J Gambl Stud DOI 10.1007/s10899-017-9683-5 REVIEW PAPER
Pathological Gambling and Motor Impulsivity: A Systematic Review with Meta-Analysis Nahian S. Chowdhury1 • Evan J. Livesey1 • Alex Blaszczynski1 Justin A. Harris1
•
Ó Springer Science+Business Media New York 2017
Abstract Motor impulsivity, which is an impairment in withholding and cancelling inappropriate responses, may account for the inability for pathological gamblers (PGs) to inhibit their urges to gamble. The aim of this systematic review was to perform a quantitative and qualitative synthesis of existing studies in order to assess whether PGs without comorbid substance use disorder have elevated motor impulsivity, relative to healthy controls. An exhaustive literature search led to the identification of 20 studies which met inclusion criteria. A meta-analysis was then conducted on the following measures: stop signal reaction time from the stop signal task; commission errors, omission errors, and Go reaction time from the Go/No-Go task; and the motor impulsiveness subscale of the Barratt Impulsiveness Scale (BIS-Motor). The results revealed a moderate to large mean effect size of stop signal reaction time, small to moderate mean effect sizes for commission errors, omission errors and Go reaction time, and a large mean effect size for the BISMotor. Significant heterogeneity in effect sizes was observed on most behavioural measures, but not for the BIS-Motor or omission errors on the Go/No-Go task. Overall, these results suggest that motor impulsivity may be one of the features of PG psychopathology, accounting for their poor inhibitory control over gambling behaviours. Moreover, other deficits in sustained attention, or more generally in executive/cognitive control, may be present in PGs. We discuss the implications, limitations of existing research, and suggested avenues for future studies, particularly the need to acknowledge heterogeneity amongst PGs and amongst different behavioural measures. Keywords Gambling Impulsivity Inhibitory control Executive function
& Nahian S. Chowdhury
[email protected] 1
School of Psychology, University of Sydney, Camperdown, NSW 2006, Australia
123
J Gambl Stud
Introduction Although many individuals gamble without experiencing difficulties, it becomes an issue for some, whereby it gradually impacts an individual’s capacity to achieve their personal, relational and vocational goals, leading to a pattern of pathological gambling (PG). The symptoms of PG (or Gambling Disorder, as named in the DSM-V) are much like the symptoms of tolerance and withdrawal in substance dependence (e.g. gambling with increasing amounts of money to experience the desired arousal and restlessness/irritability during attempts to resist gambling), and are characterised by a loss of control over one’s gambling behaviour (Grant and Chamberlain 2014). Major models have attempted to explain why PGs find it so difficult to control their urge to gamble, even in the face of negative life consequences. These models attempt to integrate various interacting factors, where certain factors are more or less prominent in each individual gambler (Sharpe 2002), or where each individual gambler can be categorised in one of three pathways representing a certain combination of factors (Blaszczynski and Nower 1998). These include social factors such as gambling accessibility and familial/cultural attitudes towards gambling, biological factors such as genetic predispositions and altered neurotransmitter systems, and psychological factors such as associative learning, mood disturbances, gambling related arousal and cognitive biases. However, consensus suggests that the characteristic that can account for most cases of PG is impulsivity (Grant et al. 2016). Impulsivity is a complex construct that has been difficult to define. Most definitions (Caswell et al. 2013; Chamberlain and Sahakian 2007; Evenden 1999; Logan et al. 1997; Moeller et al. 2001; Stanford et al. 2009) suggest that a behaviour is impulsive if it is inappropriate in a particular context, is prematurely expressed with little consideration of the future consequences and leads to undesirable outcomes in the short-term and/or longterm. Impulsivity is also complex in that it is a multifaceted construct—there are different types of impulsivity which rely on separate neurological processes and which are measured in different ways. Researchers have distinguished two primary types of impulsivity: cognitive and motor impulsivity (Brunner and Hen 1997; Caswell et al. 2013; Chamberlain and Sahakian 2007; Kraplin et al. 2014; Kraplin et al. 2015; Malloy-Diniz et al. 2007; Olmstead et al. 2009; Vassileva et al. 2007). Cognitive impulsivity refers to impulsivity in decisions, whereby an individual is biased towards smaller immediate rewards as opposed to larger delayed rewards, or where an individual makes quick decisions under conditions of uncertainty. Tasks such as the delay discounting task (e.g. Alessi and Petry 2003) and the information sampling task (e.g. Lawrence et al. 2009) have been used to measure this facet of impulsivity. Motor impulsivity on the other hand, refers to impulsivity in action, characterised by an impaired ability to inhibit (delay, withhold and interrupt) inappropriate responses. The ability to inhibit inappropriate responses is an important aspect of human functioning, as it allows individuals to evaluate the consequences of these responses, and to express more appropriate responses. Therefore, it is possible that this aspect of functioning is impaired in PG, and can account for the unsuccessful attempts to inhibit the urge to gamble. Accordingly, the goal of this systematic review will be to evaluate whether this aspect of impulsivity plays a role in the aetiology of PG. Motor Impulsivity has been measured in two ways; through self-report and through laboratory tasks. The most widely used measures are described below.
123
J Gambl Stud
Go/No-Go Task The Go/No-Go (GNG) task (Donders 1969) consists of ‘‘go’’ trials, where the participant is required to make a speeded response (e.g. pressing a key when a red square appears), as well as ‘‘no-go’’ trials, where the participant is required to withhold responding (e.g. withholding a response when a black square appears). The key dependent measure reflective of motor impulsivity is the number of responses made to no-go stimuli (commission errors), as this represents an individual’s ability to inhibit an inappropriate response. Other dependent measures include the absence of a response to a go stimulus (omission errors), which is thought to be a measure of sustained attention (Trommer et al. 1988), as well as the reaction time on go trials (Go RT), where it is thought that impulsivity leads to faster (or rash) responding on these trials (Rentrop et al. 2008).
Stop Signal Task The stop-signal task (Logan and Cowan 1984) is similar to the GNG task; however, it requires cancellation of an already cued response. Along with go trials, there are also a certain number of ‘‘stop’’ trials, where the go stimulus is followed by a ‘‘stop signal’’ after a certain interval (the ‘‘Stop Signal Delay’’, SSD), signalling to participants that they must cancel their initiated response. Researchers usually attempt to identify the SSD at which there is a 50% probability of inhibiting the response (the critical SSD). A staircasing method is used, where the SSD will increase or decrease by a certain interval (e.g. 50 ms) on each trial based on the success or failure to inhibit a response on the previous trial, which should lead to a 50% success rate in inhibition. Hence, the mean SSD identified through this staircasing procedure becomes equivalent to the critical SSD. Based on a horse-race model where ‘‘go’’ and ‘‘stop’’ processes compete, the critical SSD is then subtracted from the average Go RT to identify the Stop-Signal Reaction Time (SSRT) of the participant. This is the average amount of time required for a participant to successfully cancel their response after the onset of the stop signal. Hence, longer SSRTs are reflective of impaired inhibitory control over behaviour. Refer to Fig. 1 for a schematic representation of the stop-signal task.
Fig. 1 A schematic representation of the stop-signal task, where the go stimulus is a left arrow and the stop signal is an auditory beep. The critical SSD (the SSD at which there is a 50% chance of successful inhibition) is identified through a staircasing method, and is then subtracted from the Mean Go RT to identify the SSRT. According to a horse race model, the go process will be executed if the SSD on any trial is higher than the critical SSD, whilst the stop process will be executed if the SSD is lower than the critical SSD. Individuals who are high in motor impulsivity have longer SSRTs, such that they require longer to execute their stop response
123
J Gambl Stud
Barratt-Impulsiveness Scale The Barratt Impulsiveness Scale (BIS-11) (Patton et al. 1995) is a self-report questionnaire which contains three subscales: non-planning, attentional, and motor impulsiveness subscales. The Non-planning subscale measures an individual’s orientation towards the present and relates to delay discounting tasks. The Attentional subscale measures an individual’s ability to stay focused on daily activities. The Motor subscale measures an individual’s tendency to respond without prior thought, containing items such as ‘‘I do things without thinking’’ and ‘‘I act on the spur of the moment’’. Researchers have complemented the Stop-Signal Task and GNG task using the motor-impulsiveness subscale the BIS-11, as this subscale has been thought to reflect a similar concept of inhibitory control (Kraplin et al. 2015; Leppink et al. 2016; Meule 2013). However, weak or inconsistent correlations between self-report and laboratory-task measures of impulsivity have been found (e.g. Enticott et al. 2006). One explanation for this is that self-report questionnaires are reflective of trait impulsivity (a stable personality feature) whilst laboratory tasks are reflective of state impulsivity (a transitory or momentary feature specific to a situation) (Meule 2013), suggesting that even the attribute of motor impulsivity itself requires a multi-dimensional conceptualisation. In the following systematic review, we will perform a qualitative and quantitative synthesis of empirical studies that compare pathological gamblers with healthy controls on measures obtained from the stop-signal task (SSRT), the GNG task (commission errors, omission errors and Go RT) and the BIS-11 (Motor Impulsiveness Subscale). This will allow us to address the question of whether motor impulsivity represents a major component of PG psychopathology and what areas of further research are required. To our knowledge, two meta-analyses have previously been conducted (Lipszyc and Schachar 2010; Smith et al. 2014) comparing PGs (and other addictions) with healthy controls on Stop-Signal and/or GNG task performance, however both analyses had low power and did not account for the possible confounds of comorbid substance use disorder, restricting conclusions made about the isolated effects of PG. Furthermore, these meta-analyses only assessed laboratory-task measures of impulsivity, with no analysis of self-report measures. We aim to build on these analyses by including studies with PG samples without comorbid substance use disorder, and to investigate both laboratory-task and self-reported measures of impulsivity to allow for more multi-faceted conclusions.
Methods Literature Search An exhaustive search was conducted by the primary author in the PsycINFO, PubMed, Medline, Embase and CINAHLelectronic databases with no date restrictions, on the 22nd of November 2016. The search terms used were ‘‘gambling’’, ‘‘pathological gambling’’ and ‘‘problem gambling’’, in combination with ‘‘impulsivity’’, ‘‘impulsive behaviour’’, ‘‘executive function’’, ‘‘inhibition’’, ‘‘response inhibition’’, ‘‘cognitive control’’ and ‘‘inhibitory control’’. Once all duplicates were removed, the primary author partially screened the titles, abstracts and texts of the received articles, identifying studies which compare gamblers and a control group on a behavioural and/or self-reported measure of motor impulsivity. Once identified studies were partially screened, two of the authors
123
J Gambl Stud
independently reviewed the remaining papers for full-text screening. Initially, an agreement of 86% was reached on the exclusion and inclusion of papers. The authors then made contact to discuss any inconsistencies, reaching a final agreement of 100%.
Exclusion/Inclusion Criteria Studies were required to fulfil nine criteria. They had to (a) be presented in English (although we acknowledge that this may be a source of bias in our analysis) b) be conducted on human participants (c) compare a control group to a gambling group (d) contain a gambling group who met criteria for DSM-IV pathological gambling or DSM-V gambling disorder as assessed by diagnostic instruments such as the South Oaks Gambling screen (SOGs), (e) contain a group of gamblers who did not meet criteria for comorbid substance use disorder (aside from caffeine or nicotine dependence) (f) contain a control group which differs significantly in gambling severity (g) report a minimum of one of the following: SSRT in the Stop-Signal task, No-Go Commission errors, Go Omission Errors, Go RT in the GNG task, and BIS-Motor Impulsiveness, (h) calculate SSRT if a stop-signal task is used, rather than calculating the percentage of inhibited responses at various stop signal delays, and (i) provide sufficient information for effect size calculation. Although there were no included studies with adolescent samples, no age limit was applied to the exclusion/inclusion criteria. We included studies that had No-Go and Go trials of all relative frequencies. Samples where there were no PGs which satisfied criteria for substance use disorder, or where both groups did not differ in substance use, were included. Samples with comorbid ADHD were not excluded because there are a limited number of studies excluding ADHD psychopathology or making comparisons between comorbid and non-comorbid PGs, which would result in low statistical power for the meta-analysis.
Calculating Effect Sizes We used http://www.campbellcollaboration.org/resources/effect_size_input.php to calculate Cohen’s d and inverse variance weights (Lipsey and Wilson 2001). We used means and SDs, or F and t values to calculate effect sizes. Where multiple gambling groups existed (e.g. PG alone vs. PG with comorbid ADHD) or where repeated measures were taken (e.g. across different blocks of the GNG task), the means and SDs of those groups or repeated measures were combined. For the stop-signal task, we calculated the effect sizes for SSRT. Other measures from this task (e.g. Go RT, error rates) were not analysed as they are rarely reported. For the GNG task, we calculated the effect size for errors of commission, errors of omission, and Go RT. For the BIS, we calculated the effect size for the score on BIS-Motor Impulsiveness. Positive effect sizes were indicative of lower SSRT, fewer commission errors, fewer omission errors, quicker Go RTs and lower BISMotor Impulsiveness in the control group.
Performing the Meta-Analyses The meta-analyses were conducted using the Comprehensive Meta-Analysis Software (Version 3; Biostat, Englewood, NJ, 2014). We used a random effects model to calculate the weighted mean effect size of each measure (Borenstein et al. 2007), although we also indicate when the results differ if a fixed effects model is used. Each effect size was assigned a weight, and the weighted k effect sizes were summed. This total was divided by
123
J Gambl Stud
the sum of the weights of the k studies to obtain the weighted mean effect size. We also computed the standard error and variance of the weighted mean effect size as well as the 95% confidence interval of the effect size. We then carried out a z-transformation to test the null hypotheses that the weighted mean effect size was 0, with two-tailed significance set at a p value of 0.05. Under a fixed effects model, we also computed the Q-statistic to test whether there was significant heterogeneity amongst the effect sizes Significance was assessed against a k-1 degrees of freedom Chi squared distribution. A Q-statistic with a p \ 0.05 suggests that the dispersion of effect sizes between the k studies is substantial. We also computed the I2 statistic, an estimate of the percentage of variability due to effect size heterogeneity (Borenstein et al. 2009).
Results As shown in Fig. 2, partial text screening identified 46 studies comparing a gambling group with a control group on a measure of motor impulsivity. After full text screening, 12 studies were not included as comorbid substance use disorder was not an exclusion criterion, 6 studies using the BIS-11 were excluded as data on the motor impulsiveness subscale were not available, 4 studies were excluded as percentage inhibited response was used in the stop signal task analysis rather than SSRT (although 1 of these studies was still eligible for a comparison of BIS Motor Impulsiveness), 3 studies were excluded as the gambling group did not meet criteria for PG, 1 study was excluded as the control groups’ gambling habits were not significantly different from the PG group, and 1 study was excluded as there were insufficient data to calculate the desired effect size. This led to the exclusion of 26 studies, leaving 20 studies for the meta-analysis. Of the studies which met inclusion criteria, 5 studies reported SSRT differences (with a total of 179 PGs and 188 Controls) and 7 studies reported GNG Commission Differences (with a total 328 PGs and 350 controls). Although 4 studies reported GNG Omission differences, the effect size of 5.89 obtained from Zhou et al. (2016) was considered an outlier, and likely to reflect an error in their reporting, and was therefore excluded, leaving 3 studies for the final analysis (with a total of 189 PGs and 181 Controls). Five studies reported GNG Go RT differences (with a total of 225 PGs and 261 Controls) and 12 studies included analysis of the BIS-11 Motor Impulsiveness Subscale (with a total of 410 PGs and 464 Controls). Refer to Tables 1, 2 and 3 for the characteristics of the included studies. Of the 5 studies reporting SSRT differences, 4 involved choice reaction time tasks (e.g. responding with left or right arrows based on the direction of stimuli), with an auditory (Grant et al. 2010; Goudriaan et al. 2005, 2006) or visual (Kraplin et al. 2015; Rodriguez-Jimenez et al. 2006) stop signal presented at a staircased interval after the appearance of the go-signal on 25% of the trials. One study (Billieux et al. 2012) involved a matching task where participants were instructed to respond when a target stimulus matched a cue stimulus; on 25% of the trials, the target stimuli would turn red at a certain interval after its onset, to which participants would have to withhold their response. Four out of 5 studies reported significantly slower SSRTs in PGs compared to control. A meta-analysis of SSRT differences between PGs and Controls revealed an overall moderate-large effect size of 0.57, in adherence with Cohen’s (1992) effect size classifications of small (0.2), medium (0.5) and large (0.8). Among the 7 studies reporting on the GNG task, 2 studies (Goudriaan et al. 2005; Zhou et al. 2016) involved responding and withholding responses to a one of numerous go and no-go stimuli (e.g. withhold a response to one of four two-digit numbers), 4 studies
123
J Gambl Stud
Fig. 2 Flow chart representing the process in which the papers included in the meta-analysis were identified
(Kertzman et al. 2008, 2011, Fuentes et al. 2006, van Holst et al. 2012) involved responding and withholding responses to only one go and one no-go stimulus (e.g. a certain coloured square) and another study (Rodriguez-Jimenez et al. 2006) involved responding
123
123 33.2
55 (55)
40 (40)
Rodriguez-Jimenez et al. (2006)
37.3
32.0
35.6
Absence of pathological gambling, SOGS \ 4, no substance use in the previous year (except caffeine and nicotine), no ADHD, no psychotic or affective disorders, no organic mental disorders, no somatic disorders which would interfere with testing, no illiteracy, no IQ \ 70
No history of substance use or dependence, no major psychiatric disorders, no physical conditions which could affect cognitive and motor performance, Dutch as first language, no current use of psychotropic medication which could not be discontinued, negative urinalysis for alcohol, cannabis and benzodiazepines, age between 18 and 60
49 (40)
50 (35)
Mean age
N (male)
Criteria
N (male)
Mean age
Gambling group
Control group
Goudriaan et al. (2006)
Study
Table 1 Characteristics of study samples and effect sizes, for papers which use the stop signal task
DSM-IV pathological gambling (SOGS), no substance use in the previous year (except caffeine and nicotine), no psychotic or affective disorders, no organic mental disorders, no somatic disorders which would interfere with testing, no illiteracy, no IQ \ 70. Gamblers were seeking treatment
DSM-IV pathological gambling (DIS)a, no current treatment except for PG, no history of substance use/ dependence (except nicotine), no major psychiatric disorders except for manic disorders, OCD, ADHD and anti-social personality disorder, no physical conditions which could affect cognitive and motor performance, Dutch as first language, no current use of psychotropic medication which could not be discontinued, negative urinalysis for alcohol, cannabis and benzodiazepines, age between 18 and 60. Gamblers were seeking treatment
Criteria
-0.15 Combined the means and SDs of the ADHD and non-ADHD group
0.64
Effect size (Cohen’s d), notes
J Gambl Stud
20 (17)
52 (21)
Billieux et al. (2012)
Kraplin et al. (2015)
25.69
38.30
50.4
Irregular and unproblematic gambling, smoked less than 20 cigarettes in life, no current treatment for mental or personality disorders, no other mental disorders in the previous 12 months, no physical conditions which could affect cognitive and motor performance (e.g. ADHD), German as first language, no use of psychotropic medication in the previous two weeks, negative urinalysis for cocaine, ecstasy, methamphetamines, opioids or cannabis, no other mental disorders, age between 18 and 25
No past or present gambling activity, no history of psychiatric or neurological disorder
Absence of a history of psychiatric illness
50.4
44.1
29.89
29 (11)
20 (17)
26 (20)
DSM-IV pathological gambling in previous 12 months, no current treatment for mental or personality disorders, no physical conditions which could affect cognitive and motor performance (e.g. ADHD), German as first language, no use of psychotropic medication in the previous two weeks, negative urinalysis for cocaine, ecstasy, methamphetamines, opioids or cannabis, no other mental disorders, age between 18 and 25. Gamblers were recruited from the community via advertisements
DSM-IV pathological gambling (SOGS), no substance use disorder, no reported history of neurological disorders. Gamblers were seeking treatment
DSM-IV pathological gambling (SCIPG), no current substance abuse or dependence (except nicotine), no lifetime bipolar disorder, no dementia, no psychotic disorder. Gamblers were recruited from the community via advertisements
Criteria
0.64 None of the participants in the PG group met criteria for DSM-IV substance use disorder in the past 12 months Combined the means and SDs of the PG and PG/ ND group
1.10
0.81 Baseline data were used for gamblers and control
Effect size (Cohen’s d), notes
Diagnostic instrument used to assess pathological gambling, such as the South Oaks Gambling Screen (SOGS), Diagnostic Interview Schedule (DIS), Structured Clinical Interview for DSM (SCID) and the NORC Diagnostic Screen for Gambling Problems (NODS)
a
26 (10)
Mean age
N (male)
Criteria
N (male)
Mean age
Gambling group
Control group
Grant et al. (2010)
Study
Table 1 continued
J Gambl Stud
123
123 39.0
40.1
48 (40)
52
35.8
40.9
49 (34)
82 (45)
Goudriaan et al. (2005)
Fuentes et al. (2006)
Did not meet criteria for any psychiatric disorders (e.g. substance abuse) or lifetime diagnosis of a recurrent psychiatric syndrome
No lifetime substance abuse or dependence, no history of psychosis, no treatment for a mental disorder in past 3 years, physical conditions which could affect cognitive and motor performance, use of psychotropic medication which could not be discontinued, negative urinalysis for alcohol, cannabis and benzodiazepines
Mean age
N (male)
Mean age
N (male)
Criteria
Gambling group
Control group
Study
DSM-IV pathological gambling (SOGS). noncomorbid Pathological Gambling Group did not meet criteria for any psychiatric disorders (e.g. substance abuse). Gamblers were treatment seeking
DSM-IV pathological gambling (DIS), no lifetime substance abuse or dependence, no history of psychosis, no current treatment for mental disorders except for those under study, physical conditions which could affect cognitive and motor performance, use of psychotropic medication which could not be discontinued, negative urinalysis for alcohol, cannabis and benzodiazepines. Gamblers were treatment seeking
Criteria
Table 2 Characteristics of study samples and effect sizes, for papers which use the Go/No-Go task
–
0.80 Equiprobable Go and NoGo Trials
-0.20
–
Go RT
Combined the means and SDs of the visual and auditory version of the GNG task
Use the means and SDs of the Non-Comorbid Pathological Gamblers
–
Omission
0.51
Commission
Effect size (Cohen’s d), notes
Equiprobable Go and NoGo Trials
Go/No-Go proportions
J Gambl Stud
33.2
39.5
55 (55)
83 (56)
32.0
36.8
40 (40)
84 (56)
RodriguezJimenez et al. (2006)
Kertzman et al. (2008)
No current psychiatric disorder, no lifetime diagnosis of schizophrenia, no bipolar disorder, no ADHD, no OCD, no substance-related disorders, no neurological disorders
Absence of pathological gambling, SOGS \ 4, no substance use in the previous year (except caffeine and nicotine), no ADHD, no psychotic or affective disorders, no organic mental disorders, no somatic disorders which would interfere with testing, no illiteracy, no IQ \ 70
Mean age
N (male)
Mean age
N (male)
Criteria
Gambling group
Control group
Study
Table 2 continued
DSM-IV pathological gambling (Semi-Structured Interview, SOGS), no ongoing psychiatric treatment, no alcohol and substance use/dependence, no major psychiatric disorders, no neurological disorder, no mental retardation, no use of psychotropic medication in the previous month. Gamblers were seeking treatment
DSM-IV pathological gambling (SOGS), no substance use in the previous year (except caffeine and nicotine), no psychotic or affective disorders, no organic mental disorders, no somatic disorders which would interfere with testing, no illiteracy, no IQ \ 70 Gamblers were seeking treatment
Criteria
Both Rare NoGo Blocks and Rare Go Blocks
Equiprobable Go and NoGo Trials
Go/No-Go proportions
0.24
Omission
–
Go RT
0.34
0.70 Combined the means and SDs of all four blocks of the GNG task
0.21
Combined the means and SDs of the ADHD and non-ADHD group
0.08
Commission
Effect size (Cohen’s d), notes
J Gambl Stud
123
123
Kertzman et al. (2011)
Study
39.51
51 (35)
37.70
57 (36)
No current psychiatric disorder, no lifetime DSMIV Axis 1 disorder such as schizophrenia, bipolar disorder, no ADHD, no OCD, no substance-related disorder
Mean age
N (male)
Mean age
N (male)
Criteria
Gambling group
Control group
Table 2 continued
DSM-IV pathological gambling (Semi-Structured Interview), o alcohol and substance use/dependence, no major psychiatric disorders, no neurological disorder, no mental retardation, no use of psychotropic medication in the previous month. Gamblers were seeking treatment
Criteria
Both Rare NoGo Blocks and Rare Go Blocks
Go/No-Go proportions
0.34
-0.07
0.57
Go RT
Combined the means and SDs of all four blocks of the GNG task
Omission
Commission
Effect size (Cohen’s d), notes
J Gambl Stud
van Holst et al. (2012)
Study
34.38
16 (16)
36.20
15 (15)
Gamble less than 3 times a year, no treatment for any mental disorder other than PG, no alcohol or substance use disorder, no lifetime diagnosis of schizophrenia/ psychotic episodes, no manic disorder, no OCD, no PTSD, no history or current treatment for neurological disorders, no internal disorders, no brain trauma, no exposure to neurotoxic factors, no use of psychotropic medication, no difficulty reading Dutch, negative urinalysis for alcohol, amphetamines, benzodiazepines, opioids and cocaine
Mean age
N (male)
Mean age
N (male)
Criteria
Gambling group
Control group
Table 2 continued
DSM-IV pathological gambling (DIS, SOGS), no treatment for any mental disorder other than PG, no alcohol or substance use disorder, no lifetime diagnosis of schizophrenia/ psychotic episodes, no manic disorder, no OCD, no PTSD, no history or current treatment for neurological disorders, no internal disorders, no brain trauma, no exposure to neurotoxic factors, no use of psychotropic medication, no difficulty reading Dutch, negative urinalysis for alcohol, amphetamines, benzodiazepines, opioids and cocaine. Gamblers were seeking treatment
Criteria
Rare No-Go Trials
Go/No-Go proportions
–
Omission
0.72
Go RT
One participant did not meet criteria for DSM-IV criteria according to the DIS, but all had scores higher than 5 on the SOGS
-0.11
Commission
Effect size (Cohen’s d), notes
J Gambl Stud
123
123
Zhou et al. (2016)
Study
29
28
No diagnosis of PG or IAD, no alcohol dependence, no diagnosis of substance dependence, no neurological disorder, no history of head injury, no systemic disease that affects the CNS
23 (18)
23 (16)
Mean age
N (male)
Criteria
N (male)
Mean age
Gambling group
Control group
Table 2 continued
DSM-IV pathological gambling, no alcoholdependence, no diagnosis of substance of alcohol or substance dependence, were non-smokers, no neurological disorder, no history of head injury. Gamblers were seeking treatment
Criteria
Equiprobable Go and NoGo Trials
Go/No-Go proportions
0.80
Commission
5.89
Omission
0.02
Go RT
Effect size (Cohen’s d), notes
J Gambl Stud
48.4
36.42
30 (14)
37 (37)
NODS \=1, no substance dependence within past 12 months (excluding caffeine and nicotine), no acute and severe psychiatric disorders (i.e. acute suicidality, uncontrolled psychosis), no non-English speakers, no individuals who could not provide informed consent, negative toxicology screen for marijuana, alcohol, cocaine, opioids
45.7
41 (17)
40 (40)
Ledgerwood et al. (2009)
Lai et al. (2011)
No neurological, psychiatric or psychological disorders, no score higher than 29 on the Beck-Depression Inventory
33.2
55 (55)
Absence of pathological gambling, SOGS \ 4, no substance use in the previous year (except caffeine and nicotine), no ADHD, no psychotic or affective disorders, no organic mental disorders, no somatic disorders which would interfere with testing, no illiteracy, no IQ \ 70
32.0
40 (40)
RodriguezJimenez et al. (2006)
35.61
Mean age
N (male)
Mean age
N (male)
Criteria
Gambling group
Control group
Study
DSM-IV pathological gambling (SCID, SOGS), no neurological, psychiatric or psychological disorders other than PG, no score higher than 29 on the BeckDepression Inventory. Gamblers were seeking treatment
DSM-IV pathological gambling (NODS), had gambled in past 2 months, no substance dependence within past 12 months (excluding caffeine and nicotine), no acute and severe psychiatric disorders (i.e. acute suicidality, uncontrolled psychosis), no non-English speakers, no individuals who could not provide informed consent, negative toxicology screen for marijuana, alcohol, cocaine, opioids. Half of the gamblers were seeking treatment, and half were recruited from the community via advertisements
DSM-IV pathological gambling (SOGS), no substance use in the previous year (except caffeine and nicotine), no psychotic or affective disorders, no organic mental disorders, no somatic disorders which would interfere with testing, no illiteracy, no IQ \ 70. Gamblers were seeking treatment
Criteria
0.74 No significant differences between pathological gamblers and control groups in alcohol use and other substance use
0.54
1.52 Combined the means and SDs of the ADHD and non-ADHD group
Effect size (Cohen’s d), notes
Table 3 Characteristics of study samples and effect sizes, for papers which report on the motor-impulsiveness subscale of the Barratt Impulsivity Scale
J Gambl Stud
123
123 28.1
45.3
33.50
15 (15)
54 (19)
22 (20)
47.5
28.06
54 (27)
31 (27)
Black et al. (2013)
De Wilde et al. (2013)
24.9
33 (33)
No Signs of pathological gambling, No signs of lifetime substance use disorders (except caffeine or nicotine abuse or dependence), no psychotic disorders, no organic deterioration or amnesic disorders, no physical handicaps, no severe somatic disorders
Score of 0 on the NODS and SOGS, No substance use disorder within the last 3 months (except tobacco), no current or past diagnosis of schizophrenia, no bipolar I and II disorder, no primary neurological disorder, no major depressive episode in last three months, no significant evidence of cognitive impairment (MMSE \ 23), no history of head injury (with loss of consciousness [ 10 min)
No Axis 1 diagnosis, No history of substance use, no psychotic disorders, no neurological disorder, no significant head injury and no mental retardation
Mean age
N (male)
Criteria
N (male)
Mean age
Gambling group
Control group
Hwang et al. (2012)
Study
Table 3 continued
DSM-IV Pathological Gambling (SCID, SOGS), No signs of lifetime substance use disorders (except caffeine or nicotine abuse or dependence), no psychotic disorders, no organic deterioration or amnesic disorders, no physical handicaps, no severe somatic disorders. Gamblers were seeking treatment
DSM-IV pathological gambling (NODS, SOGS), No substance use disorder within the last 3 months (except tobacco), no current or past diagnosis of schizophrenia, no bipolar I and II disorder, no primary neurological disorder, no major depressive episode in last three months, no significant evidence of cognitive impairment (MMSE \ 23), no history of head injury (with loss of consciousness [ 10 min). Gamblers were recruited from the community via advertisements, a gambling registry and word-of-mouth
DSM-IV pathological gambling (SOGS). No history of substance use, no psychotic disorders, no neurological disorder, no significant head injury and no mental retardation. Gamblers were seeking treatment
Criteria
1.59 Combined the means and SDs of abstinent and nonabstinent pathological gamblers
0.89
1.25
Effect size (Cohen’s d), notes
J Gambl Stud
53 (32)
Kraplin et al. (2014)
36.74
37.48
23 (20)
No history of substance use or dependence, no major psychiatric disorders, no physical conditions which could affect cognitive and motor performance, Dutch as first language, no current use of psychotropic medication which could not be discontinued, negative urinalysis for alcohol, cannabis and benzodiazepines, age between 18 and 60
No substance use disorders, no physical, psychiatric or neurological problems
35.61
37.82
51 (44)
36.45
23 (20)
20 (17)
Grecucci et al. (2014)
No history of psychiatric or neurological disorders
37.15
20 (17)
Giorgetta et al. (2014)
Mean age
N (male)
Mean age
N (male)
Criteria
Gambling group
Control group
Study
Table 3 continued
DSM-IV pathological gambling (DIS), no current treatment except for PG, no history of substance use/dependence (except nicotine), no major psychiatric disorders except for manic disorders, OCD, ADHD and anti-social personality disorder, no physical conditions which could affect cognitive and motor performance, Dutch as first language, no current use of psychotropic medication which could not be discontinued, negative urinalysis for alcohol, cannabis and benzodiazepines, age between 18 and 60. Gamblers were seeking treatment
DSM-IV Pathological Gambling, no substance use disorders, no physical, psychiatric or neurological problems, no other comorbid disorders (such as anxiety disorders), no cognitive impairment. Gamblers were seeking treatment
DSM-IV Pathological gambling, no alcohol or substance use disorder, no current major depressive episodes, no neurological or medical illnesses, no history of traumatic brain injury. Gamblers were seeking treatment
Criteria
0.95
1.23
0.98
Effect size (Cohen’s d), notes
J Gambl Stud
123
123 29.89
41.56
29
26 (20)
54 (54)
23 (18)
41.56
54 (54)
23 (16)
Ciccarelli et al. (2016)
Zhou et al. (2016)
28
25.69
52 (21)
No diagnosis of PG or IAD, no alcohol dependence, no diagnosis of substance dependence, no neurological disorder, no history of head injury, no systemic disease that affects the CNS
No pre-existing psychiatric or neurological disorder, no pre-existing problem gambling
Irregular and unproblematic gambling, smoked less than 20 cigarettes in life, no current treatment for mental or personality disorders, no other mental disorders in the previous 12 months, no physical conditions which could affect cognitive and motor performance (e.g. ADHD), German as first language, no use of psychotropic medication in the previous two weeks, negative urinalysis for cocaine, ecstasy, methamphetamines, opioids or cannabis, no other mental disorders, age between 18 and 25
Mean age
N (male)
Criteria
N (male)
Mean age
Gambling group
Control group
Kraplin et al. (2015)
Study
Table 3 continued
DSM-IV pathological gambling, no alcohol-dependence, no diagnosis of substance of alcohol or substance dependence, were non-smokers, no neurological disorder, no history of head injury. Gamblers were seeking treatment
DSM-V pathological gambling, no alcohol or substance use, no neurological or medical illness, no pharmacological treatment, no comorbid Axis-I or Axis-II disorders. Gamblers were seeking treatment
DSM-IV pathological gambling in previous 12 months, no current treatment for mental or personality disorders, no physical conditions which could affect cognitive and motor performance (e.g. ADHD), German as first language, no use of psychotropic medication in the previous two weeks, negative urinalysis for cocaine, ecstasy, methamphetamines, opioids or cannabis, no other mental disorders, age between 18 and 25. Gamblers were recruited from the community via advertisements
Criteria
1.05
0.99
0.45 None of participants in the PG group met criteria for DSM-IV substance use disorder in the past 12 months Combined the means and SDs of the PG and PG/ ND group
Effect size (Cohen’s d), notes
J Gambl Stud
J Gambl Stud
to a letter only if it was preceded by a specific letter. The relative proportions of go and nogo stimuli in each of these studies are reported in Table 2. Four out of 7 studies reporting commission errors found a significantly higher commission error rate in PGs compared to controls, with a meta-analysis revealing an overall small-moderate effect size of 0.32. Upon exclusion of the effect size obtained from Zhou et al. (2016), two out of 3 studies reporting omission errors found a significantly higher omission error rate in PGs compared to controls, with a meta-analysis revealing an overall small-moderate effect size of 0.31. Three out of 5 studies reporting Go RT found significantly slower Go RT in PGs compared to control, with a meta-analysis revealing an overall non-significant small-moderate effect size of 0.35, although this effect size was significant when a fixed effects model was used. Of the 12 studies included in the analysis that used the BIS-Motor Impulsiveness subscale, all found significantly higher self-reported motor impulsiveness in PGs compared to controls. A meta-analysis revealed overall a large effect size of 0.96. Heterogeneity Analyses computing the Q statistic revealed significant heterogeneity in effect sizes for SSRT, GNG Commission errors and GNG Go RT, with no significant differences in effect sizes for BIS-Motor Impulsiveness and GNG Omission errors. Analysis of the I2 statistic showed that there was moderate-high heterogeneity for SSRT and GNG commission errors, high heterogeneity for GNG Go RT and low-moderate heterogeneity for BIS-Motor Impulsiveness, in adherence with the classification system (Higgins et al. 2003) of low (25%), moderate (50%) and high (75%). Refer to Table 4 for a summary of the meta-analysis, and to Fig. 3 for the forest plots of the effect sizes.
Discussion On the whole, the results of our meta-analysis suggest that PGs without comorbid substance use disorder have elevated motor impulsivity. Compared to controls, PGs rate themselves higher on self-reported motor impulsiveness (with large effect sizes), require more time to stop an already initiated response (with moderate to large effect sizes) and are more likely to fail to withhold a response to a no-go stimulus (with small to moderate effect sizes). Interestingly, PGs are also more likely to fail to execute a response to a go stimulus (with small to moderate effect sizes), pointing towards deficits in sustained attention, and in some cases, are slower at responding to a go stimulus (with small to moderate effect sizes). Our findings for the SSRT measure are consistent with Smith et al. (2014), where a similar overall moderate effect size for SSRT was also found for PGs. In contrast, Lipszyc and Schachar (2010) did not find a significant overall effect size for SSRT, although their analysis was based on only two studies. Our findings for the GNG measures are inconsistent with Smith et al. (2014), as their analysis revealed non-significant effect sizes for commission and omission errors for PGs, although their analysis was limited to two studies, whereas our analysis had higher statistical power. Nevertheless, there appears to be more evidence of behavioural motor impulsivity in PGs in the stop signal task compared to the GNG task.
Heterogeneity of Effect Sizes Our analysis revealed significant heterogeneity in effect sizes for most behavioural measures of motor impulsivity. The significant heterogeneity in the SSRT measure was greatly driven by the negative effect size obtained from Rodriguez-Jimenez et al. (2006), where a PG group without ADHD showed a lower (albeit non-significant) SSRT than the control
123
123
0.31
0.35
3
5
12
GNG Omission errors
GNG Go RT
Barratt Motor Impulsiveness Subscale
0.98
0.32
7
0.57
5
GNG Commission errors
Values in bold reflect significant p values
Variance
SE
CI lower
CI upper
z
p
0.45–1.59
-0.20 to 0.72
0.24–0.34
-0.11 to 0.80
-0.15 to 1.10
0.01
0.04
0.01
0.02
0.05
0.10
0.21
0.11
0.14
0.22
0.78
-0.06
0.11
0.05
0.15
1.18
0.76
0.52
0.59
1.00
9.78
1.68
2.97
2.30
2.66
0.008
<0.001
0.092
0.003
0.022
42.89
77.61
0.00
65.60
73.20
19.26
17.87
0.16
17.44
14.92
Q
11
4
2
6
4
df
I2
Mix–max
k
d
Heterogeneity analysis
Effect size analysis
Stop Signal Task - SSRT
Measure
Table 4 Results of the effect size and heterogeneity analysis
0.056
0.001
0.92
0.008
0.005
p
J Gambl Stud
J Gambl Stud
Fig. 3 Forest plots of the effect sizes for SSRT, GNG Commission errors, GNG Omission Errors, GNG Go RT and BIS Motor Impulsiveness
group. This emphasizes the need for more studies looking at the effects of comorbid ADHD on motor impulsivity. Another potential factor that may contribute to heterogeneity of the SSRT measure is variation in the Go RTs in the Stop Signal task. If participants progressively slow their response to the go signal as a strategy to avoid stopping errors, this can increase the probability of stopping above the targeted 50% success rate, leading to an underestimation of the SSRT. Hence, it would be beneficial for future researchers to report the probability of stopping along with mean Go RT, in order to determine whether any SSRT measures are affected by strategic slowing. There are a few possible reasons for the heterogeneity in GNG measures. One is to do with variations in the nature of the GNG tasks (types of stimuli used/number of stimuli requiring a response) and variations in the proportions of Go and No-Go trials. In addition, there may be differences between studies in the characteristics of the samples studied. For example, some comorbid psychopathologies may be over-represented in some samples relative to others. Nonetheless, it may simply be the case that heterogeneity in a characteristic such as motor impulsivity will always exist, with no one factor accounting for all cases of PG. It rests on future research to understand the task conditions and sample characteristics which relate to deficits in inhibitory control.
Implications Motor Impulsivity On the whole, our findings suggest that PGs are impaired on behavioural and self-reported measures of motor impulsivity (SSRT, commission error rate and BIS-motor
123
J Gambl Stud
impulsiveness). Thus, one reason for the failed attempts to control gambling in PG may be a general deficit in inhibitory control. Gamblers receive many internal (cognitions, emotions) and external signals to pursue gambling (e.g. advertisements for betting on a sports team) and to cease gambling (e.g. a friend advising them to stop gambling) in their daily lives. Our results suggest that gamblers who are highly motor impulsive may be at greater risk of PG, as they are more likely to gamble even when cued to cease gambling, and are less likely (or take much longer) to interrupt gambling once they have already begun. Of course, it should be acknowledged that behavioural measures collected in the laboratory are limited in their applicability towards such real-world situations as they tap into processes in the millisecond range and contain generic cues, whereas the failure to inhibit gambling would occur over a larger time frame and involve a more complex set of cues. Nonetheless, we believe that the results of these tasks are important as they attempt to isolate a specific cognitive mechanism (inhibitory control). Therefore, assessing whether this mechanism is impaired in PGs can inform us about aetiology at a basic level. It is likely that this is one mechanism, amongst many others (e.g. associative learning, cognitive biases) that may contribute to the failure to inhibit gambling. It also worth noting that our focus on motor impulsivity does not address differences in the compulsion to gamble, and thus understanding the relationship between compulsivity and impulsivity may be an important consideration. Further work is required on the applicability of behavioural measures to PG. For instance, stronger conclusions may be made through the use of GNG tasks and Stop-Signal tasks which contain gambling relevant stimuli. Furthermore, future studies should investigate the reliability of GNG measures and Stop-Signal measures throughout time or throughout treatment, as this may provide information as to whether such measures are indeed state measures of impulsivity or whether they may reflect stable and generalizable trait measures. For the BIS-motor impulsiveness measure, further caution should be exercised in interpreting this as a true measure of motor impulsivity reflecting inhibitory control, particularly because none of the items on this subscale explicitly ask questions related to difficulties ‘‘stopping’’ or ‘‘withholding’’ inappropriate actions. The questionnaire also has items which may be considered irrelevant to motor impulsivity, such as one’s tendency to change jobs or residence. Another concern with this measure is the inherent circularity of using questionnaires on gamblers, such that the leading nature of the questions coupled with recent involvement in gambling activities is likely to overinflate self-reported motor impulsiveness, which may be contributing to the large effect sizes. It would be more informative if such individuals judged themselves as impulsive on the basis of behaviours outside of gambling contexts, which could perhaps be communicated upon administration of the questionnaire.
Cause or Consequence? There is ongoing debate as to whether it is impulsivity which predisposes an individual to gambling or whether it is gambling that leads to the development of impulsivity. Recent evidence, suggests that PG in adulthood is predicted by impulsivity in childhood and adolescence (Auger et al. 2010; Dussault et al. 2011; Pagani et al. 2009). In relation to our analyses, this suggests that it is the core inhibitory control deficits that acts as a vulnerability factor for predisposing early gamblers towards a pattern of pathological gambling, as opposed to the deficits themselves being a product of increasing gambling frequency and intensity. Nevertheless, it is likely that the relationship between PG and motor impulsivity is bidirectional and varies between individuals.
123
J Gambl Stud
Attention Deficits Our analysis revealed a higher rate of omission errors in PGs than controls, pointing towards a deficit in sustained attention alongside the deficit in inhibitory control. However, conclusions made about the specific deficits in the GNG task are difficult to make, as the relative proportions of Go and No-Go trials are thought to reveal different sorts of information about underlying deficits (Smith et al. 2014). Tasks with Frequent Go and Rare No-Go trials are thought to be the best test of inhibition deficits, as the Go Response becomes the pre-potent/automatic response that must be withheld on No-Go trials. In contrast, tasks with frequent No-Go and Rare Go trials are thought to be the best test of sustained attention, as the No-Go Response is the pre-potent/automatic response that must be overridden on Go trials. The underlying deficit (inhibition vs. inattention) in equiprobable GNG tasks is difficult to disentangle and may instead reflect general response selection or discrimination deficits. Our analysis mostly found significant group differences in commission error rates in equiprobable GNG tasks; whilst only one out of three studies using Rare No-go trials revealed significant group differences in commission errors, suggesting evidence of inhibition per se may not be as strong as expected. Significant group differences in omission errors, in contrast, were found on GNG tasks with both equiprobable and Rare No-Go trials, providing good evidence for deficits in sustained attention. A further note of caution for conclusions about attention deficits must also be made as our samples did not exclude comorbid ADHD. Therefore, there is no clear evidence of attention deficits in PG that are not mediated by ADHD.
A Generalised Executive/Cognitive Control Deficit The finding of significantly slower Go RT for PGs in some studies is somewhat paradoxical. Intuitively, impulsive individuals are expected to respond quickly (or rashly) to stimuli, although the contrary was found in our meta-analysis. Alongside the finding of significantly more omission errors, this pattern of results may suggest that the core psychopathology in PG is not specifically motor impulsivity, but it may be a generalised deficit in executive or cognitive control (Wright et al. 2014) which consists of inhibitory control, attentional control, cognitive flexibility and working memory (Diamond 2013). This generalised deficit results in an inability to resolve conflicts between ‘‘stimulusdriven/automatic’’ and ‘‘voluntary’’ responses, leading to slow processing. Consequently, in tasks where there are differing proportions of Go and No-Go trials, there is conflict between the automatic tendency to execute a response and the voluntary tendency to withhold a response, leading impulsive individuals to be more susceptible to slow Go RTs (Kertzman et al. 2008). This explanation fits in with our results, as significant Go RT effect sizes were only observed in studies with differing proportions of Go and No-Go trials (when conflict was present), in contrast to equiprobable GNG tasks wherein there is no strong bias for going or stopping.
Clinical Implications An enriched understanding of the aetiology of PG has clinical implications. Laboratory and self-report measures of motor impulsivity may be used to distinguish different types of gamblers to help guide treatment. For example, Blaszczynski and Nower’s (1998) model categorises gamblers into three overlapping aetiological pathways; one where PG is caused
123
J Gambl Stud
purely by associative learning principles, one where PG is caused by associative learning as well as mood disturbances, and another where impulsivity is a predisposing factor in addition to emotional and behavioural factors. It is possible that those in the impulsivity pathway are more impaired on SSRT and GNG measures relative to those in the purely behavioural pathway or those in the mood-disturbance pathway. Hence, using these measures may help identify certain groups of gamblers, which may then guide treatment strategies, particularly whether the inability to inhibit inappropriate thoughts, behaviours and emotions should be a core target of intervention. Such measures may also be used as indicators of treatment resistance, treatment attrition, and the likelihood of relapse into PG.
Limitations of Current Research/Areas for Future Research To further strengthen conclusions about the relationship between PG and motor impulsivity, there are a number of gaps in the identified studies that require consideration. First, more acknowledgement of heterogeneity amongst PGs is required, including those who gamble for different reasons (to remove aversive mood states or those who relieve boredom/increase arousal), and/or those who fit in different pathways according to Blaszczynski and Nower’s (1998) model. More research is also required comparing motor impulsivity in PGs with and without ADHD. As the results of Rodriguez-Jimenez et al. (2006) would suggest, comorbid ADHD is likely to influence measures of motor impulsivity, and thus, the isolated effects of PGs on motor impulsivity are still unclear. Furthermore, the PG and control samples in the studies reported were mostly males. Although this is consistent with gender biases in the PG population, this limits generalizability to females. There is also little research on the relationship between the severity of gambling symptoms and performance on state and trait measures of motor impulsivity. For instance, it may be that ‘‘problem’’ gamblers, which are a less severe population of gamblers than PGs, may be less impaired on measures of inhibitory control. Lastly, more research is required on the neural basis for motor impulsivity in problem and pathological gamblers. Investigating whether the neural processes involved in inhibitory control are also impaired in such populations would provide a multidimensional account about PG aetiology alongside self-reported and behavioural measures.
Limitations of Current Meta-Analyses Alongside limitations of studies, our meta-analysis itself has several limitations. For the Stop Signal task, we did not include studies which measured the percentage of inhibited responses at different stop-signal delays (e.g. 50, 150, 250 and 350 ms). Of the 4 excluded studies which used this measure, 3 of these studies did not find significant differences between PGs and controls, suggesting that conclusions made about motor impulsivity in PGs may change depending on the measure that is used. Nonetheless, we did not include this measure as it does not consider the ‘‘race’’ between Go and Stop Processes. Therefore, the lack of differences may be due to the slowing of responses to the Go Signal (as results from the GNG task would suggest), such that longer SSDs need to be presented to PGs relative to controls to find failed inhibition. Another limitation of this meta-analysis is that for the GNG measures, we combined the means and SDs of different task conditions (e.g. different frequencies of Go and No-Go trials, different blocks of the GNG task), which is problematic as it is likely to be a cause of substantial heterogeneity between effect sizes on these measures. Comparing effects across different task conditions may have been beneficial and will become possible as the number of studies investigating inhibitory deficits in
123
J Gambl Stud
PGs grows. Lastly, the meta-analysis could have benefited from the inclusion of other subscales of the BIS-11, such as the attentional subscale, as this could have complemented omission error rates as a measure of sustained attention. Moreover, analysis of subscales from other questionnaires such as the UPPS Urgency Subscale, which may reflect a similar concept of motor impulsivity (Billieux et al. 2010), would have complemented the BIS-11 Motor Impulsiveness subscale as another self-report measure.
Conclusion In conclusion, our meta-analysis provides evidence that PGs without comorbid substance use have elevated motor impulsivity on both a behavioural and self-report level, suggesting that this is one feature of their psychopathology and may account for their difficulties controlling the urge to gamble in the face of negative consequences. However, future research must address the gaps in the current measures of motor impulsivity, the gaps in previous research, and the gaps in our meta-analysis in order reach a more solid conclusion. Compliance with Ethical Standards Conflict of interest The authors declare that they have no conflict of interest. Research Involving Human Participants This article does not contain any studies with human participants performed by any of the authors.
References Alessi, S. M., & Petry, N. M. (2003). Pathological gambling severity is associated with impulsivity in a delay discounting procedure. Behavioural Processes, 64, 345–354. Auger, N., Lo, E., Cantinotti, M., & O’Loughlin, J. (2010). Impulsivity and socio-economic status interact to increase the risk of gambling onset among youth. Addiction, 105, 2176–2183. Billieux, J., Gay, P., Rochat, L., & Van der Linden, M. (2010). The role of urgency and its underlying psychological mechanisms in problematic behaviours. Behaviour Research and Therapy, 48, 1085–1096. Billieux, J., Lagrange, G., Van der Linden, M., Lancon, C., Adida, M., & Jeanningros, R. (2012). Investigation of impulsivity in a sample of treatment-seeking pathological gamblers: A multidimensional perspective. Psychiatry Research, 198, 291–296. Black, D. W., Smith, M. M., Forbush, K. T., Shaw, M. C., McCormick, B. A., Moser, D. J., et al. (2013). Neuropsychological performance, impulsivity, symptoms of ADHD, and Clonginer’s personality traits in pathological gambling. Addiction Research & Theory, 21, 216–226. Blaszczynski, A., & Nower, L. (1998). A pathways model of problem and pathological gambling. Addiction, 97, 487–499. Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to metaanalysis. Chichester: Wiley. Borenstein, M., Hedges, L. V., & Rothstein, H. R. (2007). Meta-analysis: Fixed effects vs. random effects. www.Meta-Analysis.com. Brunner, D., & Hen, R. (1997). Insights into the neurobiology of impulsive behaviour from serotonin receptor knockout mice. Annals of the New York Academy of Sciences, 836, 81–105. Caswell, A. J., Morgan, M. J., & Duka, T. (2013). Inhibitory control contributes to ‘‘Motor’’—but not ‘‘Cognitive’’—impulsivity. Experimental Psychology, 60, 324–334. Chamberlain, S. R., & Sahakian, B. J. (2007). The neuropsychiatry of impulsivity. Current Opinion in Psychiatry, 20, 255–261. Ciccarelli, M., Malinconico, R., Griffiths, M. D., Nigro, G., & Cosenza, M. (2016). Reward preferences of pathological gamblers under conditions of uncertainty: An experimental study. Journal of Gambling Studies. doi:10.1007/s10899-016-9593.
123
J Gambl Stud Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159. Comprehensive Meta-Analysis (Version 3.3) [Computer software] (2014). Englewood. NJ: Biostat. Retrieved from http://www.meta-analysis.com. de Wilde, B., Goudriaan, A., Sabbe, B., Hulstijn, W., & Dom, G. (2013). Relapse in pathological gamblers: A pilot study on the predictive value of different impulsivity measures. Journal of Behavioral Addictions, 2, 23–30. Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64, 135–168. Donders, F. C. (1969). On the speed of mental processes. In W. G. Koster (Ed.) Attention and performance II. Amsterdam: North-Holland. (Original work published in 1868); 18681969. pp. 412-431. Dussault, F., Brendgen, M., Vitaro, F., Wanner, B., & Tremblay, R. E. (2011). Longitudinal links between impulsivity, gambling problems and depressive symptoms: A transactional model from adolescence to early adulthood. Journal of Child Psychology and Psychiatry, 52, 130–138. Enticott, P. G., Ogloff, J. R. P., & Bradshaw, J. L. (2006). Association between laboratory measures of executive inhibitory control and self-reported impulsivity. Personality and Individual Differences, 41, 285–294. Evenden, J. L. (1999). Varieties of impulsivity. Psychopharmacology (Berl), 146, 348–361. Fuentes, D., Tavares, H., Artes, R., & Gorenstein, C. (2006). Self-reported and neuropsychological measures of impulsivity in pathological gambling. Journal of the International Neuropsychological Society, 12, 907–912. Giorgetta, C., Giorgetta, C., Rattin, A., Guerreschi, C., Sanfey, A. G., & Bonini, N. (2014). To play or not to play: A personal dilemma in pathological gambling. Psychiatry Research, 219, 562–569. Goudriaan, A. E., Oosterlaan, J., de Beurs, E., & van den Brink, W. (2005). Decision making in pathological gambling: A comparison between pathological gamblers, alcohol dependents, persons with Tourette Syndrome, and normal controls. Cognitive Brain Research, 23, 137–151. Goudriaan, A. E., Oosterlaan, J., de Beurs, E., & van den Brink, W. (2006). Neurocognitive functions in pathological gambling: A comparison with alcohol dependence, Tourette syndrome and normal controls. Addiction, 101, 534–547. Grant, J. E., & Chamberlain, S. R. (2014). Impulsive action and impulsive choice across substance and behavioural addictions: Cause or consequence? Addictive Behaviors, 39, 1632–1639. Grant, J. E., Chamberlain, S. R., Odlaug, B. L., Potenza, M. N., & Kim, S. W. (2010). Memantine shows promise in reducing gambling severity and cognitive inflexibility in pathological gambling: A pilot study. Psychopharmacology (Berl), 212, 603–612. Grant, J. E., Odlaug, B. L., & Chamberlain, S. R. (2016). Neural and psychological underpinnings of gambling disorder: A review. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 65, 188–193. Grecucci, A., Giorgetta, C., Rattin, A., Guerreschi, C., Sanfey, A. G., & Bonini, N. (2014). Time devours things: How impulsivity and time affect temporal decisions in pathological gamblers. PLoS ONE, 9, e109197. Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003). Measuring inconsistency in metaanalyses service. BMJ, 327, 557–660. Hwang, J. Y., Shin, Y., Lim, S., Park, H. Y., Shin, N. Y., Jang, J. H., et al. (2012). Multidimensional comparison of personality characteristics of the big five model, impulsiveness, and affect in pathological gambling and obsessive-compulsive disorder. Journal of Gambling Studies, 28, 351–362. Kertzman, S., Lidogoster, H., Aizer, A., Kotler, M., & Dannon, P. N. (2011). Risk-taking decisions in pathological gamblers is not a result of their impaired inhibition ability. Psychiatry Research, 188, 2011. Kertzman, S., Lowengrub, K., Aizer, A., Vainder, M., Kotler, M., & Dannon, P. N. (2008). Go–No–Go performance in pathological gamblers. Psychiatry Research, 161, 1–10. Kraplin, A., Behrendt, S., Scherbaum, S., Dshemuchadse, M., Buhringer, G., & Goschke, T. (2015). Increased impulsivity in pathological gambling: Considering nicotine dependence. Journal of Clinical and Experimental Neuropsychology, 37, 367–378. Kraplin, A., Buhringer, G., Oosterlaan, J., van den Brink, W., Goschke, T., & Goudriaan, A. E. (2014). Dimensions and disorder specificity of impulsivity in pathological gambling. Addictive Behaviors, 39, 1646–1651. Lai, F. D. M., Ip, A. K. Y., & Lee, T. M. C. (2011). Impulsivity and pathological gambling among Chinese: Is it a state or a trait problem? BMC Research Notes, 4, 492–498. Lawrence, A. J., Luty, J., Bogdan, N. A., Sahakian, B. J., & Clark, L. (2009). Problem gamblers share deficits in impulsive decision-making with alcohol-dependent individuals. Addiction, 104, 1006–1015.
123
J Gambl Stud Ledgerwood, D. M., Alessi, S. M., Phoenix, N., & Petry, N. M. (2009). Behavioral assessment of impulsivity in pathological gamblers with and without substance use disorder histories versus healthy controls. Drug and Alcohol Dependence, 105, 89–96. Leppink, E. W., Redden, S. A., & Grant, J. E. (2016). Impulsivity in body-focused repetitive behaviour disorders: Disparate clinical associations between three distinct measures. International Journal of Psychiatry in Clinical Practise, 20, 24–31. Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, CA: Sage Publications. Lipszyc, J., & Schachar, R. (2010). Inhibitory control and psychopathology: A meta-analysis of studies using the stop signal task. Journals of the International Neuropsychological Society, 16, 1064–1076. Logan, G. D., & Cowan, W. B. (1984). On the ability to inhibit thought and action: A theory of an act of control. Psychological Review, 91, 295–327. Logan, G. D., Schachar, R. J., & Tannock, R. (1997). Impulsivity and inhibitory control. Psychological Science, 8, 60–64. Malloy-Diniz, L., Fuentes, D., Leite, W. B., Correa, H., & Bechara, A. (2007). Impulsive behaviour in adults with attention deficit/hyperactivity disorder: Characterisation of attention, motor and cognitive impulsiveness. Journal of the International Neuropsychological Society, 13, 693–698. Meule, A. (2013). Impulsivity and overeating: A closer look at the subscales of the Barratt Impulsiveness Scale. Frontiers in Psychology, 4, 1–4. Moeller, F. G., Barratt, E. S., Dougherty, D. M., Schmitz, J. M., & Swann, A. C. (2001). Psychiatric aspects of impulsivity. The American Journal of Psychiatry, 158, 1783–1793. Olmstead, M., Ougazzal, A., & Kieffer, B. L. (2009). Mu and delta opioid receptors oppositely regulate motor impulsivity in the signalled nose poke task. PLoS ONE, 4, e4410. Pagani, L. S., Derevensky, J. L., & Japel, C. (2009). Predicting gambling behavior in sixth grade from Kindergarten impulsivity a tale of developmental continuity. Archives of Paediatrics and Adolescent Medicine, 163, 238–243. Patton, J. H., Stanford, M. S., & Barratt, E. S. (1995). Factor structure of the Barratt impulsiveness scale. Journal of Clinical Psychology, 51, 768–774. Rentrop, M., Backenstrass, M., Jaentsch, B., Kaiser, S., Roth, A., Unger, J., et al. (2008). Response inhibition in borderline personality disorder: Performance in a go/nogo task. Psychopathology, 41, 50–57. Rodriguez-Jimenez, R., Avila, C., Jimenez-Arriero, M. A., Ponce, G., Monasor, R., Jimenez, M., et al. (2006). Impulsivity and sustained attention in pathological gamblers: Influence of Childhood ADHD history. Journal of Gambling Studies, 22, 451–461. Sharpe, L. (2002). A reformulated cognitive-behavioural model of problem gambling: A Biopsychosocial perspective. Clinical Psychology Review, 22, 1–25. Smith, J. L., Mattick, R. P., Jamadar, S. D., & Iredale, J. M. (2014). Deficits in behavioural inhibition in substance abuse and addiction: A meta-analysis. Drug and Alcohol Dependence, 145, 1–33. Stanford, M. S., Mathias, C. W., Dougherty, D. M., Lake, S. L., Anderson, N. E., & Patton, J. H. (2009). Fifty years of the Barratt Impulsiveness Scale: An update and review. Personality and Individual Differences, 47, 385–395. Trommer, B. L., Hoeppner, J. B., Lorber, R., & Armstrong, K. J. (1988). The Go–No-Go paradigm in attention deficit disorder. Annals of Neurology, 24, 610–614. van Holst, R. J., van der Meer, J. N., McLaren, D. G., van den Brink, W., Veltman, D. J., & Goudriaan, A. E. (2012). Interactions between affective and cognitive processing systems in problematic gamblers: A functional connectivity study. PLoS ONE, 7, e49923. Vassileva, J., Gonzalez, R., Bechara, A., & Martin, E. M. (2007). Are all drug addicts impulsive? Effects of antisociality and extent of multidrug use on cognitive and motor impulsivity. Addictive Behaviours, 32, 3071–3076. Wright, L., Lipszyc, J., Dupuis, A., Thayapararajah, S. W., & Russell, S. (2014). Response inhibition and psychopathology: A meta-analysis of go/no-go task performance. Journal of Abnormal Psychology, 123, 429–439. Zhou, Z., Zhou, H., & Zhu, H. (2016). Working memory, executive function and impulsivity in Internetaddictive disorders: A comparison with pathological gambling. Acta Neuropsychiatrica, 28, 92–100.
123