Durgs consumption

Individual differences in substance dependence: at theintersection of brain, behaviour and cognition

Travis E. Baker1, Tim Stockwell1,2, Gordon Barnes3 & Clay B. Holroyd1

Department of Psychology1, Center of Addiction Research of British Columbia2, and Child and Youth Care3, University of Victoria, Canada

ABSTRACT adb_243 458..466

Recent theories of drug dependence propose that the transition from occasional recreational substance use to harmfuluse and dependence results from the impact of disrupted midbrain dopamine signals for reinforcement learning onfrontal brain areas that implement cognitive control and decision-making. We investigated this hypothesis in humansusing electrophysiological and behavioral measures believed to assay the integrity of midbrain dopamine system andits neural targets. Our investigation revealed two groups of dependent individuals, one characterized by disrupteddopamine-dependent reward learning and the other by disrupted error learning associated with depression-proneness.These results highlight important neurobiological and behavioral differences between two classes of dependent usersthat can inform the development of individually tailored treatment programs.

Keywords addiction, cognitive control, event-related brain potentials, feedback error-related negativity, midbraindopamine system, reinforcement learning.

Correspondence to: Travis E. Baker, Department of Psychology, University of Victoria, P.O. Box 3050 STN CSC, Victoria, BC V8W 3P5, Canada. E-mail:[email protected]

INTRODUCTION

Are we in control of our own decisions? Most of us feel incontrol, but individuals who suffer from severe drugdependence exhibit impaired control over their substanceuse despite often catastrophic consequences on personalhealth, finances and social relationships. Yet, despite thewidespread availability and prevalence of addictive sub-stances in most societies (Anderson 2006), only somedrug users ultimately become dependent (Kessler et al.2005). Over the last several decades, multidisciplinaryefforts in addictions research have indicated that sub-stance dependence results from a confluence of riskfactors related to biology, cognition and learning, person-ality, genetics and the social environment, but there is asyet little direct evidence in humans of the neuroadaptivemechanisms that mediate the transition from occasional,controlled drug use to the impaired control that charac-terizes severe dependence (Hyman 2007).

Notably, all addictive drugs stimulate the midbraindopamine system (MDS) (Di Chiara & Imperato 1988),which projects to and regulates brain structures underly-ing cognitive control and decision-making, namely pre-frontal cortex (Cohen, Braver & Brown 2002), anterior

cingulate cortex (ACC) (Holroyd & Coles 2002) and thebasal ganglia (BG) (Cohen & Frank 2008). MDS neuronsdistribute information about rewarding events such thatphasic bursts and dips in dopamine neuron activity areelicited when events are, respectively, ‘better thanexpected’ [positive reward prediction error (RPE)] and‘worse than expected’ (negative RPE) (Schultz 1998). Inkeeping with formal models of reinforcement learning,these RPEs ‘propagate back in time’ in trial-and-errorlearning tasks from reward delivery to the earliest predic-tive indicator of reward. Accordingly, it has been sug-gested that the dopamine RPEs serve as reinforcementlearning signals, gradually optimizing behavior by asso-ciating predictive cues and behaviors with forthcomingrewards (Schultz 1998). In this way, the dopamine RPEsignals appear to increase the ‘incentive salience’ or‘wanting’ of rewards, that is, the motivation to work forthe reward in a given behavioral context, as distinct fromthe affective enjoyment or ‘liking’ of the reward whenconsumed (McClure, Daw & Montague 2003).

In view of the role played by the MDS in reinforcementlearning, addiction has recently been hypothesized to befundamentally a problem of learning and memory(Hyman 2005). According to this view, drugs of abuse

HUMAN/CLINICAL STUDY

Addiction Biologydoi:10.1111/j.1369-1600.2010.00243.x

© 2010 The Authors, Addiction Biology © 2010 Society for the Study of Addiction Addiction Biology, 16, 458–466

effectively increase the magnitude of the positive RPEscarried by the MDS by raising extracellular dopaminelevels either directly or indirectly (Di Chiara & Imperato1988). Whereas natural rewards and external cuesassociated with reward produce transient increases indopamine neuron activity only when these events areunexpected, addictive drugs and drug-related cuesincrease dopamine levels even when these events areexpected, thereby augmenting the size of the elicited posi-tive RPE signals (Rice & Cragg 2004). In turn, these exag-gerated signals induce changes to synaptic connectivity(Hyman, Malenka & Nestler 2006) that rewire anddisrupt the neural targets of the MDS in ACC, BG andorbitofrontal cortex (OFC) (Robinson & Kolb 2004;Homayoun & Moghaddam 2006), causing the motiva-tional value of states that precede drug consumption togrow without bound (Redish 2004). Because these brainareas implement neural processes that are central tocognitive control and decision-making—includinggoal-directed action selection, response activation andinhibition, performance monitoring and reward-basedlearning (Miller & Cohen 2001; Cohen et al. 2002;Holroyd & Coles 2002; Cohen & Frank 2008)—addictivedrugs are sometimes said to ‘usurp’ the cognitive controlsystem (Hyman 2007).

Here, we hypothesized that the impact of disruptedRPEs on brain networks involved in cognitive control anddecision-making precipitate the compulsive drug use thatdefines severe dependence. To investigate this hypothesis,we indirectly assayed the integrity of the dopaminesystem and frontal brain areas involved in cognitivecontrol and decision-making in young adults using acombination of electrophysiological and behavioralmeasures together with surveys of substance use andpersonality.

Specifically, to assess the neural integrity of the MDSand its projections to frontal cortex, event-related brainpotentials (ERPs) were recorded from participants as theynavigated a ‘virtual T-maze’ to find rewards (Baker &Holroyd 2009). It is believed that the impact of dopamineRPEs on motor-related areas in ACC modulate the ampli-tude of a component of the ERP called the feedback error-related negativity (fERN) (Holroyd & Coles 2002; Baker &Holroyd 2009). Like the dopamine RPE signals, the fERNis sensitive to events that first indicate when events arebetter or worse than expected (Holroyd & Coles 2002;Holroyd & Krigolson 2007; Baker & Holroyd 2009;Holroyd et al. 2009). Further, genetic (Marco-Pallareset al. 2009), pharmacological and neuropsychological(Overbeek, Nieuwenhuis & Ridderinkhof 2005) evidenceimplicates dopamine in fERN production, although thespecific mechanism is debated (Jocham & Ullsperger2009). We predicted that if substance dependence resultsin part from the impact of disrupted dopamine RPE

signals on frontal brain structures involved in cognitivecontrol, then the fERN should be abnormal in Dependentbut not Non-dependent individuals.

In addition, immediately following the T-maze partici-pants engaged in the Probabilistic Selection Task (PST)(Frank, Seeberger & O’reilly 2004), a trial-and-errorlearning task that is believed to be sensitive to dopaminedysfunction. The PST has provided insight into individualdifferences related to Parkinson’s disease, attention-deficit hyperactivity disorder, schizophrenia, normalaging, genetic makeup, the effect of dopaminergic ago-nists and antagonists, and ‘top-down’ modulation of theBG by OFC and ACC (Cohen & Frank 2008). Accordingto an influential neurocomputational theory of theBG-MDS, positive dopamine RPEs facilitate approachlearning in this task by reinforcing a BG ‘Go’ pathway viaD1 receptors, whereas negative dopamine RPEs facilitateavoidance learning by reinforcing a BG ‘No-go’ pathwayvia D2 receptors (Cohen & Frank 2008). For example,prior studies revealed that people with Parkinson’sdisease, who have low striatal dopamine levels, werebetter at avoidance learning than approach learning;dopamine medications reversed this bias as predicted bythe models (Frank et al. 2004). We predicted that if sub-stance dependence results in part from the impact of dis-rupted dopamine RPE signals on brain structuresinvolved in decision-making, then performance in thistask should be abnormal in Dependent but not Non-dependent individuals.

MATERIALS AND METHODS

Participants

We first collected survey data (substance use history, per-sonality risk factors associated with addiction and familyhistory) from 412 first- and second-year undergraduatestudents. Of these participants, 70 agreed to return toparticipate in an electrophysiological and behavioralexperiment on a subsequent day (two additional partici-pants were excluded because of a reported head injury).The computer-based survey was comprised of severalseparate inventories, namely, the Alcohol, Smoking andSubstance Involvement Screening Test (ASSIST) (Hume-niuk & Ali 2006), a validated screening test developed bythe World Health Organization for identifying the degreeof problematic substance use (i.e. tobacco, alcohol, can-nabis, cocaine, amphetamine-type stimulants, sedatives,hallucinogens, inhalants, opioids and ‘other drugs’); theSeverity of Alcohol Dependence Questionnaire, whichassesses the severity of alcohol abuse and dependence(Stockwell, Murphy & Hodgson 1983); the Addiction-Prone Personality (APP) Scale (Anderson et al. 1999), a21-item self-report questionnaire that assesses the role of

Substance dependence 459


personality in the susceptibility to addiction; and the Sub-stance Use Risk Profile Scale (SURPS) (Conrod & Woicik2002), a 23-item self report questionnaire that providesa measure on four dimensions of personality traits—depression-proneness, anxiety-sensitivity, impulsivityand sensation seeking—that are risk factors for sub-stance use.

For the purpose of this study, participants were classi-fied as either Dependent or Non-dependent substanceusers according to their scores on the Global Continuumof Substance Risk (GCR) scale of the ASSIST. Specifically,participants with GCR scores falling within the bottom(score < 16) and top (score > 41) quartiles of our samplewere classified as Non-dependent (18 participants) andDependent (18 participants) users, respectively. Thesescores are comparable with the cut-offs established inprevious validation studies of the ASSIST for non-dependence (score < 15) and dependence (score > 39.5)(Newcombe, Humeniuk & Ali 2005). The DependentGroup tended to abuse alcohol, cannabis and tobacco, butsome individuals also reported taking amphetamines,cocaine, sedatives and/or hallucinogens (see supportinginformation; details are provided at the end of the paper).The study was conducted in accordance with the ethicalstandards prescribed in the 1964 Declaration of Helsinki.

Procedure

ERP task—virtual T-maze

The virtual T-maze is a guessing/reinforcement learningtask that elicits robust fERNs (Baker & Holroyd 2009).Participants navigated the virtual T-maze by pressing leftand right buttons corresponding to images of a left andright alley presented on a computer screen. After eachresponse, an image of the chosen alley appeared, followedby a feedback stimulus (apple or orange) indicatingwhether the participant received 0 or 5 cents on that trial;unbeknown to the participants, the feedback was randomand equiprobable. The experiment consisted of fourblocks of 50 trials each separated by rest periods. ERPswere created for each electrode and subject by averagingthe single-trial electroencephalography (EEG) accordingto feedback type (for a complete description of EEG DataAcquisition and Analysis methods, please see SOM).

For each participant, the fERN was measured atchannel FCz, where it reaches maximum amplitude(Miltner, Braun & Coles 1997; Holroyd & Krigolson2007). To isolate the fERN from other overlapping ERPcomponents, the fERN was evaluated for each participantas a difference wave by subtracting the Reward feedbackERPs from the corresponding No-reward feedback ERPs(Miltner et al. 1997; Holroyd & Krigolson 2007). Themean amplitude of this difference wave was obtained byaveraging the difference wave within a 200–320 ms

window following feedback onset. The P2 and P3 compo-nents were also measured for the purpose of comparison.The P2 was measured base-to-peak at a frontal-centralchannel (FCz) for the Reward and No-reward ERPs. TheP3 amplitude was measured by identifying the maximumpositive-going value of the Reward and No-reward ERPsrecorded at electrode site Pz, within a window extendingfrom 300 to 600 ms following the presentation of thefeedback stimulus (see SOM for further details).

Behavioral task—the probabilistic selection task

Consistent with standard practice, the feedback stimuli inthe T-maze task were delivered at random, providing ameans to identify the fERN using the difference waveapproach (Holroyd & Coles 2002; Holroyd & Krigolson2007; Baker & Holroyd 2009; Holroyd et al. 2009), butfor this reason the task did not provide a meaningful per-formance measure. Hence, immediately after participantscompleted the T-maze, we asked them to engage in thePST, a task designed to identify individual biases to learn-ing from positive or negative feedback (Frank et al. 2004).In brief, during an initial Learning Phase, participantswere exposed to three pairs of stimuli presented inrandom order (for more details, please see SOM). Theresponse mappings were probabilistic such that onestimulus in each of the three pairs was rewarded on 80%,70% and 60% of the trials, respectively, with the remain-ing stimulus in each pair rewarded on the complemen-tary percentage of trials. Given that these stimulusprobabilities are not optimal for extracting the fERN usingthe difference wave approach, EEG data were not recordedduring this task (Holroyd & Krigolson 2007; Holroyd et al.2009). Participants learned by trial-and-error to choosethe more frequently rewarded stimulus over the alterna-tive in each pair. Critically, they could do so either bylearning that particular stimuli were associated withrelatively more reward, by learning that particularstimuli were associated with relatively more punishment,or both. During the Test Phase, participants were exposedto all possible combinations of these stimuli in a randomorder and were required to select the symbol in each pairthat they believed to be correct, but without receiving anyfeedback about their choices. If participants learned morefrom positive feedback during the Learning Phase, thenthey should reliably choose the Good Stimulus in all noveltest pairs in which it is present. On the other hand, if theylearned more from negative feedback during the LearningPhase, then they should reliably avoid the Bad Stimulusin all novel test pairs in which it is present. Participantswho did not perform better than chance on Test Phasetrials consisting of the easiest stimulus pair were elimi-nated from further analysis. In total, the data of six par-ticipants were discarded.

460 Travis E. Baker et al.


As in previous studies, we identified two subgroups ofparticipants (Frank, Woroch & Curran 2005; Frank,D’Lauro & Curran 2007). Participants who tended to pickthe stimulus that was most frequently rewarded duringthe Learning Phase (the ‘Good Stimulus’), which dependson learning from positive reinforcement, were classifiedas ‘Positive Learners’, whereas participants who tendedto avoid the stimulus that was most frequently punishedduring the Learning Phase (the ‘Bad Stimulus’), whichdepends on learning from negative reinforcement, wereclassified as ‘Negative Learners’. Six subjects displayedequally good performance in choosing the Good Stimulusand avoiding the Bad Stimulus and were not included ineither group (but were included in a continuous measureof relative learning biases; see next). Group comparisonsconfirmed that Positive Learners (n = 32) were betterthan Negative Learners (n = 27) at choosing the GoodStimulus, t(57) = 6.28, P < 0.001, whereas Negative

Learners were better than Positive Learners at avoidingthe Bad Stimulus, t(57) = -4.7, P < 0.001 (for moredetails, see SOM).

RESULTS

Electrophysiological results

Figure 1 (a and b) illustrates the ERPs elicited by theReward and No-reward feedback and the associated dif-ference waves, averaged across participants separately forthe Non-dependent and Dependent Groups. The ERPs forthe Non-dependent Group revealed a typical fERN occur-ring at about 250 ms following feedback presentation(Holroyd & Coles 2002) (Fig. 1a), whereas the ERPs forthe Dependent Group were nearly identical, exhibitinglittle difference between conditions (Fig. 1b). Figure 1cpresents the associated difference waves together,

(a)

(c) (d)

(b)

Figure 1 Event-related brain potential (ERP) data associated with frontal-central electrode channel FCz. Grand-average ERPs associated withReward (blue dotted lines) and No-reward (red dashed lines) outcomes and associated difference waves (black solid lines) for the (a)Non-dependent Group and the (b) Dependent Group. (c) Feedback error-related negativity (fERN) difference waves for the Non-dependentGroup (solid lines) and Dependent Group (dashed lines). In a–c, 0 ms corresponds to time of feedback delivery. (d) fERN amplitude as afunction of LearnerType (Positive and Negative) derived from performance on the Probabilistic SelectionTask, for the Non-dependent Group(solid line) and the Dependent Group (dashed line). Bars indicate the standard error of the mean. Negative voltages are plotted up byconvention



revealing a truncated fERN in the Dependent Group(M = -1.9 mV, SE = �0.4) relative to the Non-dependentGroup (M = -3.6 mV, SE = �0.5), t(34) = -2.34, P <0.05. Further analysis indicated that the amplitudes ofthe P200 and the P300 were about the same for the twogroups (P > 0.05), confirming that the effect of interestwas isolated to the predicted ERP component—thefERN—and thus did not reflect an overall processingdifference between the groups. Moreover, this effectremained statistically significant when variability associ-ated with personality-related risk factors for substanceuse (i.e. depression-proneness, anxiety, impulsivity andsensation seeking), as measured by the SURPS and APP,was controlled for using ANCOVA, F(1, 36) = 3.9,

P < 0.05. Hence, the degree of substance use appears tohave affected fERN amplitude independently of the per-sonality traits that precipitated the substance use in thefirst place.

Behavioral results

Overall, a two-way ANOVA on Test Phase accuracy withGroup (Non-dependent, Dependent) and Stimulus type(Positive, Negative) as factors revealed a main effect ofGroup, indicating that the Non-dependent Group per-formed more accurately (83%) than the DependentGroup (66%) did, F(1,31) = 9.2, P < 0.005, ES = 0.23(Fig. 2a). Specifically, Non-dependent participants

(a) (b)

(c) (d)

Figure 2 Performance on the Probabilistic Selection Task (PST). Accuracy in the Test Phase of the PST for the Dependent and Non-dependent Groups, separately for the Choose Good and Avoid Bad conditions, for (a) all participants, (b) Negative Learners only and (c)Positive Learners only. Note that chance accuracy is 50%. (d) Learning Index Scores derived from the PST accuracy data, separately for Positiveand Negative Learners. Dependent Group data are indicated by circles and dotted lines and Non-dependent Group data are indicated bysquares and solid lines. Bars indicate standard errors of the mean



avoided choosing the Bad Stimulus more often (84%)than Dependent participants did (65%), t(28) = 2.7,P < 0.01, and there was a trend such that the Non-dependent participants chose the Good Stimulus (84%)more often than the Dependent participants did (66%),t(28) = 1.8, P < 0.08. No between-group differences inperformance were found during the Learning Phase ofthe task (see SOM for details).

We examined this between-group difference in TestPhase accuracy by classifying the Dependent and Non-dependent participants as either Negative or PositiveLearners. For Negative Learners, both groups tended toavoid choosing the Bad Stimulus about equally often,t(12) = 1.2, P > 0.05, but the Non-dependent partici-pants (n = 7) tended to chose the Good Stimulus moreoften (75%) than the Dependent participants (n = 7) did(44%), t(12) = 2.7, P < 0.05 (Fig. 2b). Likewise, for Posi-tive Learners, both groups tended to choose the GoodStimulus about equally often, t(14) = 0.92, P > 0.05, butthe Non-dependent participants (n = 7) tended to avoidchoosing the Bad Stimulus more often (76%) than theDependent participants (n = 9) did (50%), t(14) = 2.8,P < 0.01 (Fig. 2c).

We investigated this issue further by determining foreach subject the degree to which they used their preferredstrategy relative to their non-preferred strategy. For eachsubject, we computed the Learning Index Score (LIS),defined as LIS = (preferred accuracy – non-preferredaccuracy)/(preferred accuracy + non-preferred accu-racy); higher LIS scores indicate a greater preference forone strategy over the other. The LIS scores for Dependentand Non-dependent participants are shown separately forPositive and Negative Learners in Fig. 2d. A two-wayANOVA on LIS as a function of Group (Non-dependent,Dependent), and learner type (Positive, Negative) revealeda main effect of Group, F(1, 28) = 14.46, P < 0.001,ES = 0.35, indicating that Dependent participants exhib-ited a larger learning bias (Mean = 0.30) compared withNon-dependent participants (Mean = 0.10); all othermain effects and interactions were not significant,P > 0.05. Taken together, these results indicate that theDependent and Non-dependent participants performedthe task about equally well when allowed to use theirpreferred strategies, but that the Dependent participantswere severely impaired relative to the Non-dependent par-ticipants when required to use their non-preferred strate-gies. Thus, the overall performance difference acrossgroups illustrated in Fig. 2a resulted mainly from theDependent participants responding at chance accuracywhen forced to rely on their less favored methods forresponse selection. Note that this finding argues against ageneral cognitive or learning impairment in the Depen-dent participants, which would be expected to impact bothstrategies equally.

Interaction between fERN and PST results

Given that the reward processing system that producesthe fERN might be sensitive to learning style, we exam-ined fERN amplitude as a function of both Group andLearner Type. A two-way ANOVA on fERN amplitudewith Group (Non-dependent, Dependent) and LearnerType (Positive, Negative) as factors revealed a significantinteraction between Group and Learner Type, F(1,28) = 4.3, P < 0.05, ES = 0.13 (Fig. 1d). Post hoc analysisrevealed that the ERP effect of interest was mainly drivenby a reduced fERN in Dependent Negative Learners(M = -1.2 mV, SE = �0.5) relative to Non-dependentNegative Learners (M = -4.7 mV, SE = �0.4), P < 0.01;all other paired comparisons were non-significant(P > 0.05, corrected for multiple comparisons using Bon-feronni correction). In other words, the reduced fERN inthe Dependent Group relative to the Non-dependentGroup was associated with the participants who werebetter in the PST at avoiding the Bad Stimulus than atchoosing the Good Stimulus. Further, a two-way ANOVAon each of the personality trait scores as a function ofLearner Type and Group revealed that none of these wererelated to Learner Type or Group (P > 0.05) except forDepression-proneness (Conrod & Woicik 2002). Specifi-cally, Dependent Positive Learners scored higher on theDepression-proneness scale (M = 14, SE = �1.1) than didDependent Negative Learners (M = 10, SE = �0.6),t(14) = 2.4, P < 0.05. In other words, Dependent partici-pants who scored high on the Depression-proneness scalewere relatively successful in the PST at choosing the GoodStimulus but relatively impaired at avoiding the BadStimulus. Further analysis indicated that scores onthe Depression-proneness were about the same for theNon-dependent Positive Learner compared with theNon-dependent Negative Learner Group (P > 0.05),confirming that the effect of interest was isolated toDependent Group and thus did not reflect an overalldifference in Depression-proneness between learningstrategies. Taken together, these results indicate thatDependent individuals who fail to learn from rewardfeedback produce a truncated neural response tofeedback, whereas Dependent individuals who fail tolearn from error feedback exhibit higher levels ofDepression-proneness.

DISCUSSION

Our findings are indicative of two separate groups ofdependent drug users, one characterized by impairedreward learning and the other characterized by impairederror learning. According to a neurocomputationaltheory of the fERN, this electrophysiological signal isargued to be elicited by the impact of RPEs carried by the



MDS onto motor areas in ACC, where they are utilized forthe adaptive modification of behavior according to prin-ciples of reinforcement learning (Holroyd & Coles 2002).Importantly, the difference in the ERPs elicited by positiveand negative feedback has recently been shown to resultmainly from reward processing induced by positive feed-back (Cohen, Elger & Ranganath 2007; Holroyd, Pakzad-Vaezi & Krigolson 2008). In line with this observation, wefound that for the dependent individuals who wereimpaired at reward learning, a negative-going deflectionin the ERP following Reward trials mirrored the negative-going deflection in the ERP following No-reward trials. Inother words, reward feedback failed to induce dopamine-dependent reward processing in these individuals.Further, computational simulations of the BG-MDS haveindicated that disrupted positive dopamine RPEs tend toupset reward learning while sparing error learning(Cohen & Frank 2008) as we observed (Fig. 2b). Thesefindings are consistent with the proposal that substancedependence is associated with the impact of impaireddopamine-mediated reinforcement learning signals onneural areas for cognitive control and decision-making.

It remains to be determined whether the drug use wasa consequence or the cause of this reward processingimpairment. On the one hand, the findings survived sta-tistical control of several important personality-relatedrisk factors for drug use. Further, the results are consis-tent with the observation that all drugs of abuse stimu-late the dopamine system (Di Chiara & Imperato 1988),resulting in maladaptive synaptic changes (Hyman et al.2006) that disrupt neural networks in ACC, OFC and BG(Robinson & Kolb 2004; Homayoun & Moghaddam2006), which in turn desensitizes the system to non-drugrewards (Koob & Le Moal 2005) like the small monetaryincentives used here (Volkow et al. 2009). These consid-erations suggest that heavy drug use may have modifiedthe MDS and its neural targets in this population. On theother hand, it is also possible that abnormal dopaminesignals resulted directly from dopamine-related geneticpolymorphisms associated with addiction-proneness(Kreek et al. 2005), impaired reward learning and sparederror learning (Cohen & Frank 2008), and reduced fERNamplitudes (Marco-Pallares et al. 2009). In fact, wesuspect that both factors may be involved, such that independence-prone individuals the reinforcing propertiesof addictive drugs exploit genetic vulnerabilities to thedopamine system.

By contrast, we found that the dependent individualswho were impaired at error learning scored high on thedepression-proneness scale when compared with depen-dent individuals who were not impaired at error learning.It is interesting to note that depression and drug depen-dence are highly comorbid, not only because depressedindividuals tend to take drugs of abuse for the purpose of

self-medication (Markou, Kosten & Koob 1998), but alsobecause substance use can lead to depression (Rehm,Taylor & Room 2006). Further, depressed individualssometimes rely on the analgesic properties of alcohol andother drugs to ameliorate negative affect (Conrod &Woicik 2002), which directs their thought processesaway from negative self-rumination toward positive self-reflection (Stephens & Curtin 1995). In this way, theanalgesic properties of drugs can reinforce behaviorsthat protect against negative, self-relevant information(Markou et al. 1998). Hence, we suggest that thedepression-prone dependent individuals in this studytended to ignore error feedback in favor of positive feed-back during the Training phase of the PST, leading tobetter performance on the ‘Choose Good’ trials relative tothe ‘Avoid Bad’ trials during the Test Phase of the PST.Consistent with this view, substance use could impairerror learning directly by altering OFC structure andfunction (Robinson & Kolb 2004; Homayoun & Moghad-dam 2006), thereby disrupting ‘top-down’ regulation ofthe BG Go and No-go pathways (Cohen & Frank 2008;Wheeler & Fellows 2008). The transition of these indi-viduals from a propensity to use addictive substances todependence could also be facilitated by dopamine-relatedgenetic vulnerabilities associated with addiction-proneness (Kreek et al. 2005), impaired negative learningand spared positive learning (Klein et al. 2007; Cohen &Frank 2008), and reduced error-related brain activationin ACC (Klein et al. 2007).

Although our participants were not screened for thepresence of comorbid disorders, such as attention-deficithyperactivity disorder and major depression, the experi-mental results remained robust even when the effectsof personality traits related to anxiety, depression-proneness, impulsivity and sensation seeking were con-trolled for statistically. Nevertheless, future studies shouldexamine this possible confounding factor. It was also thecase that the participants were not screened for acutedrug use before starting the experimental session. Asidefrom the fact that they did not display any obvious signs ofrecent drug or alcohol use while being tested, we believethat our results are uncontaminated by acute drug usefor the following reasons. First, dopamine agonists suchas caffeine, nicotine and amphetamine increase ERNamplitude (Overbeek et al. 2005; Jocham & Ullsperger2009), but the dependent individuals in our study exhib-ited decreased, rather than increased, fERNs. Second,depressants such as alcohol tend to depress other ERPcomponents such as the P300 in addition to the ERN(Holroyd & Yeung 2003; Polich & Criado 2006). By con-trast, despite the large reduction in fERN amplitude in thedependent participants in our study, the P200 and P300components appeared entirely normal—indicating thatthe effects of drug use were in fact limited to the fERN.



Given that substance users bring with them diverselife histories, personalities, biological/genetic profiles anddrug preferences, substance dependence has provenextremely challenging to treat. An obvious next stepwould be the inclusion of neurobiological markers ofsubstance dependence in individually tailored treatmentprograms. For instance, combined assessment of electro-physiological, cognitive and genetic profiles could poten-tially improve upon current therapeutic approaches andbetter predict vulnerability to relapse. By highlightingimportant neurobiological and behavioral differencesbetween two classes of dependent users, this researchmay represent an important step in this promisingdirection.

Acknowledgements

This research was supported by a British ColumbiaMental Health and Addiction Research Seed Grant and aNatural Sciences and Engineering Research Council Dis-covery Grant (RGPIN 312409-05). The first author wassupported by a Canadian Institute of Health ResearchDoctoral Award, through the Integrated Mentor Programin Addictions Research Training. We are grateful toMichael Frank for providing us with the PST task and forhelp with data analysis, as well as the research assistantsin the Brain and Cognition Laboratory for assistance withdata collection.

Authors Contribution

TEB, TS, GB and CBH designed the study; TEB collectedand analyzed all questionnaire, ERP and behavioral data;TEB and CBH wrote the manuscript.

References

Anderson P (2006) Global use of alcohol, drugs and tobacco.Drug Alcohol Rev 25:489–502.

Anderson RE, Barnes GE, Patton D, Perkins TM (1999) Person-ality in the development of substance abuse. Pers Psychol Eur17:141.

Baker TE, Holroyd CB (2009) Which way do I go? Neural activa-tion in response to feedback and spatial processing in a virtualT-maze. Cereb Cortex 19:1708–1722.

Cohen JD, Braver TS, Brown JW (2002) Computational perspec-tives on dopamine function in prefrontal cortex. Curr OpinNeurobiol 12:223–229.

Cohen MX, Elger CE, Ranganath C (2007) Reward expectationmodulates feedback-related negativity and EEG spectra. Neu-roimage 35:968–978.

Cohen MX, Frank MJ (2008) Neurocomputational models ofbasal ganglia function in learning, memory and choice. BehavBrain Res 199:141–156.

Conrod PJ, Woicik P (2002) Validation of a four-factor model ofpersonality risk for substance abuse and examination of abrief instrument for assessing personality risk. Addict Biol7:329.

Di Chiara G, Imperato A (1988) Drugs abused by humans pref-erentially increase synaptic dopamine concentrations in themesolimbic system of freely moving rats. Proc Natl Acad Sci US A 85:5274–5278.

Frank MJ, D’Lauro C, Curran T (2007) Cross-task individualdifferences in error processing: neural, electrophysiologicaland genetic components. Cogn Affect Behav Neurosci 7:297–308.

Frank MJ, Seeberger LC, O’reilly RC (2004) By carrot or by stick:cognitive reinforcement learning in parkinsonism. Science306:1940–1943.

Frank MJ, Woroch BS, Curran T (2005) Error-related negativitypredicts reinforcement learning and conflict biases. Neuron47:495–501.

Holroyd CB, Coles MG (2002) The neural basis of human errorprocessing: reinforcement learning, dopamine, and the error-related negativity. Psychol Rev 109:679–709.

Holroyd CB, Krigolson OE (2007) Reward prediction errorsignals associated with a modified time estimation task. Psy-chophysiology 44:913–917.

Holroyd CB, Krigolson OE, Baker R, Lee S, Gibson J (2009) Whenis an error not a prediction error? An electrophysiologicalinvestigation. Cogn Affect Behav Neurosci 9:59–70.

Holroyd CB, Pakzad-Vaezi KL, Krigolson OE (2008) The feedbackcorrect-related positivity: sensitivity of the event-related brainpotential to unexpected positive feedback. Psychophysiology45:688–697.

Holroyd CB, Yeung N (2003) Alcohol and error processing.Trends Neurosci 26:402–404.

Homayoun H, Moghaddam B (2006) Progression of cellularadaptations in medial prefrontal and orbitofrontal cortex inresponse to repeated amphetamine. J Neurosci 26:8025–8039.

Humeniuk R, Ali R (2006) Validation of the Alcohol, Smokingand Substance Involvement Screening Test (ASSIST) and PilotBrief Intervention: a Technical Report of Phase II Findings ofthe WHO ASSIST Project. WHO Library Cataloguing-in-Publication Data.

Hyman SE (2005) Addiction: a disease of learning and memory.Am J Psychiatry 162:1414–1422.

Hyman SE (2007) The neurobiology of addiction: implicationsfor voluntary control of behavior. Am J Bioeth 7:8–11.

Hyman SE, Malenka RC, Nestler EJ (2006) Neural mechanismsof addiction: the role of reward-related learning and memory.Annu Rev Neurosci 29:565–598.

Jocham G, Ullsperger M (2009) Neuropharmacology of perfor-mance monitoring. Neurosci Biobehav Rev 33:48–60.

Kessler RC, Chiu WT, Demler O, Merikangas KR, Walters EE(2005) Prevalence, severity, and comorbidity of 12-monthDSM-IV disorders in the National Comorbidity Survey Repli-cation. Arch Gen Psychiatry 62:617–627.

Klein TA, Neumann J, Reuter M, Hennig J, von Cramon DY,Ullsperger M (2007) Genetically determined differences inlearning from errors. Science 318:1642–1645.

Koob GF, Le Moal M (2005) Plasticity of reward neurocircuitryand the ‘dark side’ of drug addiction. Nat Neurosci 8:1442–1444.

Kreek MJ, Nielsen DA, Butelman ER, LaForge KS (2005) Geneticinfluences on impulsivity, risk taking, stress responsivity andvulnerability to drug abuse and addiction. Nat Neurosci8:1450–1457.

Marco-Pallares J, Cucurell D, Cunillera T, Kramer UM, Camara E,Nager W, Bauer P, Schüle R, Schöls L, Münte TF, Rodriguez-Fornells A. (2009) Genetic variability in the dopamine system



(dopamine receptor D4, catechol-O-methyltransferase) modu-lates neurophysiological responses to gains and losses. BiolPsychiatry 2:154–161.

Markou A, Kosten TR, Koob GF (1998) Neurobiological similari-ties in depression and drug dependence: a self-medicationhypothesis. Neuropsychopharmacology 18:135–174.

McClure SM, Daw ND, Montague PR (2003) A computationalsubstrate for incentive salience. Trends Neurosci 26:423–428.

Miller EK, Cohen JD (2001) An integrative theory of prefrontalcortex function. Annu Rev Neurosci 24:167–202.

Miltner WHR, Braun CH, Coles MGH (1997) Event-related brainpotentials following incorrect feedback in a time-estimationtask: evidence for a ‘generic’ neural system for error detection.J Cogn Neurosci 9:788–798.

Newcombe DA, Humeniuk RE, Ali R (2005) Validationof the World Health Organization Alcohol, Smoking andSubstance Involvement Screening Test (ASSIST): report ofresults from the Australian site. Drug Alcohol Rev 24:217–226.

Overbeek JM, Nieuwenhuis S, Ridderinkhof KR (2005) Disso-ciable components of error processing: on the functional sig-nificance of the Pe Vis-à-vis the ERN/Ne. J Psychophysiology19:319–329.

Polich J, Criado JR (2006) Neuropsychology and neuropharma-cology of P3a and P3b. Int J Psychophysiol 60:172–185.

Redish AD (2004) Addiction as a computational process goneawry. Science 306:1944–1947.

Rehm J, Taylor B, Room R (2006) Global burden of disease fromalcohol, illicit drugs and tobacco. Drug Alcohol Rev 25:503–513.

Rice ME, Cragg SJ (2004) Nicotine amplifies reward-relateddopamine signals in striatum. Nat Neurosci 7:583–584.

Robinson TE, Kolb B (2004) Structural plasticity associated withexposure to drugs of abuse. Neuropharmacology 47 (Suppl.1):33–46.

Schultz W (1998) Predictive reward signal of dopamineneurons. J Neurophysiol 80:1–27.

Stephens RS, Curtin L (1995) Alcohol and depression: effects onmood and biased processing of self-relevant information.Psychol Addict Behav 9:211.

Stockwell T, Murphy D, Hodgson R (1983) The severity ofalcohol dependence questionnaire: its use, reliability andvalidity. Br J Addict 78:145–155.

Volkow ND, Fowler JS, Wang GJ, Baler R, Telang F (2009)Imaging dopamine’s role in drug abuse and addiction. Neu-ropharmacology 56 (Suppl. 1):3–8.

Wheeler EZ, Fellows LK (2008) The human ventromedial frontallobe is critical for learning from negative feedback. Brain131:1323–1331.

SUPPORTING INFORMATION

Additional Supporting Information may be found in theonline version of this article.

Appendix S1 Materials and Methods.Figure S1 Probabilistic Learning Task.Figure S2 Substance Preference.

Please note: Wiley-Blackwell are not responsible for thecontent or functionality of any supporting materials sup-plied by the authors. Any queries (other than missingmaterial) should be directed to the corresponding authorfor the article.



Copyright of Addiction Biology is the property of Wiley-Blackwell and its content may not be copied or

emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.

However, users may print, download, or email articles for individual use.

Durgs consumption

Documents

Transcript of Durgs consumption