Gambling severity predicts midbrain response to near-miss outcomes (2010)

J Neurosci. 2010 May 5;30(18):6180-7. doi: 10.1523/JNEUROSCI.5758-09.2010.

Chase HW1, Clark L.

Author information

  • 1School of Psychology, University of Nottingham, Nottingham NG7 2RD, United Kingdom.

Abstract

Gambling is a common recreational activity that becomes dysfunctional in a subset of individuals, with DSM ‘pathological gambling’ regarded as the most severe form. During gambling, players experience a range of cognitive distortions that promote an over-estimation of the chances of winning. Near-miss outcomes are thought to fuel these distortions. We observed previously that near-misses recruited overlapping circuitry to monetary wins in a study in healthy volunteers (Clark et al. 2009).

The present study sought to extend these observations in regular gamblers and relate brain responses to an index of gambling severity. Twenty regular gamblers, who varied in their involvement from recreational players to probable pathological gamblers, were scanned whilst performing a simplified slot-machine task that delivered occasional monetary wins, as well as near-miss and full-miss non-win outcomes. In the overall group, near-miss outcomes were associated with a significant response in the ventral striatum, which was also recruited by monetary wins. Gambling severity, measured with the South Oaks Gambling Screen, predicted a greater response in the dopaminergic midbrain to near-miss outcomes. This effect survived controlling for clinical co-morbidities that were present in the regular gamblers. Gambling severity did not predict win-related responses in the midbrain or elsewhere.

These results demonstrate that near-miss events during gambling recruit reward-related brain circuitry in regular players. An association with gambling severity in the midbrain suggests that near-miss outcomes may enhance dopamine transmission in disordered gambling, which extends neurobiological similarities between pathological gambling and drug addiction.

Keywords: Gambling, Cognitive, Addiction, Dopamine, Striatum, Midbrain

Introduction

Gambling is a form of entertainment that can become dysfunctional in some individuals: ‘pathological gambling’ is a DSM-IV Impulse Control Disorder (American Psychiatric Association, 2000) with symptoms that include withdrawal and tolerance (Potenza, 2006). Accumulating data indicate neurobiological alterations in the brain reward system in problem gamblers (Reuter et al., 2005, Tanabe et al., 2007, Potenza, 2008). For example, an fMRI study using a guessing task with monetary wins and losses found an attenuation of win-related activity in the ventral striatum and medial prefrontal cortex (PFC) of pathological gamblers (Reuter et al., 2005). Similar changes have been described in drug abusers (Goldstein et al., 2007, Wrase et al., 2007), and are thought to indicate dysregulation of dopaminergic input to these structures. Dopaminergic involvement in gambling is supported by reports of problem gambling as a medication side-effect in patients with Parkinson’s Disease (Dodd et al., 2005, Steeves et al., 2009).

Neuroimaging investigations of problem gambling to date have neglected the complex cognitions that gamblers frequently experience (Ladouceur and Walker, 1996). On games of chance like roulette or lottery, gamblers often misperceive some level of skill involvement (the ‘illusion of control’) (Langer, 1975). These cognitive distortions are more prevalent in problem gamblers (Miller and Currie, 2008) and are directly fostered by certain features of gambling games (Griffiths, 1993), including the presence of near-misses: non-win outcomes that are proximal to a jackpot. Near-misses are able to promote continued gambling despite their objective non-win (loss) status (Kassinove and Schare, 2001, Cote et al., 2003). The neural mechanisms that underpin near-miss effects have broader relevance to the understanding of reinforcement learning: on games of skill (e.g. soccer), near-misses (e.g. hitting the post) provide a valid signal of skill acquisition and imminent reward, and thus a reinforcement learning system may usefully assign value to these outcomes. However, in games of chance, near-misses do not indicate future success, and their potency suggests that gambling games may harness brain mechanisms that naturally handle skill situations (Clark, 2010).

Using a slot-machine task in healthy volunteers, we found that near-misses were associated with significant activity in brain regions (ventral striatum, anterior insula) that responded to monetary wins (Clark et al., 2009). The present study aimed to extend these observations in a group of regular gamblers. First, we aimed to corroborate our finding that near-miss outcomes would recruit components of the brain reward system in regular gamblers. Second, we sought to identify areas within this system where brain activity during gambling was associated with gambling severity. Although previous fMRI studies have explored problem gambling using case-control designs, it is increasingly recognised that disordered gambling is dimensional in nature: gamblers who do not meet DSM criteria frequently describe obvious gambling-related harms (e.g. debt, interpersonal conflict), and these harms increase steadily with gambling involvement (e.g. gambling frequency or expenditure) (Currie et al., 2006). To reflect this continuum of disordered gambling, we used voxel-wise regression to identify brain areas where win- and near- miss- related activity was predicted by individual variation in gambling severity.

Methods

Participants

Regular gamblers (n=24, 3 female) were recruited via advertisement. Four subjects were excluded from analysis due to excessive movement during scanning, leaving a reported group size of 20 (2 female). Subjects attended an fMRI scanning session at the Wolfson Brain Imaging Centre, Cambridge, U.K.. The protocol was approved by the Norfolk & Norwich Research Ethics Committee (COREC 06/Q0101/69) and all volunteers provided written informed consent. Volunteers were reimbursed £40 for participation, and had the opportunity to win further money on the task (unbeknownst to subjects, this was a fixed amount of £15).

Gambling behaviour was assessed using the SOGS (Lesieur and Blume, 1987), a 16-item self-report scale assessing core symptoms and negative consequences of gambling (e.g. loss-chasing, borrowing money, lying about gambling, family conflict). Prior to the scanning session, subjects attended a screening session involving a structured psychiatric interview with a postdoctoral psychologist (Structured Clinical Interview for DSM-IV Axis I Disorders; SCID) (First et al., 1996). Given the high co-morbidity between problem gambling and other mental health problems (Kessler et al., 2008), we opted to tolerate psychiatric co-morbidities to avoid over-selection of a clinically-unrepresentative sample. Co-morbidities were as follows: current dysthymia and/or drug-related mood disorder (n=5), lifetime major depressive disorder (n=4), current bipolar disorder (n=1), anxiety or panic disorder (n=2), lifetime drug dependence (n=3), current alcohol/drug abuse (n=8), current alcohol dependence (n=1). Three subjects were currently receiving psychotropic medication (antidepressant n=2, benzodiazepine n=1). In addition, urinalysis (SureStep™, Bedford, UK) on the day of the fMRI scan detected positive tests for cannabis (THC) in 4 participants. Self-report questionnaire measures were used to quantify current psychiatric symptoms: Beck Depression Inventory (version 2) (Beck et al., 1996), Beck Anxiety Inventory (BAI) (Beck et al., 1988), Adult ADHD Self-Report Scale (Kessler et al., 2005), Padua Inventory for OCD symptoms (Burns et al., 1996), and the Alcohol Use Questionnaire (AUQ) (Townshend and Duka, 2002).

Procedure

During the fMRI scan, subjects completed 3 blocks of 60 trials on the Slot Machine Task (Clark et al., 2009), lasting approximately 45 minutes. Subjects were practiced on the task (10 trials with 2 hypothetical wins) prior to entering the scanner, and during scanning, responses were registered using a button box. Trial structure and display screen are displayed in Figure 1. On each trial, two reels are presented on the screen, with a horizontal ‘payline’ visible centrally. Six icons are displayed on each reel in the same order. The six icons were selected from 16 alternatives at the start of the scan task, to heighten a sense of involvement.

Figure 1 

Task Design. The slot machine task presents two reels, with the six identical play icons displayed on each reel, and a horizontal ‘payline’ across the centre of the screen. On trials with a white screen background (as displayed), the volunteer …

Each trial proceeded as follows: during the selection phase, one of the six icons was selected on the left reel (selection phase; 5s fixed duration). Following selection, the right reel was spun for 2.8-6s (anticipation phase), and decelerated to a standstill, beginning the outcome phase (4s fixed). At the end of each trial, there was an inter-trial interval of variable duration (2-7s). In the outcome phase, if the right reel stopped on the selected icon (i.e. matching icons were displayed in the payline), a £0.50 win was delivered; all other outcomes won nothing. Trials where the right reel stopped one position above or below the payline were designated ‘near-misses’. Non-win trials where the reel stopped in one of the three remaining positions (i.e. more than one position from the payline) were designated ‘full-misses’. During the selection phase, on trials with a white screen background, the participant selected the play icon using two buttons to scroll through the shapes, and a third button to confirm selection (participant-chosen trials) within the 5s window. On trials with a black screen background, the computer selected the play icon, and the subject was required to confirm the selection by pressing the third button within the 5s window (computer-chosen trials). Participant-chosen (n=90) and computer-chosen trials (n=90) were presented in a fixed pseudo-random order. If selection/ confirmation was not completed within the 5s window, a “Too Late!” message was presented, followed by the inter-trial interval. Outcomes were pseudo-randomized to ensure a fair number of wins (1/6, total 30 = £15), near-misses (2/6, total 60) and full-misses (3/6, total 90).

On 1/3 trials, subjective ratings were acquired at two points during the trial, using onscreen 21-point visual analogue scales. Following selection, subjects rated “How do you rate your chances of winning?” and following the outcome, subjects rated ”How much do you want to continue to play the game?”. No time limit was imposed for the subjective ratings. Data from the subjective ratings were converted to a standardised z score, based on each individual’s mean and standard deviation for that rating, to account for the variability in anchoring across subjects. Subjective ratings were analysed using paired t tests (for ‘chances of winning’) and repeated-measures analysis of variance (for ‘continue to play’) with outcome (3 levels: win, near-miss, full-miss) and control (2 levels: participant-chosen, computer-chosen) as factors.

Imaging Procedure

Scanning was performed on a Siemens TimTrio 3 Tesla magnet using a 32 slice axial oblique sequence, with a repetition time of 2s (TE 30ms, flip angle 78°, voxel size 3.1 × 3.1 × 3.0mm, matrix size 64 × 64, field of view 201mm × 201mm, bandwidth 2232Hz/Px). Three runs of 60 trials were completed (630 repetitions), with 6 dummy scans discarded at the start of each run to allow for equilibrium effects. A high-resolution magnetization-prepared rapid acquisition gradient-echo sequence (MP-RAGE) structural image was acquired after the functional runs for use in the spatial normalization.

FMRI data were analyzed using SPM5 (Statistical Parametric Mapping, Wellcome Department of Cognitive Neurology, London, UK). Preprocessing consisted of slice timing correction, within-subject realignment, spatial normalization, and spatial smoothing using a 10mm Gaussian kernel. Subjects’ movement parameters were screened for excessive movement (defined as >5mm within a run), resulting in exclusion of 4 participants (1 female) from all analysis. Time series were high pass filtered (128s). Volumes were normalized to the International Consortium for Brain Mapping (ICBM) templates that approximate to Talairach & Tournoux (1988) space, using a matrix calculated by normalizing the segmented MP-RAGE structural image for each subject onto the ICBM grey and white matter templates.

A canonical hemodynamic response function (HRF) was modelled to the onsets of the selection phase, the anticipation phase and the outcome phase on each trial, in order to minimize unexplained variance in the design matrix. To analyze outcome-related brain responses, the events were classified into 8 trial types, comprising a 2 (choice: participant-chosen, computer-chosen) by 4 (win, near-miss before the payline, near-miss past the payline, full-miss) factorial design. Movement parameters from realignment were included in the design matrix as covariates of no interest. The HRF was used as a covariate in a general linear model, and a parameter estimate was obtained for each voxel, for each event type, reflecting the strength of covariance between the data and the canonical HRF. Contrast images were calculated between parameter estimates from different trial types, and individual contrast images were then taken to a second-level random-effects group analysis.

Four contrasts were computed to assess the outcome-related brain responses in the overall group of regular gamblers: 1) All monetary wins (i.e. both participant- and computer- chosen trials) minus all non-win outcomes. 2) Near-misses (on both participant- and computer- chosen trials) minus full-miss outcomes (on both participant- and computer-chosen trials). 3) Near-miss by personal control interaction: areas differentially recruited by near-misses compared to full-misses as a function of participant versus computer control (ie. 1, −1, −1, 1). 4) Win activity on participant-chosen trials minus win activity on computer-chosen trials. To explore these effects as a function of gambling severity, voxel-wise univariate regressions were run using the SOGS score as a predictor variable. Given our a priori hypotheses about the role of brain reward circuitry in gambling distortions and problem gambling, we implemented the win contrast (all monetary wins minus all non-wins, thresholded at pFWE<.05 corrected) from our previous study (Clark et al., 2009) as a mask for these contrasts, as well as the regression analyses, using the PickAtlas tool (Maldjian et al., 2003). These region-of-interest analyses were thresholded at p<.05 corrected for multiple comparisons using random field theory (Worsley et al., 1996), i.e. Family Wise Error (FWE) corrected, with a cluster threshold of 10 voxels to reduce the rate of false positives (Forman et al., 1995). This cluster threshold was selected on the grounds that the smallest region of a priori interest (midbrain substantia nigra / ventral tegmental area) has an estimated size of 20-25 voxels (Duzel et al., 2009). Signal change was extracted from activated foci using the MARSBAR tool (Brett et al., 2002) for the purposes of plotting the data. Whole-brain analyses are also presented using an exploratory threshold of p<.001 uncorrected.

Results

Variation in Gambling Severity

The regular gamblers were predominantly male (n=18) with a mean age of 33.7 (sd 1.8), mean years of education of 14.5 (sd 0.5) and mean NART-estimated full IQ of 111.5 (sd 7.3). The preferred form of gambling in the group was off-course sports betting (horse-racing or football), but slot machines, cards and lotteries were also common (see Supplementary Table 1). All but one subject were currently active gamblers, playing at least once a week on their preferred form of gambling; the participant who was no longer gambling had been abstinent for one year. Thirteen of the group met the SOGS threshold of >=5 for probable Pathological Gambling (overall range 0-20, mean 7.25, median 6.5) (see Supplementary Figure 1). Maximum expenditure in a single day varied from £10-£100 (n=5), £100-£1000 (n=8), £1000-£10,000 (n=5), to more than £10,000 (n=2). Descriptive data for the questionnaire measures of clinical symptoms are reported in Supplementary Table 2.

Subjective Ratings during the Slot Machine Task

The post-selection ratings of “How do you rate your chances of winning?” were significantly higher on participant-chosen trials compared to computer-chosen trials (t(19)=5.2, p<0.001). This effect of personal control was attenuated as a function of gambling severity as measured by the SOGS (r20=−0.53, p=0.016). The post-outcome ratings of “How much do you want to continue to play?” were analysed using two-way ANOVA to reveal a main effect of feedback (F(2,38)=40.179, p<0.001), no main effect of agency (F(1,19)<1), and an agency by feedback interaction (F(2,38)=3.604, p=0.037) (see Supplementary Table 3). Participant-chosen wins were rated more highly than computer-chosen wins (t(19)=2.199, p=0.040), but personal control did not influence the ratings for near-miss (t(19)=−1.272, p=0.217) or full-miss (t(19)=−0.998, p=0.331) outcomes. ‘Continue to play’ ratings were higher after wins compared to either kind of non-win, regardless of personal control (t(19)>3.889, p<0.002 in all cases), whereas near misses and full misses did not differ on participant-chosen trials (t(19)=1.104, p=0.283) or computer-chosen trials (t(19)<1). Thus, there was no detectable effect of the near-miss outcomes on self-report ratings in the regular gamblers.

fMRI Responses to Gambling Outcomes

Brain regions sensitive to unpredictable monetary wins were identified by contrasting all winning outcomes against all non-win outcomes, within an independent ROI defined from the win contrast in our previous study (Clark et al., 2009). Significant signal change was observed in a number of areas linked to reward and reinforcement learning: right ventral striatum (putamen) (peak voxel: x, y, z = 20, 10, −6, Z=3.66, 133 voxels, pFWE=.029) and thalamus (x, y, z = 2, −6, 2; Z=4.71, 14 voxels, pFWE=.001), with subthreshold foci in the left ventral striatum (x, y, z = −16, 2, −6, Z=3.39, pFWE=.065), anterior insula bilaterally (x, y, z = 28, 20, −6, Z=3.46, pFWE=.054; x, y, z = 36, 16, −8, Z=3.36, pFWE=.070; x, y, z = −36, 18, −6, Z=3.47, pFWE=.052), and midbrain proximal to substantia nigra / ventral tegmental area (SN/VTA) (x, y, z = −8, −20, −14, Z=3.36, pFWE=.071) (visible in Figure 2A, thresholded at p<.001 for display purposes). An independent contrast assessed brain responses to near-miss outcomes, compared to full-misses. There was significant signal change in the right ventral striatum (putamen) (x, y, z = 18, 6, −2, Z=3.67, 52 voxels, pFWE=.032) and the left parahippocampal gyrus (BA 28) bordering on the striatum (x, y, z = −16, −2, −10, Z=4.32, 27 voxels, pFWE=.003) (see Figure 2B). The contrasts of participant-chosen wins minus computer-chosen wins, and the interaction contrast for near-miss activity as a function of personal control, did not yield any significant activation within the ROI mask.

Figure 2 

A) Win-related activation (win > non-win outcomes) in the regular gamblers, using a region of interest mask of win activity from an independent sample (Clark et al. 2009). Activity is displayed at p<0.001 uncorrected, k=10, to illustrate …

Effects of Gambling Severity of fMRI Responses to Gambling Outcomes

Gambling severity (SOGS score) was entered as a single regressor in the contrast of monetary wins minus all non-wins, using the win-sensitive ROI mask. There were no significant voxels where SOGS score predicted either increases or decreases in win-related activity. However, a regression analysis for the near-miss minus full-miss contrast indicated that SOGS gambling severity was positively related to the brain response to near-miss outcomes in the midbrain (48 voxels: x, y, z = −6, −18, −16, Z=4.99, pFWE<.001; x, y, z = 10, −18, −12, Z=3.90, pFWE=.014) (see Figure 3). In addition, we also observed gambling severity was negatively related to the brain response to near-miss outcomes in the left caudate (x, y, z = −12, 8, 6, Z=3.91, 11 voxels, pFWE=.013). This cluster lay at the dorsal tip of the ROI, overlapping the internal capsule, and we could not identify win- (contrast 1) or near-miss- (contrast 2) related activity at this focus in the present dataset, even at a liberal threshold (p<.005 uncorrected). Furthermore, the extracted signal from ventral striatum and midbrain clusters were positively correlated on both win (r20=0.72, p<.001) and near-miss outcomes (r20=0.43, p=.06), as seen in previous studies (D’Ardenne et al., 2008, Schott et al., 2008, Kahnt et al., 2009). Therefore, although this caudate peak met our significance threshold, we are cautious about inferring a role for this region in gambling near-misses.

Figure 3 

A) Effect of gambling severity (South Oaks Gambling Screen; SOGS) on near-miss related activation, within the region of interest mask (displayed at p<0.001 uncorrected, k=10). B) Extracted signal for the near-miss minus full-miss contrast in the …

The smoothing kernel (10mm) implemented in our primary analysis limited our ability to resolve activation within the midbrain. We re-modelled the fMRI data using a smaller 4mm smoothing kernel. In a whole-brain analysis using an exploratory threshold (p<.001 uncorrected), two activations in the midbrain (x = −8, y = −18, z = −18, Z = 3.37, p<0.001; x = 12, y = −16, z = −12, Z = 3.28, p=0.001) reflected the effect of SOGS gambling severity on near-miss related activation (displayed in Figure 4A at a threshold of p<.005 uncorrected). These activations are consistent with a compound SN/VTA signal (Duzel et al., 2009).

Figure 4 

A) The association between gambling severity (SOGS score) and near-miss related activation (near-miss minus full-miss) in the midbrain (z=−18 and z=−12), using a smaller (4mm) smoothing kernel. Activity thresholded at p<0.005 uncorrected …

The regular gamblers displayed a number of clinical co-morbidities that covaried moderately with their gambling severity. To examine whether the midbrain association was specifically associated with gambling severity rather than these co-morbidities, we included continuous measures of depression (BDI), anxiety (BAI), ADHD symptomatology (ASRS), impulsivity (BIS), OCD symptoms (Padua scale) and alcohol use/abuse (AUQ scale) as additional covariate regressors in the SOGS regression. In each case, the midbrain activation (peak voxel: x = −6, y = −18, z = −16) for the SOGS association was detectable with a Z statistic between 2.20-2.56 (p=.014 to p=.005 uncorrected). In contrast, the negative association between SOGS and near-miss related activity in the caudate did not survive controlling for depressive (BDI) and OCD (Padua scale) symptoms, at a liberal threshold of p<0.05 uncorrected.

These data indicate that in a correlational design, a stronger midbrain response to near-miss outcomes was associated with disordered gambling. Previous case-control studies of pathological gamblers indicate an overall attenuation of reward-related activity (Reuter et al., 2005). To investigate this apparent discrepancy, we conducted a post-hoc between-groups analysis comparing the overall brain response to reward (wins minus non-win outcomes) in our regular gamblers against the non-gambling volunteers from our previous study (Clark et al., 2009). This was conducted as a whole-brain analysis using an exploratory significance threshold (p<.001 uncorrected). Consistent with Reuter et al, the regular gamblers displayed a weaker response to monetary wins in several reward-sensitive regions including the striatum and rostral anterior cingulate cortex (see Figure 4B and Supplementary Table 5), after covarying for group differences in age. There were no overall group differences in the near-miss response. A mixed-model ANOVA of the subjective ratings data in the regular gamblers and healthy non-gamblers revealed no significant group differences, although notably, in the pooled group (n=34) there was a marginally significant effect of participant-chosen near-miss outcomes to increase ratings of ‘Continue to play’ (t(33)=1.87, p=.07) relative to participant-chosen full-misses (see Supplementary Material and Supplementary Table 6).

Discussion

The present study investigated brain responses during a computerised slot machine task in a group of regular gamblers who varied in their involvement from recreational, social players to moderately-severe probable pathological gamblers. Unpredictable monetary wins on the task recruited a network of reward-sensitive regions including the ventral striatum. Our task further enabled the direct comparison of near-miss non-wins against full-miss non-wins, and this contrast revealed a response to near-misses in striatal regions also responsive to wins, despite the objective non-win status of these outcomes. This analysis in regular gamblers extends our recent findings in healthy volunteers with modest gambling involvement (Clark et al., 2009), highlighting the recruitment of brain reward circuitry by near-miss outcomes. The specific aim of the present study was to associate these fMRI responses with individual variation in gambling severity, in order to examine the relevance of these responses to an emerging literature on the neurobiology of problem gambling (Reuter et al., 2005, Potenza, 2008). Scores on our index of gambling severity (SOGS) ranged from 0 to 19 (see Supplementary Figure 1), with a score of 5 indicative of probable pathological gambling. This emphasises the continuous nature of gambling harms in the non-clinical population (Currie et al., 2006), and indicates that a regression-based method of analysis is appropriate for exploring neural markers of disordered gambling. While the SOGS score was not associated with the brain response to monetary wins, gambling severity was predicted by the neural response to near-miss outcomes, in the midbrain. This activation was proximal to the dopaminergic nuclei in the SN/VTA, a finding that was further substantiated by a re-analysis of our data using a smaller (4mm) smoothing kernel (Bunzeck and Duzel, 2006, D’Ardenne et al., 2008, Murray et al., 2008, Shohamy and Wagner, 2008, Duzel et al., 2009). Moreover, the association between midbrain activity to near-misses and gambling severity was not readily explained by other clinical symptoms (depression, impulsivity, OCD, alcohol use) that are moderately prevalent in regular gamblers (Kessler et al., 2008).

The observed midbrain association is consistent with the role of dopamine transmission in disordered gambling, indicated by previous studies of peripheral markers (Bergh et al., 1997, Meyer et al., 2004) and the phenomenon of medication-induced pathological gambling in Parkinson’s Disease (Dodd et al., 2005, Steeves et al., 2009). This syndrome is particularly linked to D3-preferent dopamine agonist medications, and it is notable that D3-receptors are abundant in human SN (Gurevich and Joyce, 1999). The ability of near-miss outcomes to enhance dopamine transmission in more severe problem gamblers could underlie the potency of these outcomes to invigorate gambling (Kassinove and Schare, 2001, Cote et al., 2003, Clark et al., 2009). Electrophysiological studies recording from midbrain neurons have demonstrated the well-known role for this system in signalling reward and coding reward prediction errors (Schultz, 2002, Montague et al., 2004). Human neuroimaging studies confirm midbrain BOLD responses in monetary reward tasks (e.g. Bjork et al., 2004, D’Ardenne et al., 2008, Schott et al., 2008), which correlate with a direct index of striatal dopamine release ([11C]raclopride displacement) (Schott et al., 2008). Certainly, it is likely that reward prediction errors were engendered on near-miss trials in the current task: a positive prediction error occurs as the reel decelerates and the subject anticipates a winning outcome. This is immediately followed by a negative prediction error, as the reel stops one position from the winning payline. Recent data indicate that midbrain BOLD signal may be particularly aligned with positive prediction errors (D’Ardenne et al., 2008), consistent with gamblers’ more general style of over-estimating their chances of winning (Ladouceur & Walker 1996). Two further aspects of midbrain firing seen in electrophysiological data may be relevant to the current fMRI findings. First, midbrain neurons display generalisation, where they fire to stimuli that are similar to those predictive of reward (Tobler et al., 2005, Shohamy and Wagner, 2008). It is a testable hypothesis that problem gamblers show excessive generalisation of reward-predictive stimuli, mediated by midbrain hyper-reactivity. Second, midbrain neurons may show adaptive coding within a task, where their maximal response is scaled to the available reward (Tobler et al., 2005). This may explain why we did not observe a midbrain association with gambling severity on winning outcomes, despite an overall midbrain response to wins. However, we did not explicitly demonstrate a significant difference in the strength of the SOGS-midbrain association on near-miss and win trials. From the positive trend-line in Figure 3C, it is conceivable that a SOGS-midbrain association may be detectable for winning outcomes in a larger sample.

A previous case-control study in pathological gamblers reported reduced BOLD signal in ventral striatum and medial PFC in response to monetary wins (Reuter et al., 2005). This finding was interpreted as evidence for a reward deficiency account of pathological gambling, where a hypoactive reward system confers vulnerability to a range of addictions (Bowirrat and Oscar-Berman, 2005). The task employed in the Reuter et al. study was a simple two-choice guessing task that is unlikely to elicit the complex distortions of probability and skill perception that are central to gambling behavior (Ladouceur and Walker, 1996, Clark, 2010). We conducted a between-groups analysis comparing the regular gamblers from the present study against volunteers with modest gambling involvement from our previous study (Clark et al., 2009). Although the circuitry recruited by monetary wins was strikingly similar across the two groups, the regular gamblers displayed an attenuated response to winning that was significant in the ventral striatum and medial PFC, corroborating Reuter et al (2005). Critically, the present data show that this state of overall reward deficiency is coupled with excessive recruitment of brain reward circuitry under conditions of cognitive distortion (near-misses), which varies as a function of gambling severity. It is likely that these two effects cancelled out in the between-groups comparison of near-miss activity, where no differences were observed.

Two further points of comparison with our previous study are noteworthy. First, our earlier study reported an interaction between near-misses and personal control in the medial PFC (Clark et al., 2009). We were unable to substantiate this interaction effect in the regular gamblers. Indeed, the regular gamblers did not display significant recruitment of this region even in the basic win contrast, and neuropsychological studies indicate specific impairments on probes of medial PFC integrity in problem gamblers (Goudriaan et al., 2006, Lawrence et al., 2009). Our previous study also hinted at a key role of the insula in the motivational effect of near-misses. In the present study, insula activations were restricted to the overall win contrast, at a level just below FWE significance, and these responses did not covary with gambling severity. We believe these insula responses convey information about peripheral physiology (e.g. heart rate increases) during gambling (e.g. Craig, 2003), and it may be harder to engender this arousal in regular gamblers who have much experience with high-stimulation games. Psychophysiological studies in regular players have shown qualitative differences between gambling in laboratory settings versus naturalistic (e.g. casino) settings (Anderson and Brown, 1984, Meyer et al., 2004). Future work combining fMRI and psychophysiological monitoring is needed to assess the relationship between evoked arousal and brain activity during gambling (c.f. Critchley et al., 2001).

Some limitations of the current study should be noted. First, while we covaried for several common co-morbidities, some relevant conditions including nicotine dependence and personality disorders (Cunningham-Williams et al., 1998) were not assessed. Second, the between-groups comparison against our earlier study was not planned and the groups were not well matched for age and gender. We covaried for age but not gender, as our group of regular gamblers was almost exclusively male. Disordered gambling is more prevalent in men (Kessler et al., 2008), but further studies are required to test whether our effects generalise to female gamblers. Third, the self-report ratings did not demonstrate a significant subjective effect of the near-misses in the regular gamblers. This is likely an issue of statistical power given the fragility of visual analogue ratings: in our previous study, the subjective effects were observed in a larger behavioral experiment in 40 volunteers. A marginally significant effect of the (participant-chosen) near-misses to increase motivation to play was observed in a pooled analysis of the two fMRI datasets (n=34, see Supplementary Table 6). Finally, our inference that dopamine is involved in gambling near-misses must be treated with a suitable degree of caution given the indirect nature of the BOLD signal and limited spatial resolution of fMRI (see Duzel et al., 2009 for review). Other neurotransmitters implicated in gambling behavior, including serotonin, are present in the midbrain, and are modulated by motivational stimuli, albeit without phasic responses (Nakamura et al., 2008). Pharmacological challenge designs will be needed to explore these questions directly; for example, Zack & Poulos (2004) reported that the indirect dopamine agonist, amphetamine, increased urges to gamble and attentional bias in problem gamblers. One clinical implication of such findings is that drugs that reduce dopamine transmission may have therapeutic benefit in reducing cognitive distortions in problem gamblers.

Supplementary Material

Supp1

Acknowledgements

Supported by a project grant from the Economic and Social Research Council and Responsibility in Gambling Trust to LC and TW Robbins (RES-164-25-0010). Completed within the Behavioural and Clinical Neuroscience Institute, supported by a consortium award from the Medical Research Council (U.K.) and the Wellcome Trust. We are grateful to the participants, and the radiographic staff at the Wolfson Brain Imaging Centre, Cambridge, U.K.

References

  1. American Psychiatric Association . Diagnostic and statistical manual of mental disorders – Text revision. 4th ed. American Psychiatric Association; Washington, DC: 2000.
  2. Anderson G, Brown RI. Real and laboratory gambling, sensation-seeking and arousal. Br J Psychol. 1984;75:401–410. [PubMed]
  3. Beck AT, Epstein N, Brown G, Steer RA. An inventory for measuring clinical anxiety: psychometric properties. J Consult Clin Psychol. 1988;56:893–897. [PubMed]
  4. Beck AT, Steer RA, Brown GK. Manual for Beck Depression Inventory-II. Psychological Corporation; San Antonio, TX.: 1996.
  5. Bergh C, Eklund T, Sodersten P, Nordin C. Altered dopamine function in pathological gambling. Psychol Med. 1997;27:473–475. [PubMed]
  6. Bjork JM, Knutson B, Fong GW, Caggiano DM, Bennett SM, Hommer DW. Incentive-elicited brain activation in adolescents: similarities and differences from young adults. J Neurosci. 2004;24:1793–1802. [PubMed]
  7. Bowirrat A, Oscar-Berman M. Relationship between dopaminergic neurotransmission, alcoholism, and Reward Deficiency syndrome. Am J Med Genet B Neuropsychiatr Genet. 2005;132:29–37. [PubMed]
  8. Brett M, Anton J-L, Valabregue R, Poline J-B. Region of interest analysis using an SPM toolbox [abstract] NeuroImage. 2002;16
  9. Bunzeck N, Duzel E. Absolute coding of stimulus novelty in the human substantia nigra/VTA. Neuron. 2006;51:369–379. [PubMed]
  10. Burns GL, Keortge SG, Formea GM, Sternberger LG. Revision of the Padua Inventory of obsessive compulsive disorder symptoms: distinctions between worry, obsessions, and compulsions. Behav Res Ther. 1996;34:163–173. [PubMed]
  11. Clark L. Decision-making during gambling: an integration of cognitive and psychobiological approaches. Philos Trans R Soc Lond B Biol Sci. 2010;365:319–330. [PMC free article] [PubMed]
  12. Clark L, Lawrence AJ, Astley-Jones F, Gray N. Gambling near-misses enhance motivation to gamble and recruit win-related brain circuitry. Neuron. 2009;61:481–490. [PMC free article] [PubMed]
  13. Cote D, Caron A, Aubert J, Desrochers V, Ladouceur R. Near wins prolong gambling on a video lottery terminal. J Gambl Stud. 2003;19:433–438. [PubMed]
  14. Craig AD. Interoception: the sense of the physiological condition of the body. Curr Opin Neurobiol. 2003;13:500–505. [PubMed]
  15. Critchley HD, Mathias CJ, Dolan RJ. Neural activity in the human brain relating to uncertainty and arousal during anticipation. Neuron. 2001;29:537–545. [PubMed]
  16. Cunningham-Williams RM, Cottler LB, Compton WM, 3rd, Spitznagel EL. Taking chances: problem gamblers and mental health disorders–results from the St. Louis Epidemiologic Catchment Area Study. Am J Public Health. 1998;88:1093–1096. [PMC free article] [PubMed]
  17. Currie SR, Hodgins DC, Wang J, el-Guebaly N, Wynne H, Chen S. Risk of harm among gamblers in the general population as a function of level of participation in gambling activities. Addiction. 2006;101:570–580. [PubMed]
  18. D’Ardenne K, McClure SM, Nystrom LE, Cohen JD. BOLD responses reflecting dopaminergic signals in the human ventral tegmental area. Science. 2008;319:1264–1267. [PubMed]
  19. Dodd ML, Klos KJ, Bower JH, Geda YE, Josephs KA, Ahlskog JE. Pathological gambling caused by drugs used to treat Parkinson’s Disease. Arch Neurol. 2005;62:1377–1381. [PubMed]
  20. Duzel E, Bunzeck N, Guitart-Masip M, Wittmann B, Schott BH, Tobler PN. Functional imaging of the human dopaminergic midbrain. Trends Neurosci. 2009;32:321–328. [PubMed]
  21. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for DSM-IV Axis I Disorders, Clinician Version. American Psychiatric Press, Inc; Washington D.C.: 1996.
  22. Forman SD, Cohen JD, Fitzgerald M, Eddy WF, Mintun MA, Noll DC. Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of a cluster-size threshold. Magn Reson Med. 1995;33:636–647. [PubMed]
  23. Goldstein RZ, Alia-Klein N, Tomasi D, Zhang L, Cottone LA, Maloney T, Telang F, Caparelli EC, Chang L, Ernst T, Samaras D, Squires NK, Volkow ND. Is decreased prefrontal cortical sensitivity to monetary reward associated with impaired motivation and self-control in cocaine addiction? Am J Psychiatry. 2007;164:43–51. [PMC free article] [PubMed]
  24. Goudriaan AE, Oosterlaan J, de Beurs E, van den Brink W. Neurocognitive functions in pathological gambling: a comparison with alcohol dependence, Tourette syndrome and normal controls. Addiction. 2006;101:534–547. [PubMed]
  25. Griffiths M. Fruit machine gambling: the importance of structural characteristics. J Gambl Stud. 1993;9:101–120.
  26. Gurevich EV, Joyce JN. Distribution of dopamine D3 receptor expressing neurons in the human forebrain: comparison with D2 receptor expressing neurons. Neuropsychopharmacol. 1999;20:60–80. [PubMed]
  27. Kahnt T, Park SQ, Cohen MX, Beck A, Heinz A, Wrase J. Dorsal striatal-midbrain connectivity in humans predicts how reinforcements are used to guide decisions. J Cogn Neurosci. 2009;21:1332–1345. [PubMed]
  28. Kassinove JI, Schare ML. Effects of the “near miss” and the “big win” on persistence at slot machine gambling. Psychology of Addictive Behaviors. 2001;15:155–158. [PubMed]
  29. Kessler RC, Adler L, Ames M, Demler O, Faraone S, Hiripi E, Howes MJ, Jin R, Secnik K, Spencer T, Ustun TB, Walters EE. The World Health Organization Adult ADHD Self-Report Scale (ASRS): a short screening scale for use in the general population. Psychol Med. 2005;35:245–256. [PubMed]
  30. Kessler RC, Hwang I, LaBrie R, Petukhova M, Sampson NA, Winters KC, Shaffer HJ. DSM-IV pathological gambling in the National Comorbidity Survey Replication. Psychol Med. 2008;38:1351–1360. [PMC free article] [PubMed]
  31. Ladouceur R, Walker M. A cognitive perspective on gambling. In: Salkovskis PM, editor. Trends in Cognitive and Behavioural Therapies. Wiley & Sons; Chichester, U.K.: 1996. pp. 89–120.
  32. Langer EJ. The illusion of control. J Pers Soc Psychol. 1975;32:311–328.
  33. Lawrence AJ, Luty J, Bogdan NA, Sahakian BJ, Clark L. Problem gamblers share deficits in impulsive decision-making with alcohol-dependent individuals. Addiction. 2009;104:1006–1015. [PMC free article] [PubMed]
  34. Lesieur HR, Blume SB. The South Oaks Gambling Screen (SOGS): a new instrument for the identification of pathological gamblers. Am J Psychiatry. 1987;144:1184–1188. [PubMed]
  35. Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH. An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. Neuroimage. 2003;19:1233–1239. [PubMed]
  36. Meyer G, Schwertfeger J, Exton MS, Janssen OE, Knapp W, Stadler MA, Schedlowski M, Kruger TH. Neuroendocrine response to casino gambling in problem gamblers. Psychoneuroendocrinology. 2004;29:1272–1280. [PubMed]
  37. Miller NV, Currie SR. A Canadian population level analysis of the roles of irrational gambling cognitions and risky gambling practices as correlates of gambling intensity and pathological gambling. J Gambl Stud. 2008;24:257–274. [PubMed]
  38. Montague PR, Hyman SE, Cohen JD. Computational roles for dopamine in behavioural control. Nature. 2004;431:760–767. [PubMed]
  39. Murray GK, Clark L, Corlett PR, Blackwell AD, Cools R, Jones PB, Robbins TW, Poustka L. Incentive motivation in first-episode psychosis: a behavioural study. BMC Psychiatry. 2008;8:34. [PMC free article] [PubMed]
  40. Nakamura K, Matsumoto M, Hikosaka O. Reward-dependent modulation of neuronal activity in the primate dorsal raphe nucleus. J Neurosci. 2008;28:5331–5343. [PMC free article] [PubMed]
  41. Potenza MN. Should addictive disorders include non-substance-related conditions? Addiction. 2006;101(Suppl 1):142–151. [PubMed]
  42. Potenza MN. The neurobiology of pathological gambling and drug addiction: an overview and new findings. Philos Trans R Soc Lond B Biol Sci. 2008;363:3181–3189. [PMC free article] [PubMed]
  43. Reuter J, Raedler T, Rose M, Hand I, Glascher J, Buchel C. Pathological gambling is linked to reduced activation of the mesolimbic reward system. Nat Neurosci. 2005;8:147–148. [PubMed]
  44. Schott BH, Minuzzi L, Krebs RM, Elmenhorst D, Lang M, Winz OH, Seidenbecher CI, Coenen HH, Heinze HJ, Zilles K, Duzel E, Bauer A. Mesolimbic functional magnetic resonance imaging activations during reward anticipation correlate with reward-related ventral striatal dopamine release. J Neurosci. 2008;28:14311–14319. [PubMed]
  45. Schultz W. Getting formal with dopamine and reward. Neuron. 2002;36:241–263. [PubMed]
  46. Shohamy D, Wagner AD. Integrating memories in the human brain: hippocampal-midbrain encoding of overlapping events. Neuron. 2008;60:378–389. [PMC free article] [PubMed]
  47. Steeves TD, Miyasaki J, Zurowski M, Lang AE, Pellecchia G, Van Eimeren T, Rusjan P, Houle S, Strafella AP. Increased striatal dopamine release in Parkinsonian patients with pathological gambling: a [11C] raclopride PET study. Brain. 2009;132:1376–1385. [PMC free article] [PubMed]
  48. Talairach J, Tournoux P. Co-planar stereotaxic atlas of the human brain. Thieme Medical Publishers; New York: 1988.
  49. Tanabe J, Thompson L, Claus E, Dalwani M, Hutchison K, Banich MT. Prefrontal cortex activity is reduced in gambling and nongambling substance users during decision-making. Hum Brain Mapp. 2007;28:1276–1286. [PubMed]
  50. Tobler PN, Fiorillo CD, Schultz W. Adaptive coding of reward value by dopamine neurons. Science. 2005;307:1642–1645. [PubMed]
  51. Townshend JM, Duka T. Patterns of alcohol drinking in a population of young social drinkers: a comparison of questionnaire and diary measures. Alcohol Alcohol. 2002;37:187–192. [PubMed]
  52. Worsley KJ, Marrett S, Neelin P, Vandal AC, Friston KJ, Evans AC. A unified statistical approach for determining significant signals in images of cerebral activation. Hum Brain Mapp. 1996;4:58–73. [PubMed]
  53. Wrase J, Schlagenhauf F, Kienast T, Wustenberg T, Bermpohl F, Kahnt T, Beck A, Strohle A, Juckel G, Knutson B, Heinz A. Dysfunction of reward processing correlates with alcohol craving in detoxified alcoholics. NeuroImage. 2007;35:787–794. [PubMed]
  54. Zack M, Poulos CX. Amphetamine primes motivation to gamble and gambling-related semantic networks in problem gamblers. Neuropsychopharmacol. 2004;29:195–207. [PubMed]