial intervals (LITIs) but not during baseline (BL) sessions.*Significant difference from low-impulsive rats during the same LITI.

4 BIOL PSYCHIATRY 2013;]:]]]–]]] J.E. Murray et al.

Dose � Lever interaction: F6,78 ¼ 3.50, p ¼ .004), confirming ourearlier results (15,23). Thus, although dopamine receptor blockadein the DLS was ineffective during the early stage of cocaineseeking (Figure 5A) (effect of Dose: F3,45 ¼ 1.03, p ¼ .389 andLever � Dose interaction: F3,45 ¼ 1.06, p ¼ .375), it dose-dependently reduced cocaine seeking when performed at thetransition stage (Figure 5B) (main effect of Dose, F3,45 ¼ 3.41, p ¼.025; and a Lever � Dose interaction, F3,45 ¼ 3.45, p ¼ .024). Posthoc analyses revealed that this effect was attributable to the 10-and 15-μg/side doses of α-flupenthixol (HSD ¼ 26.59). When cue-controlled cocaine seeking was well-established, bilateral DLS α-flupenthixol infusions resulted in an even more pronounceddecrease in cocaine-seeking responses measured during the 15-min drug-free interval (Figure 5C) (main effect of Dose: F3,39 ¼9.69, p � .001 and Lever � Dose interaction: F3,39 ¼ 9.01, p �.001). At this stage, all doses of α-flupenthixol significantlyreduced cocaine seeking relative to vehicle (HSD ¼ 40.30).

Impulsivity Is Associated with a Delayed Transition to DLSDopamine Control over Cocaine Seeking

The progressive recruitment of DLS dopamine control overcocaine seeking observed in the entire population was modulatedby impulsivity status. Thus, HI and LI rats displayed different time-courses in their sensitivity to DLS dopamine receptor blockade overthe transition from early to well-established, habitual, cue-controlledcocaine seeking (Session � Dose � Group contrasts: F1,12 ¼ 8.07, p� .05). Thus, whereas DLS α-flupenthixol infusions had no signifi-cant effect on active lever presses in HI (Figure 6A) and LI rats(Figure 6B) during the early seeking tests (main effects of Dose orDose � Lever interaction: Fs # 2.83, p $ .063), they dose-dependently decreased cocaine seeking in LI rats (Figure 6C) (maineffect of Dose: F3,21 ¼ 3.89, p ¼ .023, and a Dose � Leverinteraction: F3,21 ¼ 3.86, p ¼ .024) but not in HI rats (Figure 6D)(Fs � 1) during the transition seeking tests. Post hoc analysesrevealed that cocaine seeking in LI rats behavior was decreasedafter infusions of 10- and 15-μg/side doses of α-flupenthixol relativeto vehicle and inactive lever presses (HSD ¼ 40.62).

In the well-established seeking tests, after rats had been trainedto seek cocaine under the control of contingent presentations of

www.sobp.org/journal

the drug associated CSs, during the FI15(FR10:S) stage of thesecond-order schedule, responding was dose-dependentlydecreased by bilateral infusions of α-flupenthixol into the DLS inboth HI and LI rats. LI rats continued to display dose-dependenteffects of α-flupenthixol infusions into the DLS (Figure 6E), while thissensitivity to DLS dopamine receptor blockade now emerged in HIrats (Figure 6F) (main effect of Dose: F3,15 ¼ 5.23, p ¼ .011 andF3,21 ¼ 4.11, p ¼ .019, respectively, Dose � Lever interaction: F3,15 ¼5.20, p ¼ .012 and F3,21 ¼ 3.59, p ¼ .031, respectively). Thus, the 10and 15 μg/side doses of α-flupenthixol markedly reduced active-lever presses relative to vehicle such that significant differencesbetween active and inactive lever pressing were no longer observed(HSD ¼ 69.58 and HSD ¼ 55.62 for LI and HI rats, respectively).

Although a shift in the time course of the recruitment ofdopamine-dependent DLS control over cue-controlled cocaine seek-ing was observed between HI and LI rats, the two groups differedneither in the propensity to initiate cocaine self-administration overthe five FR1 acquisition sessions (main effect of Session: F4,56 ¼ 3.124,p ¼ .022 but no effect of Group: F1,14 ¼ 1.606, p ¼ .226, or Group �Session interaction: F � 1) nor in their performance to the increasingbehavioral demands associated with each stage of the establishmentof a second-order schedule of reinforcement for the drug. Indeed, no

Figure 4. Schematic representations of the localization of injection sitesin high-impulsive (A) and low-impulsive (B) rats with guide cannulaeplaced in the anterior dorsolateral striatum. Reprinted from Paxinos andWatson (27) with permission from Elsevier, copyright 1998.

J.E. Murray et al. BIOL PSYCHIATRY 2013;]:]]]–]]] 5

differences were observed in cocaine-seeking responses between HIand LI rats either during the five FR10(FR4:S) sessions that precededthe intermediate stage assessment (all Fs � 1) or during the FI15(FR10:S) sessions that preceded the late stage assessment (maineffect of Group: F1,12 ¼ 1.367, p ¼ .265, and Group � Sessioninteraction: F14,168 ¼ 1.167, p ¼ .305), despite an overall increase inactive lever presses over the sessions, indicative of the progressiveincrease in the influence of contingent presentations on CS oninstrumental cocaine-seeking responses over time (main effect ofSession: F14,168 ¼ 1.872, p ¼ .033).

Figure 5. Progressive recruitment of dopamine-dependent dorsolateralstriatum control over cue-controlled cocaine seeking. Active and inactivelever presses (�1 SEM) during (cocaine-free) tests of drug seeking withα-flupenthixol injections into the dorsolateral striatum of high- and low-impulsive rats combined at the early (A), transition (B), and well-established(C) stages of training. *Significant difference in active lever responding fromthe 0 μg test. �Significant difference between active and inactive leverresponses for each dose tested. FI, fixed-interval; FR, fixed-ratio.

Discussion

Cocaine-induced intrastriatal processes eventually resulting inDLS dopamine-dependent drug-seeking habits (3,14,15,23,30–31)are increasingly considered to be a pivotal mechanism during thedevelopment of addiction (16). Although impulsivity character-ized by low ventral striatal D2/3 dopamine receptor availability(19) has been identified as a key marker of the individualpropensity to switch from controlled to compulsive drug use(18), the ways in which impulsivity and its underlying neuralsubstrates interact with drug-induced intrastriatal adaptations areunknown. According to our earlier speculation (28) and acomputational model of addiction based on striatal function(20), the trait of high impulsivity and associated low dopamineD2/3 ventral striatal dopamine receptors (19) has been suggestedto facilitate drug-induced recruitment of DLS-dependent habitualcontrol over cocaine-seeking behavior. By contrast, integrativehypotheses suggest that addiction develops when the neuro-biological underpinnings of impaired executive, corticostriatal-dependent, inhibitory control, lying at the core of impulsivity, addto and converge with those associated with drug-induced intra-striatal shifts subserving the development of cue-controlled drug-seeking habits (6,7,21,32,33).

The findings in the present study support the latter view byproviding evidence that increased impulsivity does not facilitate

or accelerate the progressive recruitment of dopamine-dependent DLS control over behavior that has been shown tounderlie both drug-seeking habits and compulsive cocaine

www.sobp.org/journal

Figure 6. Delayed transition to dorsolateral striatum control over cocaine-seeking behavior in high impulsive rats. Active and inactive lever presses (�1SEM) during (cocaine-free) tests of drug seeking with α-flupenthixol injections into the dorsolateral striatum of low- and high-impulsive rats at the early(A, B, respectively), transition (C, D, respectively), and well-established (E, F, respectively) stages of training. *Significant difference in active leverresponding from the 0 μg test. �Significant difference between active and inactive lever responses for each dose tested. FI, fixed-interval; FR, fixed-ratio.

6 BIOL PSYCHIATRY 2013;]:]]]–]]] J.E. Murray et al.

seeking (3,6,15,16,23). Instead, high impulsivity was associatedwith a delay in striato-striatal neuroadaptations leading to theprogressive devolution of control over cocaine seeking to DLSdopamine-dependent processes. This thereby indicates that theinteraction between impulsivity and cocaine-induced recruit-ment of dopamine-dependent dorsolateral striatal control overbehavior underlying the eventual transition to compulsive drug

www.sobp.org/journal

seeking (16) might depend upon interactive, co-occurring corti-costriatal and striato-striatal processes. It might therefore bespeculated that compulsive drug seeking arises from the devel-opment of qualitatively aberrant, rigid, maladaptive habitsin vulnerable individuals that are characterized by premor-bid alterations in corticostriatal-dependent inhibitory controlprocesses.

J.E. Murray et al. BIOL PSYCHIATRY 2013;]:]]]–]]] 7

Thus, in HI rats, there was a shift in the time-course of theeffects of bilateral intra-DLS infusions of the dopamine receptorantagonist α-flupenthixol to reduce active lever presses duringthe 15-min drug-seeking challenge tests. Although DLS dopaminereceptor blockade had no effect on cue-controlled cocaine-seeking responses at the early performance test stage, itsignificantly decreased active lever presses at the later, habitualtest stage, the two test stages when there were no significantdifferences between HI and LI rats. These data, in agreement withour previous work (23), thereby demonstrate that—regardless ofdifferences in impulse control—all subjects eventually developDLS dopamine-dependent cocaine-seeking habits after pro-tracted drug-seeking performance (3,8,15,23). However, at theintermediate stage of training, cocaine-seeking responses weredecreased by DLS dopamine receptor blockade specifically in LIbut not HI rats.

This delayed recruitment of the DLS in the control overcocaine seeking suggests that low availability of ventral striatumdopamine D2 receptors might influence drug-induced adapta-tions underlying the progressive ventral to dorsal striatal shift thatoccurs in the course of addiction in humans (12,34) and duringextended periods of cocaine self-administration in nonhumanprimates (8,9,11,35) and rats (10). We and others have suggestedthat this ventral to dorsal striatal shift depends upon thedopamine-dependent ascending spiraling circuitry (36,37) func-tionally linking the ventral with the dorsolateral striatum(13,15,31,38), even though the mechanisms whereby this circuitryis recruited remain to be established. Added to the recentdemonstration that the progressive cocaine-induced ventral tothe dorsal striatum decrease in dopamine D2 receptors andmessenger RNA (mRNA) levels demonstrated in primates (39–41)and rats (10) is also delayed in HI as compared with LI rats (10),despite lower baseline levels of D2 mRNA in the nucleusaccumbens shell and dopaminergic neurons of the former (10),the present results suggest that low D2 receptor availability in theventral striatum retards intra-striatal cocaine-induced plasticityprocesses. This is consistent with the demonstration that individ-ual vulnerability to develop addiction-like behavior for cocaine,that we have demonstrated to be highly predicted by highimpulsivity (18), is associated with impaired cocaine-inducedplasticity in the ventral striatum (42).

Although protracted cocaine exposure results in markeddecreases in striatal D2 dopamine receptor and mRNA levels,an adaptation suggested to contribute to the development ofaddiction (39,43–45), cocaine self-administration in HI rats thatdisplay spontaneous low D2 mRNA and receptor levels in theventral striatum results in a normalization of D2 receptor levels(46) that parallels a reduction in impulsivity. This observationtherefore suggests that the potential delay in dorsal striatalrecruitment after cocaine exposure observed in HI rats mightbe attributed to cocaine-induced remediation of low D2dopamine receptors in the ventral striatum and the associatedimpulsivity that occurs early on after cocaine self-administration. Indeed, this hypothesis is supported by a recentmicro positron emission topography study in LI and HI rats (46).This has important implications at the psychological level inthat it suggests that, for HI rats, instrumental actions forcocaine might remain goal-directed for longer than in LI rats,a consequence partly determined by a dopamine deficiencystate in the ventral striatum. This is consistent with theobservation that HI rats are more focused on a food goal thanLI rats, spending more time at the food delivery magazinewhen trained in the 5-CSRTT. Moreover goal-trackers in a

Pavlovian conditioned approach task motivated by food weremore impulsive in a delay discounting task than sign-trackers(47), a dimension of impulsivity that is also expressed by HIrats selected in the 5-CSRTT (48). These observations indicatethat impulsivity is associated with a dominance of goal-direct-ed behavior during early experience in instrumental andPavlovian tasks.

The present results show that the psychological mecha-nisms whereby impulsivity and habits contribute to addictiondo not depend upon a facilitation of the development of thelatter by the former. However, it is pivotal to dissociate thepropensity to develop habits, which in itself is not an aberrantprocess, from the inability to regain control over maladaptivehabits that have become inflexible, such as those that are seenin addicted individuals who compulsively seek and take drugs.This further suggests that vulnerability to addiction does notlie in the propensity of an individual to develop habits butinstead in the rigid nature of drug-seeking habits and theinability of an individual to regain control over these malad-aptive habits. This inflexibility of drug-seeking habits mightstem from either cortical (49) or striatal components of weakinhibitory control or in the persistence of aberrant neuro-biological adaptations that have accumulated during therecruitment of dorsolateral striatal control over behavior toovercome the apparent lack of striatal neuroplasticity thatcharacterizes HI rats (10).

This work was supported by Medical Research Council (MRC)grants to BJE and JWD (G1002231, G0701500) and by a joint coreaward from the MRC and Wellcome Trust (MRC G1000183; WT093875/Z/10/Z) in support of the Behavioral and Clinical Neuro-science Institute at Cambridge University.

We acknowledge funding support within the MRC ImperialCollege-Cambridge University-Manchester University (ICCAM) strate-gic addiction cluster (G1000018). DB is a member of the Groupe deRecheche (GDR) 3557 and is supported by an INSERM AVENIR grant,the ANR “heraddictstress,” the IREB and the University of Poitiers. Wethank Emily Jordan, David Theobald, and Alan Lyon for theirtechnical assistance.

The authors reported no biomedical financial interests orpotential conflicts of interest.

Supplementary material cited in this article is available online athttp://dx.doi.org/10.1016/j.biopsych.2013.09.011.

1. Chen BT, Yau HJ, Hatch C, Kusumoto-Yoshida I, Cho SL, Hopf FW, BonciA (2013): Rescuing cocaine-induced prefrontal cortex hypoactivityprevents compulsive cocaine seeking. Nature 496:359–362.

2. Pelloux Y, Dilleen R, Economidou D, Theobald D, Everitt BJ (2012):Reduced forebrain serotonin transmission is causally involved in thedevelopment of compulsive cocaine seeking in rats. Neuropsycho-pharmacology 37:2505–2514.

3. Zapata A, Minney VL, Shippenberg TS (2010): Shift from goal-directedto habitual cocaine seeking after prolonged experience in rats. JNeurosci 30:15457–15463.

4. Jentsch JD, Taylor JR (1999): Impulsivity resulting from frontostriataldysfunction in drug abuse: implications for the control of behavior byreward-related stimuli. Psychopharmacology 146:373–390.

5. Dickinson A, Wood N, Smith J (2002): Alcohol seeking by rats: Actionor habit? Q J Exp Psychol B 55:331–348.

6. Corbit LH, Nie H, Janak PH (2012): Habitual alcohol seeking: timecourse and the contribution of subregions of the dorsal striatum. BiolPsychiatry 72:389–395.

7. Everitt B, Robbins T (2005): Neural systems of reinforcement for drugaddiction: From actions to habits to compulsion. Nat Neurosci 8:1481–1489.

www.sobp.org/journal

8 BIOL PSYCHIATRY 2013;]:]]]–]]] J.E. Murray et al.

8. Porrino LJ, Daunais JB, Smith HR, Nader MA (2004): The expandingeffects of cocaine: Studies in a nonhuman primate model of cocaineself-administration. Neurosci Biobehav Rev 27:813–820.

9. Porrino L (2004): Cocaine self-administration produces a progressiveinvolvement of limbic, association, and sensorimotor striatal domains.J Neurosci 24:3554–3562.

10. Besson M, Pelloux Y, Dilleen R, Theobald D, Belin-Rauscent A, RobbinsTW, et al. (2013): Cocaine modulation of fronto-striatal expression ofzif268, D2 and 5-HT2c receptors in high and low impulsive rats.Neuropschopharmacology 38:1963–1973.

11. Porrino L, Smith HR, Nader MA, Beveridge TJ (2007): The effects ofcocaine: A shifting target over the course of addiction. Prog Neuro-psychopharmacol Biol Psychiatry 31:1593–1600.

12. Vollstadt-Klein S, Wichert S, Rabinstein J, Buhler M, Klein O, Ende G,et al. (2010): Initial, habitual and compulsive alcohol use is charac-terized by a shift of cue processing from ventral to dorsal striatum.Addiction 105:1741–1749.

13. Xie C, Shao Y, Ma L, Zhai T, Ye E, Fu L, et al. (2012): Imbalancedfunctional link between valuation networks in abstinent heroin-dependent subjects [published online ahead of print December 4].Mol Psychiatry.

14. Vanderschuren LJ, Di Ciano P, Everitt BJ (2005): Involvement of the dorsalstriatum in cue-controlled cocaine seeking. J Neurosci 25:8665–8670.

15. Belin D, Everitt BJ (2008): Cocaine-seeking habits depend upondopamine-dependent serial connectivity linking the ventral with thedorsal striatum. Neuron 57:432–441.

16. Jonkman S, Pelloux Y, Everitt BJ (2012): Differential roles of thedorsolateral and midlateral striatum in punished cocaine seeking. JNeurosci 32:4645–4650.

17. American Psychiatric Association (1994): Diagnostic and StatisticalManual of Mental Disorders, 4th ed. Washington, DC: AmericanPsychiatric Press.

18. Belin D, Mar A, Dalley J, Robbins T, Everitt B (2008): High impulsivitypredicts the switch to compulsive cocaine-taking. Science 320:1352–1355.

19. Dalley JW, Fryer T, Brichard L, Robinson E, Theobald D, Laane K, et al.(2007): Nucleus accumbens D2/3 receptors predict trait impulsivityand cocaine reinforcement. Science 315:1267–1270.

20. Piray P, Keramati MM, Dezfouli A, Lucas C, Mokri A (2010): Individualdifferences in nucleus accumbens dopamine receptors predict devel-opment of addiction-like behavior: A computational approach. NeuralComput 22:2334–2368.

21. Belin D, Belin-Rauscent A, Murray JE, Everitt BJ (2013): Addiction:Failure of control over maladaptive incentive habits. Curr Opin Neuro-biol 23:564–572.

22. Everitt B, Robbins T (2000): Second-order schedules of drug reinforce-ment in rats and monkeys: Measurement of reinforcing efficacy anddrug-seeking behaviour. Psychopharmacology 153:17–30.

23. Murray JE, Belin D, Everitt BJ (2012): Double dissociation of thedorsomedial and dorsolateral striatal control over the acquisition andperformance of cocaine seeking. Neuropsychopharmacology 37:2456–2466.

24. Robbins T (2002): The 5-choice serial reaction time task: Behaviouralpharmacology and functional neurochemistry. Psychopharmacology163:362–380.

25. Bari A, Dalley J, Robbins T (2008): The application of the 5-choice serialreaction time task for the assessment of visual attentional processesand impulse control in rats. Nature Protocol 3:759–767.

26. McNamara R, Dalley JD, Robbins TW, Everitt BJ, Belin D (2010): Trait-like impulsivity does not predict escalation of heroin self-administration in the rat. Psychopharmacology 212:453–464.

27. Paxinos G, Watson C (1998): The Rat Brain in Stereotaxic Coordinates,4th ed. San Diego: Academic Press.

28. Everitt BJ, Belin D, Economidou D, Pelloux Y, Dalley J, Robbins TW(2008): Neural mechanisms underlying the vulnerability to developcompulsive drug-seeking habits and addiction. Philos Trans R Soc LondB Biol Sci 363:3125–3135.

29. Hedges LV (1994): Fixed effect models. In: Cooper H, Hedges LV,editors. Handbook of Research Synthesis. New York: Russell SageFoundation, 301–321.

www.sobp.org/journal

30. Ito R, Dalley J, Robbins T, Everitt BJ (2002): Dopamine release in thedorsal striatum during cocaine-seeking behavior under the control ofa drug-associated cue. J Neurosci 22:6247–6253.

31. Willuhn I, Burgeno LM, Everitt BJ, Phillips PE (2012): Hierarchicalrecruitment of phasic dopamine signaling in the striatum during theprogression of cocaine use. Proc Natl Acad Sci U S A 109:20703–20708.

32. Belin-Rauscent A, Everitt BJ, Belin D (2012): Intrastriatal shifts mediatethe transition from drug-seeking actions to habits. Biol Psychiatry 72:343–345.

33. Everitt BJ, Robbins TW (2013): From the ventral to the dorsal striatum:Devolving views of their roles in drug addiction [published onlineahead of print February 21]. Neurosci Biobehav Rev.

34. Volkow N, Wang GJ, Telang F, Fowler JS, Logan J, Childress AR, et al.(2006): Cocaine cues and dopamine in dorsal striatum: Mechanism ofcraving in cocaine addiction. J Neurosci 26:6583–6588.

35. Letchworth SR, Nader MA, Smith HR, Friedman DP, Porrino L (2001):Progression of changes in dopamine transporter binding site densityas a result of cocaine self-administration in rhesus monkeys. J Neurosci21:2799–2807.

36. Haber S, Fudge JL, McFarland NR (2000): Striatonigrostriatal pathwaysin primates form an ascending spiral from the shell to the dorsolateralstriatum. J Neurosci 20:2369–2382.

37. Ikemoto S (2007): Dopamine reward circuitry: Two projection systemsfrom the ventral midbrain to the nucleus accumbens–olfactorytubercle complex. Brain Res Rev 56:27–78.

38. Keramati M, Gutkin B (2013): Imbalanced decision hierarchy in addictsemerging from drug-hijacked dopamine spiraling circuit. PLoS One 8:e61489.

39. Volkow N, Fowler J, Wang G, Hitzemann R (1993): Decreaseddopamine D2 receptor availability is associated with reduced frontalmetabolism in cocaine abusers. Synapse 14:169–177.

40. Moore RJ, Vinsant SL, Nader MA, Porrino L, Friedman DP (1998): Effectof cocaine self-administration on dopamine D2 receptors in rhesusmonkeys. Synapse 30:88–96.

41. Nader M, Morgan D, Gage H, Nader S, Calhoun T, Buchheimer N, et al.(2006): PET imaging of dopamine D2 receptors during chronic cocaineself-administration in monkeys. Nat Neurosci 9:1050–1056.

42. Kasanetz F, Deroche-Gamonet V, Berson N, Balado E, Lafourcade M,Manzoni O, Piazza PV (2010): Transition to addiction is associatedwith a persistent impairment in synaptic plasticity. Science 328:1709–1712.

43. Volkow ND, Fowler JS, Wang GJ, Goldstein RZ (2002): Role ofdopamine, the frontal cortex and memory circuits in drug addiction:insight from imaging studies. Neurobiol Learn Mem 78:610–624.

44. Volkow ND, Fowler JS, Wang GJ, Baler R, Telang F (2009): Imagingdopamine’s role in drug abuse and addiction. Neuropharmacology56(suppl 1):3–8.

45. Asensio S, Romero MJ, Romero FJ, Wong C, Alia-Klein N, Tomasi D,et al. (2010): Striatal dopamine D2 receptor availability predicts thethalamic and medial prefrontal responses to reward in cocaineabusers three years later. Synapse 64:397–402.

46. Caprioli D, Hong YT, Sawiak SJ, Ferrari V, Williamson DJ, Jupp B, et al.(2013): Baseline-dependent effects of cocaine pre-exposure on impul-sivity and D2/3 receptor availability in the rat striatum: Possiblerelevance to the attention-deficit hyperactivity syndrome. Neuropsy-chopharmacology 38:1460–1471.

47. Flagel SB, Robinson TE, Clark JJ, Clinton SM, Watson SJ, Seeman P,et al. (2010): An animal model of genetic vulnerability to behavioraldisinhibition and responsiveness to reward-related cues: Implicationsfor addiction. Neuropsychopharmacology 35:388–400.

48. Robinson ES, Eagle DM, Economidou D, Theobald DE, Mar AC, MurphyER, et al. (2009): Behavioural characterisation of high impulsivity onthe 5-choice serial reaction time task: Specific deficits in ‘waiting’versus ‘stopping’. Behav Brain Res 196:310–316.

49. Jupp B, Caprioli D, Saigal N, Reverte I, Shrestha S, Cumming P, et al.(2013): Dopaminergic and GABA-ergic markers of impulsivity in rats:Evidence for anatomical localisation in ventral striatum and prefrontalcortex. Eur J Neurosci 37:1519–1528.