Searching for an Alternative: Effectiveness of a Visual Search Attentional
Bias Modification Training in Heavy Drinking Young Adults
By Carolien Torenvliet
Status of thesis: FinalVersion Date: December 6, 2017 ABSTRACT The current study examined the efficacy of online visual search attentional bias modification (VS-ABM) in reducing attentional biases (AB), alcohol intake and craving in heavy drinking young adults. In addition, the study tested whether (behavioral) impulsivity and levels of attentional control would moderate the effect of VS-ABM on AB. Sixty-eight participants completed the study (31 underwent VS-ABM training, and 37 underwent control training). Before and after training, measures of drinking, craving and AB were taken. AB were measured, using an alcohol visual search task (AVST) and a visual probe task (VPT). Before training, questionnaire-based measures of attentional control and impulsivity were also taken. Effects on AB showed inconsistent results as VS-ABM resulted in a significant reduction in attentional bias as measured by the VPT, but not the AVST. No statistically significant reductions in drinking and craving were observed. However, additional exploratory findings suggest that for those participants that showed higher AB at baseline, VS-ABM training resulted in a larger reduction of AB post training. Greater reductions in AB post-training were also observed at a trend level for participants with higher baseline scores relating to severity of alcohol use and behavioral impulsivity. The data suggest that VS-ABM can alter AB in some people more than in others. Future research should focus on testing the efficacy of VS-ABM in a motivated clinical population under more controlled conditions.
Key Words: Alcohol use, young adults, attentional bias, VS-ABM, impulsivity, attentional control
Student ID number: 10346163 Phone number: 0681004984 Email: c.torenvliet@uva.nl
Address: Pieter Nieuwlandstraat 61-B Postal code: 1093 XL, Amsterdam
Daily supervisor: Dr. Kyriaki Nikolaou UvA supervisor: Prof. Reinout Wiers Second assessor: Dr. Bram van Bockstaele Research Centre: Universiteit van Amsterdam Number of credits (EC): 25 (i.e. 700 hours)
Introduction
Alcohol use is remarkably present during young adulthood, with the largest percentage of drinkers (87.4%), heavy drinkers (18.1%), and excessive drinkers (14.1%) aged 20 to 30 years (CBS, 2017). Neurobiological research suggests that this- and other impulsive behavior during young adulthood can be explained by a tension between well-developed bottom-up processing exaggerating reactivity towards motivational stimuli, and underdeveloped top-down processes managing cognitive control (Casey & Jones, 2010). The distinction between bottom-up and top-down processes in young adulthood is also reflected in dual process models of cognition, that suggest that behavior can be explained as the result of two - sometimes opposing – processes: an ‘impulsive’ (similar to bottom-up) and an ‘reflective’ process (similar to top-down). The models make a distinction between involuntary automatic behavior and voluntary reflective behavior (Strack & Deutsch, 2004) using the ‘horse and rider’-metaphor in which the reflective rider aims to tame the impulsive horse (Friese, Hofmann, & Wiers, 2011). As the pre-frontal cortex (represented by the rider) is still developing during young adulthood, yet subcortical areas (represented by the horse) have already matured, the brain is naturally biased towards impulsive behavior during this developmental stage (Ernst, Pine, & Hardin, 2006; Geier & Luna, 2009). Together with an environment where alcohol is always available, socially accepted and provides status (Schultz, Nolan, Cialdini, Goldstein, & Griskevicius, 2007; Neighbors, Lee, Lewis, Fossos, & Larimer, 2007), young adulthood becomes a danger zone for developing potentially maladaptive alcohol-related behaviors.
Dominant impulsive- and weakened reflective processes are implicated in the development of alcohol use disorders (Wiers, Gladwin, Hofmann, Salemink, & Ridderinkhof, 2013). Impulsive reactions to alcohol and alcohol-related stimuli are strengthened by prolonged alcohol abuse and are reflected in the development of habitual alcohol-seeking behaviors (Everitt & Robbins, 2005), and incentive salience attributed to alcohol-related stimuli (i.e. cue triggered ‘wanting’; Berridge & Robinson, 2003). At the same time, reflective processes which aim to control impulsive reactions are weakened (e.g. Wiers et al., 2013). For example, alcohol abuse has been shown to impair executive functioning (Tapert et al., 2004), and areas of neuroplasticity during young adulthood such as the pre-frontal cortex, are particularly vulnerable to alcohol-induced neurotoxicity (Crews & Boettiger, 2009). Thus, excessive alcohol use in young adulthood may enhance the imbalance between reflective and impulsive processes, which may lead to maladaptive drinking or to alcohol use disorders later in life (Crews, He, & Hodge, 2007).
As young adulthood may be a vulnerable period for developing maladaptive drinking, yet also a time where cognitive plasticity flourishes (Crone, 2009), young adults might benefit eminently from a cognitive training that aims to reduce impulsive responses to alcohol and alcohol-related stimuli. A number of cognitive interventions for alcohol use disorders have focused on targeting attentional biases (AB) to alcohol-related stimuli (e.g. Schoenmakers, Wiers, Jones, Bruce, & Jansen, 2007; Fadardi & Cox,
2009). AB to alcohol-related stimuli are described as deficits of selective attention, whereby due to the incentive value acquired by these cues through repeated pairings with the drug reward, attention is preferentially grabbed by these alcohol-related cues. Biases consist of two components: (1) initial orienting towards alcohol-related stimuli, (2) difficulty to disengage from alcohol-related stimuli (Phaf & Kan, 2007; Field, Munafò, & Franken, 2009). A bidirectional relation of AB with alcohol intake and craving is confirmed by experimental research, with high AB inducing larger craving and higher alcohol intake (Field & Eastwood, 2005), and alcohol intake enhancing AB at low, priming doses (Field, Wiers, Christiansen, Fillmore, & Verster, 2010; Nikolaou, Field, Critchley, & Duka, 2013). Moreover, AB predicted relapse rates after periods of abstinence (Wiers et al., 2013) and an increase of AB during treatment was related to increased relapse rates during or after treatment (Cox, Hogan, Kristian, & Race, 2002).
Two paradigms have been commonly used to assess AB in various populations: the visual probe task and the alcohol-related Stroop task. In the visual probe task (VPT), a picture of alcohol-containing and non-alcohol-containing beverages are displayed next to each other. A target (a down- or upwards-pointing arrow) replaces the picture of the non-alcohol-containing or alcohol-containing drink and participants indicate the direction of the target as quickly as possible via key press. Response times on non-alcohol target replacements are compared to response times on alcohol target replacements with faster responses on alcohol trials compared to non-alcohol trials indicating a bias towards alcohol and vice versa. When assessing AB with the alcohol-related Stroop task, participants name the color of an alcohol-related or neutral printed word or picture border while ignoring that word/picture. Again, color naming speed on alcohol trials is compared to neutral trials as an indication of AB.
In the past decade, researchers have developed Attentional Bias Modification (ABM) training paradigms that aim to reduce biases to alcohol-related stimuli. These training paradigms consist of modified versions of the VPT and alcohol-related Stroop task. Thus, during visual probe ABM training, for example, only non-alcohol trials are presented, so that throughout the training participants must focus attention in each trial attention on the non-alcoholic stimulus. Studies using single session visual probe ABM in student samples (Schoenmakers et al., 2007; Field, Duka & Eastwood, 2007), showed a decrease in AB towards the trained stimuli. However, effects did not generalize to new stimuli or to an alternative AB task, nor were there effects on drinking/craving measures. When five sessions of visual probe ABM were compared to placebo training in a clinical sample with in- and outpatients (Schoenmakers, et al., 2010), trained patients showed reduced AB which generalized to new stimuli. No effect was found on drinking/craving measures, but the trained group took longer to relapse than the placebo group. In a study using ABM based on an alcohol-related Stroop task (Fadardi & Cox, 2009), four training sessions decreased AB in the trained stimuli in hazardous and harmful drinkers. The study did not test the effects
of ABM on AB to new stimuli, but it reported a decrease in alcohol consumption, improvements in readiness to change, and a sense of increased control over drinking in harmful drinkers, which was maintained at the three-month follow-up. However, in this study participants served as their own wait-list controls, whereby a 4-week waiting period prior to training was used to assess this. In a further study with self-selected problem drinkers using online training with the alcohol-related Stroop task (Wiers et al., 2015), no significant differences between real and placebo trained groups were found, as both groups showed an equal decrease in drinking from pre- to post-training. Thus, the results of ABM interventions have been rather small in non-clinical samples, with no significant changes in AB and/or no change in drinking and craving. However, findings in clinical populations have been more promising. Recent evidence from the use of ABM to ameliorate symptoms of anxiety, suggest that insufficient ABM-induced effects on AB may be the reason for minimal or no effects on behavior and clinical outcomes (Van Bockstaele, Salemink, Bögels, & Wiers, 2017).
Recently, researchers in the field of anxiety developed an ABM intervention based on the Visual Search task, which might also be promising for training attention away from alcohol-related stimuli. During Visual Search ABM (VS-ABM; Dandeneau, Baldwin, Baccus, Sakellaropoulo, & Pruessner, 2007) anxious people search for a happy face in a grid of angry faces. VS-ABM trains attention away from negative stimuli, while also training people to allocate attention to, and actively search for, an alternative positive emotion (De Voogd, Wiers, Prins, & Salemink, 2014). The effectiveness of VS-ABM in decreasing AB is demonstrated by an experimental study in which VS-ABM, but not visual probe training, enhanced attention for positive information in a healthy, youth sample (age 11-18; De Voogd et al., 2016). Furthermore, VS-ABM was able to reduce self-reported social phobia in social phobic adolescents, and diagnosis severity ratings and the number of diagnoses in anxious children (De Voogd et al., 2014; Waters, Pittaway, Mogg, Bradley, & Pine, 2013). Importantly, in these studies AB were assessed using the VPT and a newly developed Visual Search Task (VST). The psychometric properties of the VST have been shown to be better than those of the VPT or of the alcohol-related Stroop task (Van Bockstaele et al., 2017), both of which are thought to have poor reliability (Ataya et al., 2012). To date, no study has investigated the effectiveness of an alcohol-related VS-ABM in either a healthy- or an alcohol-dependent sample in reducing AB to alcohol-related stimuli. The current study explored the possibility that VS-ABM decreases alcohol AB, craving, and alcohol intake in heavy drinking young adults. Effects on AB were examined using both the VPT and the alcohol VST (AVST) to look at whether the effect of VS-ABM generalized to other measures of AB.
In addition to exploring the efficacy of ABM, the current study also examined whether VS-ABM may be beneficial for some people more than others. Models of AB in alcohol dependence highlight impulsivity and attentional control as possible moderators of change (Field & Cox, 2008; Wiers,
et al., 2013). This seems plausible following the dual process models which state that behavior is a consequence of both impulsive and reflective processes. Heightened impulsivity and lowered attentional control as a consequence of both neurobiological characteristics of young adults and alcohol abuse might enlarge AB. Consequently, larger effects of VS-ABM might be expected for individuals with heightened impulsivity and lowered attentional control. A recent meta-analysis showed a robust relationship between impulsivity (primarily behavioral impulsivity) and AB (Coskunpinar & Cyders, 2013). Highly impulsive individuals also show narrower focus on alcohol than low impulsive individuals (Hicks, Fields, Davis, & Gable, 2015), which might indicate higher AB. In a longitudinal study, Janssen, Larsen, Vollebergh, & Wiers (2015) proposed that the interaction of alcohol use and (general) impulsivity would predict later AB. Even though AB predicted later alcohol use in this study, AB was not predicted by impulsivity or its interaction with alcohol use. Possibly, general trait impulsivity does not relate to AB as strongly as behavioral impulsivity, as was also suggested by Coskunpinar & Cyders (2013). By contrast, Weckler et al. (2017) showed that highly impulsive young adults benefited more than their less impulsive peers from an approach bias training in quitting smoking . This finding highlights the possible importance of general impulsivity next to behavioral impulsivity. Attentional control might be related to AB as it involves reflective processes. A correlational study (Garland, Boettiger, Gaylord, Chanon, & Howard, 2012) showed that high levels of AB in alcohol dependent patients related with low trait mindfulness. An important feature of mindfulness is the ability to control thoughts (Kabat-Zinn, 2003). This supports the hypothesis that low attentional control may be related to high AB, and that individuals with low attentional control might thus benefit more from ABM. However, studies that focused more narrowly on attentional control have resulted in mixed findings. A correlational study measuring executive control by using a flanker-task based Attention Network Task (ANT; Van Hemel-Ruiter, de Jong, Ostafin, & Wiers, 2015) found a relation between AB and alcohol use only in young adolescents with low executive control as predicted by dual process models. In an earlier study, the opposite effect was observed whereby only adolescents and young adults with high self-reported attentional control higher alcohol use was related to higher alcohol AB (Willem, Vasey, Beckers, Claes, & Bijttebier, 2013). Thus, both (behavioral) impulsivity and attentional control have previously shown mixed results regarding their relation to AB. Also, no research has examined their role in the efficacy of VS-ABM. Therefore, the hypothesis tested in the current research is that heightened (behavioral) impulsivity and/or lowered attentional control related to higher effectiveness of VS-ABM in reducing alcohol AB.
The main objectives of the current study were to investigate the effectiveness, and mechanisms of multiple session online VS-ABM in heavy drinking, non-dependent young adults. A modified, alcohol-related VS-ABM paradigm was used, in which participants searched for a non-alcoholic beverage in a grid of alcoholic beverages. Training effects on this task were compared with placebo training. Following the model shown in Figure 1, it was predicted that VS-ABM would decrease AB, and that this effect would generalize to a different measure of AB (VPT). In addition, the study examined the effects of VS-ABM on alcohol intake and craving. These effects were expected to be moderated by high (behavioral) impulsivity and low attentional control.
Methods
ParticipantsA subset of an opportunity sample (n = 93) was used with non-dependent heavy alcohol drinkers (between 10-40 Dutch standard drinking units per week). Participants were recruited from the University of Amsterdam through an advertisement on the University’s DPMS (Digitaal Proefpersoonuren Management Systeem) requesting for social drinkers who enjoyed drinking 10 or more drinks on a weekly basis. 21 participants were excluded because they did not complete the training. One participant was excluded for not completing the first- and last AVST assessment. For the purpose of the current study, only data from students were analyzed, giving a total sample of 68 participants (94% of the total included sample, 9 = male, 59 = female). Age ranged from 18 to 26 (mean = 19.78, SD = 1.09). All participants were right handed and able to understand Dutch. Participants did not suffer from- or were receiving medication for mental or neurocognitive illness, had no DSM diagnosis for substance abuse, were not regular cannabis users or heavy tobacco smokers (more than 20 a day). Participants were randomly allocated (and stratified by gender) to either the control training (n = 37, 31 females) or the VS-ABM training (n = 31, 29 females). The training was presented double blind: Participants were not informed about which training group they were in and experimenters did not know if participants were in the real or the placebo training. This was achieved using the online platform LOTUS, which was developed by the technical department of the University of Amsterdam and is designed for the delivery of experimental procedures online. The study was approved by the University of Amsterdam, department of Psychology Ethics Committee.
Materials
Attentional bias measures
All stimuli used in the following tasks originated from the Amsterdam Beverage Picture Set (ABPS; Pronk, van Deursen, Beraha, Larsen, Wiers, 2015)
The Alcohol Visual Search Task (AVST) is a measure of AB to alcohol-related stimuli, based on the VS-AB task in anxiety research (Dandeneau et al., 2007). On each trial, participants first clicked on a central fixation cross to ensure participants fixated on the center of the screen before each trial. To avoid accidental clicking at this stage, responses were not recorded for the first 200 ms. After the fixation cross participants saw a grid of 16 pictures, of which 15 were ‘distracters’ and 1 was a ‘target’. They were instructed to look for and click on the target with the mouse as quickly and as accurately as they could, while ignoring the distracters. Participants could take as long as they needed to complete each trial with no maximum response time. After each correct response intertrial interval time was 500 ms. After incorrect responses, a feedback screen was shown for 2 seconds. The task contained 2 ‘neutral’ and 2 ‘alcohol’ blocks, with 18 trials each. Blocks were presented in a counterbalanced order. In the neutral blocks, the target picture was a non-alcohol-containing drink and all distracters were alcohol-containing drinks. In the alcohol blocks this was reversed. One neutral and one alcohol block contained active representations of the picture stimuli (e.g. a person opening a bottle), and one neutral and alcohol block contained passive representations of the stimuli (e.g. picture of a free-standing bottle). AB was measured as a difference score between the average reaction time in the neutral block minus the average reaction time in the alcohol block (AB = neutral – alcohol). High difference scores indicated high AB towards alcohol-related stimuli. The same set of stimuli was used in the pre-training AVST and during VS-ABM training. Post-training AB were measured using a different set of stimuli to assess generalization.
Boendermaker, Maceiras, Boffo, and Wiers (2016) who were the first to use this task for assessing alcohol AB did not assess split half reliability, yet a reasonable split-half reliability was found in anxiety research (r = .43-.59; Van Bockstaele et al., 2017). However, convergent validity with other AB-measures, including VPT, was poor (all rs <.20; Van Bockstaele et al., 2017).
The Visual Probe Task (VPT; Schoenmakers et al., 2007) also assesses AB for alcohol-related stimuli. After a central fixation cross was shown for 1000 ms, a picture of an alcohol-containing drink and a picture of a non-alcohol containing drink were displayed next to each other for 2000 ms. Immediately after picture offset, a target (down- or upwards pointing arrow) appeared in the location of one of the two pictures. Participants indicated the direction of the target (up/down) by pressing the corresponding key on the keyboard (E-key or I-key). Intertrial interval time was 500 ms after each response. AB was measured as a difference in reaction time in response to the target, when the target replaced an image of an alcohol-containing vs. a non-alcohol-alcohol-containing drink. Here too, high scores were indicative of increased biases to alcohol-related stimuli. Internal reliability is considered poor (close to zero, Ataya et al., 2012). However, this task is widely used to assess AB in alcohol dependence, so keeping poor psychometric properties in
mind, it was still a useful addition to this study, and allowed us to compare the effects of VS-ABM training on a widely used task of AB.
Questionnaires
The Alcohol Use Disorders Identification Test (AUDIT; Babor, de la Fuente, Saunders, & Grant, 1992) is a 10-item brief screening scale to identify individuals with alcohol problems. Items range from 0 (never/no) to 4 (daily/yes). Total scores range was 0-40, where higher scores indicated larger alcohol problems. Internal consistency and test-retest reliability are good (α > .80, r = .81; Reinert & Allen, 2002).
The Desires for Alcohol Questionnaire (DAQ; Love, James, & Willner, 1998) is a 14-item questionnaire that assesses craving. It yields four factors representing ‘positive and negative reinforcement’, ‘strong desires’, ‘mild desires’ and ‘control over drinking’. An adjusted 5-point Likert scale ranging from 1 (not at all) to 5 (strongly agree) was used (Kramer et al., 2010), to provide a more normal distribution of the responses.
In the Timeline Follow Back (TLFB; Sobell & Sobell, 1995) participants provided retrospective estimates of daily drinking of the last month, and between training sessions. Convergent validity is good (r = .7-.8; Dennis, Funk, Godley, Godley, & Waldron, 2004).
The Attentional Control Scale (ACS; Derryberry & Reed, 2002) is a 20-item questionnaire measuring attentional control on a 4-point scale ranging from 1 (almost never) to 4 (always). Total scores range was 20-80, with higher scores indicating better attentional control. Internal consistency estimates range from fair to good (α = .71, α = .88; Olafsson, 2011).
The Barratt Impulsiveness Scale 11 (BIS-11; Patton & Stanford, 1995) is a 30-item questionnaire assessing facets of impulsivity on a 4-point Likert scale ranging from 0 (rarely never) to 3 (almost always). Total scores range was 0-90, where higher scores indicate more impulsivity. Three facets of impulsivity are measured (range 0-30): ‘motor impulsivity’, ‘attentional impulsivity’ and ‘non-planning’. In addition to the total score obtained in the questionnaire, the subscale motor impulsivity is used as a measure of behavioral impulsivity as it has been shown to correlate highest with impulsive reactions in tasks such as the stop signal task (e.g. De Wit, 2009; Moeller et al., 2001). Internal consistency and test-retest reliability are good (α = .83, r=.83; Stanford et al., 2009).
The Beck’s Depression Inventory (Beck, Steer, & Brown, 1996) is a 21-item questionnaire measuring depressive symptoms on a 4-point scale ranging from 0 (not really) to 3 (very much). Total scores range from 0-63, with higher scores indicating more depressive symptoms. Cut-off points were 10 for mild depression, 29 for severe depression.
The Core Scale (Presley, Cashin, Meilman, & Lyerla, 1996) is a measure of drug use. A part of the Core questionnaire was used in the pre-training session to measure the use of a variety of drugs over the past month (tobacco, alcohol, marijuana, hallucinogens, amphetamines, sedatives, cocaine, opiates, inhalants, stimulants and other party drugs) Participants responded on a scale of 1 – 11 (1 = 0 times, 2 = 1-10 times... 11 = > 91 times).
Demographic Data included participant’s age, gender, current country of residence, place of birth, mother and fathers place of birth, mother tongue language, marital status, current medication, highest level of obtained education and income.
Tailored Advice Questionnaire included questions related to why participants wanted to do this training, i.e. to reduce drinking or completely stop drinking. They were also asked how important it was for them to reduce/stop drinking and the confidence they had in pursuing this goal. They rated importance and confidence on a scale ranging from 1 (not very) to 10 (very).
Training paradigms
VS-ABM training was based on the VS-ABM used in anxiety research (Dandeneau et al., 2007). On each training trial participants searched for the image of the non-alcoholic beverage in a 4x4 grid, containing 15 pictures of distracting images of alcoholic beverages. After each trial a control grid was shown (containing flower pictures). Each training session contained 144 trials, with a total duration of 20 minutes. Training was performed on a computer.
The Placebo training was identical to that used by Dandeneau et al., (2007). Participants searched for a 5-petal flower in a 4x4 grid containing 15 pictures of 7-5-petal flowers. After each trial a VS-ABM grid is shown (containing alcohol pictures), to guarantee that both groups are exposed to the same number of alcohol-related stimuli. Other characteristics will be the same as in the VS-ABM training.
Procedure
A total of 8 sessions were completed online, using the LOTUS online server. During the pre-training session participants were asked to complete the demographic questions, followed by the completion of the AUDIT, BDI, BIS, ACS, Core, TLFB, and DAQ. At the end of the AUDIT, tailored advice about the risks of drinking was given to participants depending on the extent of their alcohol use. Following questionnaire completion, participants performed the pre-training AVST and VPT assessment to measure baseline levels of AB. After the pre-training session participants completed a total of 6 training sessions in 3 weeks. During each of these training sessions all participants completed the DAQ, TLFB and allotted visual search training. One week after the last training session participants had to complete the post-training assessment session. In the post-training session participants completed the BDI, Core, TLFB, DAQ, post-training AVST and VPT assessment again to measure AB after training. One month post-training, a follow up session was carried out where participants completed the BDI, Core, TLFB and DAQ. Unfortunately, response rates in the follow-up session were too low to be of use. After follow-up participants had the opportunity to continue training or to start (for the control group). For this, participants were able to log into the website as they had done previously during training.
Results
1Data cleaning and outlier removal
Data cleaning for both the pre- and post-training VPT assessments was conducted in the same way as reported previously (Elfeddali, de Vries, Bolman, Pronk, & Wiers, 2016). One participant was removed both at pre- and post-training VPT for having too low accuracy scores. Low accuracy was defined as having accuracy scores 3 SDs below the group’s mean accuracy for a specific session.
The same procedure for data cleaning and outlier removal was applied to the data derived from both the pre- and the post-training AVST assessments. Thus, participants were considered outliers and their data was removed from all analyses if their accuracy at an assessment session was 3 SDs below the group’s mean accuracy for that session (pre-training: n=4, post-training: n=2). In addition, for each assessment using the AVST, trials with reaction times higher than 3 SDs from an individuals’ mean reaction time in each condition (alcohol/non-alcohol) were also removed (pre-training total removal: 39
1 All analyses were performed with a Bonferroni correction to reduce type-I errors. Outcomes with p-values lower than .05, yet higher than the corresponding α-level, are referred to as marginally significant.
alcohol trials (1.7%) 35 non-alcohol trials (1.5%); post-training total removal: 34 alcohol trials (1.4%), 36 non-alcohol trials (1.5%)).
Pre-training AVST Assessment task reliability analyses
Following outlier removal, we first checked for differences between blocks containing active image representations and blocks containing passive image representations. No differences in reaction time or accuracy between these blocks were expected, as the inclusion of active and passive stimuli intended to allow for more variance in the assessment of biases and more stimulus variability during training. As expected, reaction times and accuracy scores did not differ between active and passive stimuli at pre-training (all p’s >.10). As a result, for all subsequent analyses, blocks with active and passive images were not analyzed separately. In order to assess task reliability, for each participant, we computed the average reaction time of odd and of even trials for the alcohol and neutral blocks of the pre-training AVST assessment. AB scores were then computed for odd trials, by subtracting the average reaction time on odd trials derived from the alcohol blocks, from the average reaction time on odd trials derived from the neutral blocks. AB scores were computed in a similar way for even trials. AB scores derived from odd and even trials were then correlated to assess split half reliability at pre-training (as in Van Bockstaele et al., 2017). Split half reliability (Spearman-Brown) of AVST AB was satisfactory (r = .68, p <.001).
Analyses of baseline indices
At baseline, groups did not differ on age, sex, and nationality, depression and cannabis, nicotine and other drug use (see Table 1). Furthermore, groups were also well matched at baseline on AB scores, as assessed using both the VPT and AVST assessments (Table1). Groups did also not differ on baseline measures of variables related to severity of alcohol problems (AUDIT), alcohol intake (TLFB), and craving (DAQ), (behavioral) impulsivity (BIS), and attentional control (ACS, Table 1). Lastly groups did not differ on baseline measures of motivation towards the training (TAQ). Analyses of baseline motivation also indicated that 65 participants wanted to reduce their drinking, whilst only 3 wanted to quit drinking completely. Average levels of motivation were moderately high (>5). Although this questionnaire is a proxy to a motivational state, this indicates that the sample was committed to changing.
In general participants showed significantly faster responses on alcohol blocks compared to neutral blocks of the AVST (paired-t (63) = -3.81, p<.001), suggesting a general AB towards alcohol-related stimuli at baseline. This was not the case for the VPT (paired-t (6) = -.799, p = .427).
Baseline AB scores, as assessed using the AVST did not significantly correlate with AUDIT scores, baseline craving scores, or with baseline indices of drinking, as assessed using the TLFB (DMT4
and TPW, see Table 2). However, baseline AB scores as assessed using the VPT correlated significantly with TLFB DMT4, and there were marginal relationships with VPT AB scores and AUDIT and one of the craving indices (mild desires; see Table 2).
Table 1. Demographic baseline characteristics of both training groups VS-ABM training
(n = 31)
Control training (n = 37)
Mean/n SD/% Mean/n SD/% t/χ2-value
Age 19.87.94 1.45 20.16 1.89 -.717 Females, n (%) 29 93.55 30 81.08 1.327 Dutch Nationality, n (%) 25 80.65 33 89.19 .419 BDI 2.13 1.87 2.12 2.99 .042 Cannabis usea 1.10 .30 1.19 .40 -1.112 Nicotine usea 2.29 2.40 2.69 3.29 -.567
Other drug usea 7.35 .61 7.31 .58 .340
AVST AB 246.30 801.45 398.92 588.97 -.877 VPT AB -.55 23.78 4.56 21.33 -.926 AUDIT 11.55 4.77 12.89 5.98 -1.010 TLFB DMT4b 3.42 2.61 4.00 2.59 -.913 TLFB TPWc 12.71 10.91 12.50 8.55 .089 DAQ Reinforcementd 7.42 3.722 7.84 3.60 W = 515 DAQ Strong desiresd 4.8791 1.75 5.14 1.70 W = 456 DAQ Mild desires 8.71 3.86 9.57 4.37 -.850 DAQ Control over drinkingd 3.00 1.65 3.41 1.69 W = 488.5
ACS 51.93 7.89 51.05 8.71 .429 BIS Total 71.97 4.36 72.19 4.50 -.204 BIS Motor 23.87 3.25 23.76 3.02 .143 TAQ Importance 6.12 2.13 6.18 1.61 .133 TAQ Confidence 6.27 1.72 5.42 2.18 1.796 TAQ Reduce, n (%) 30 96.77 35 94.59 .190 a
Usage in the past month as measured by the Core. b
TLFB DMT5 = Time Line Follow Back Days More Than 4 - indicates binge drinking in the past month. c TLFB TPW = Time Line Follow Back Total (number of drinks) Per Week.
d
Pre-training differences on these scales were tested non-parametrically using the Wilcoxon Rank-Sum test.
Table 2. Correlations between AB and primary outcome measures at baseline VPT AB TLFB DMT4 TLFB TPW AUDIT DAQ Reinf. DAQ Strong DAQ Mild DAQ Control AVST AB -.16 .03 .15 .10 -.02 -.02 -.03 -.13 VPT AB .32** .10 .28* .16 .20 .25* .13 * = p < .05, ** = p <.01
Analyses of training effects on attentional biases
To analyze the effect of the training on AB as assessed using the AVST, a 2x2 mixed ANOVA was performed, with Group (Trained vs. Control) as the between subjects factor and Time (pre-training vs. post-training) as the within subjects factor. There was no significant Group x Time interaction (F
(1,126) = .485, p = .487), suggesting that the VS-ABM training did not result in a significant change in AB from pre- to post-training by comparison to control training. Figure 2 shows an average decline in AB in the trained group, yet a substantial dispersion in this decline indicated by the error bar for the trained group at post-training illustrates a large variance in effectiveness of VS-ABM across participants. Specific characteristics of participants related to this variance were further explored in the “Moderator- and exploratory analyses”-section below. By looking at the graph (Figure 2), it was also noticed that mean AB were reduced largely in the trained group by comparison to the control group. Therefore, post-hoc analyses were used to analyze whether general AB apparent at pre-training were no longer significant in the trained and/or control group at post-training. This was indeed the case for the trained group (paired-t (28) = -.11, p = .917), yet not for the control group (paired-t (38) = -3.90, p < .001).
Figure 2. AVST AB in each group at pre- and post-training. Error-bars display standard errors of the mean.
To analyze the effect of VS-ABM on AB as assessed using the VPT, a 2x2 mixed ANOVA was performed, with Group (Trained vs. Control) as the between subjects factor and Time (pre-training vs. post-training) as the within subjects factor. There was a significant Group x Time interaction (F (1,130) = 13.65, p<.001), suggesting that VS-ABM training resulted in a significant change in AB from pre- to post-training by comparison to control training. Main effects of time and group were also significant (respectively F (1,130) = 9.65, p <.001, and, F (1,130) = 22.65, p <.001. Figure 3 shows that the effect was in the expected direction, hence a larger decline in alcohol AB from pre- to post training in the trained group compared to the control group (t (46.43) = -4.059, p <.001).
Bayesian analyses
Due to the fact that we found disparity in the effects of training on AB when examining the effects on the VPT relative to the AVST, Bayes factors (B) were used to assess strength of evidence for the alternative hypothesis, H1 (reduced bias post- relative to pre-training), over the null, H0 (i.e. not effect of training on AB) for each task individually (Wagenmakers et al., 2017). A B of above 3 indicates substantial evidence for H1 over H0 and below 1/3 substantial evidence for the H0 over Hl. All Bayes factors, B, reported here represent the evidence for H1 relative to H0. Bs between 3 and 1/3 indicate data insensitivity (see Dienes, 2014). Here, BH(0, x) refers to a Bayes factor in which the predictions of H1 were modeled as a half-normal distribution with an SD of x ( Dienes & McLatchie, 2017); the half-normal can be used when a theory makes a directional prediction where x scales the size of effect that could be expected. As a prior, the effects from De Voogd et al. (2016) were used, because their training also consisted of 6 sessions and was theoretically identical to the training in the current study. Thus x was the pre- to post-training mean difference in AB reported in the VS-ABM group for AVST AB and VPT AB respectively (μ = 2027.20 and μ = 1.50, De Voogd et al., 2016).
Using the Zoltan Dienes application2 a B of .36 for found the pre-post AVST AB difference in the trained group (μ = 264.16, SE = 432.67), indicating an insensitive effect. The same Bayesian technique was used to assess the evidence for the alternative hypothesis in the VPT AB effect, hence that pre-training VPT AB was larger at pre-training compared to post-training in the trained group (μ = 33.85, SE = 7.68). A B of 2.25 was found, indicating an insensitive effect. Thus, although a significant difference was found using frequentist statistics when examining the effect of the training on biases as assessed using the VPT, Bayesian analysis could not confirm the alternative hypothesis that VPT AB was larger at pre-training compared to post-training in the trained group. Thus, it seems that in our study, it is uncertain whether the training resulted in reliable changes in biases.
Exploratory analyses
Because AVST and VPT showed inconsistent results regarding the effect of VS-ABM on AB, more in depth exploratory analyses were performed to assess where these inconsistencies came from. This was done by analyzing effects on the AVST and VPT specifically for alcohol trials and non-alcohol trials. Two 2x2x2 ANOVA’s were performed with Group (Trained vs. Control) as the between subjects factor and Time and Stimulus (pre-training vs. post-training and alcohol vs. non-alcohol) as within subject factors. No three-way interaction was found for either task (p’s >.20, means and SEs can be found in Appendix A, Table A1). For the AVST, contrasts showed a significant Group x Time interaction (F
2
(1,252) = 6.44, p = .012), indicating a general speeding from pre- to post-training in the trained group by comparison to the control group regardless of stimulus type. For the VPT contrasts showed no Group x Time interaction (F (1,260) = .02, p = .880), yet in the trained group a stimulus specific effect was found as indicated by significant speeding at non-alcohol trials from pre- to post-training (paired-t (30) = -2.24, p = .03). Notably, this effect was not found for the trained group at alcohol trials (paired-t (30) = .891, p = .380). In the control group, no significant differences from pre- to post-training were observed for either trial type (non-alcohol: paired-t (35) = -1.70, p = .102, alcohol: paired-t (35) = -1.84, p = .074). Taken together, this indicates that VS-ABM resulted in stimulus-specific speeding in the trained group for non-alcohol trials in the VPT, and a general speeding for both trial types in the AVST.
Correlations between AVST AB and VPT AB were inspected to assess whether discrepancies between the two tasks in the decrease of AB observed at the group level, were also observed on an individual subject level as an indication of convergent validity. Surprisingly, correlational analyses revealed a significant positive correlation between post-training AVST AB and VPT AB (r (63) = .36, p = .003), yet not between pre-post differences in AVST AB and VPT AB (r (65) = .15, p = .233). Arguably, this is due to the abovementioned differences in specific- and general speeding on the AVST and VPT.
Analyses of training effects on alcohol drinking and craving
Scores for all craving subscales except ‘mild desires’ were extremely positively skewed at post-training, therefore a Wilcoxon Rank-Sum analyses (using R, equivalent to Mann-Whitney) were performed to assess post-training differences in craving between the two groups. There were no significant group differences at post-training for ‘positive/negative reinforcement (W = 544, p = .702), and ‘control over drinking’ (W = 522.5, p = .446). A trend was found for ‘strong desires’ (W = 443, p = .060), with a lower mean in the trained group (4.87) by comparison to the control group (5.14). To analyze the effect of the training on ‘mild desires’, a 2x2 mixed ANOVA was performed, with Group (Trained vs. Control) as the between subjects factor and Time (pre-training vs. post-training) as the within subjects factor. There was no significant Group x Time interaction (F (1,132) = .025, p = .876). Combined these results suggest that the VS-ABM training did not result in a significant change in craving from pre-to post-training by comparison to control training for any of the subscales.
For alcohol intake, each participant’s alcohol use was first plotted. The plots represented alcohol intake from baseline to the post-training session. Average number of drinks per days between training was used as a measure of alcohol intake, as training days between sessions varied substantially between participants. For the first session (training) a general recall question about the week before pre-training was used (TLFB: ‘total number of drinks last week’) divided by 7. The drinking score for all other sessions was calculated by dividing the number of drinks reported in a given training session, by the
number of days since the last training session. For each participant, the Area Under the Curve (AUC) of these plots was calculated. These indices were then compared between the groups.
As can be seen in Figure 5, marginal differences in alcohol intake between training sessions were found between the trained group and the control group. AUC analysis showed that the area under the curve did not differ significantly between groups (t (57.36) = -.79, p = .430). This suggests that VS-ABM training did not result in a significant change in drinking from pre-to post-training by comparison to control training. Including the strength of motivation (TAQ Importance) as a covariate did not change the results. There was no significant interaction between group and the covariate (F (1, 57) = .337, p = .564), nor a main effect of the strength of motivation (F (1) = 3.103, p = 0.084).
Furthermore, AUC was also not significantly correlated with neither the pre-post differences in AVST AB nor the pre-post differences in VPT AB (AVST: r = -.06, p = .615, VPT: r = .17, p = .185).
Figure 5. Average number of drinks per day between each training session. Error-bars display 95% CI. Moderator- and exploratory analyses
To explore the dispersion in the effect of VS-ABM training on AVST AB, several regression analyses were performed to see whether any particular participant characteristic could specifically account for the effectiveness of VS-ABM. Hypothesized moderators were general impulsivity (BIS Total), motor impulsivity (BIS Motor) and attentional control (ACS). Additional exploratory moderators were severity of alcohol use at pre-training (AUDIT), binge drinking at pre-training (TLFB DMT4), and pre-training AB (AVST AB). At interest were interactions between the moderator and group. Instead of using pre-training AB as covariate, post-training scores were regressed on pre-training scores to create an adjusted post-training scores expressed as:
adj_posti = posti – posti’, where
Next, adj_post training was implemented in the regression equation, expressed as:
adj_posti = β0 + β1*moderatori + β2*moderatori*groupi + εi (2) This is based on the residual change score method explained by Kisbu-Sakarya, MacKinnon, & Aiken (2013). Using equation 1, adjusted AVST post-scores were created in order to be able to visualize the effects (Figure 6). Using equation 2, a regression analysis for each moderator was analyzed. This approach is statistically similar to using ANCOVAs with pre-training AVST AB as a covariate. The main advantage of using residual change scores is that the adjusted post-training scores can be visualized clearly. A difference between both methods is that post-training scores are regressed on overall regression coefficients to adjust for pre-training differences, whereas with ANCOVA separate coefficients for each group are created. However, as both groups do not differ on average on pre-training AB, both methods would have given nearly identical results.
Figure 6. Moderators of change reaching/approaching significance
For binge drinking at pre-training, general impulsivity and attentional control no significant effects were found. Figure 6 shows a similar pattern for the three moderators reaching/approaching significance. Higher values on the moderator were related with higher values on adjusted post-training AB scores in the control group, but related with lower values on adjusted post-training AB scores in the trained group. For motor impulsivity, severity of alcohol use at pre-training and pre-training AB, it seems that VS-ABM training worked best (ie. resulted in low post-training AB) for participant with high impulsivity, more severe alcohol use at training and higher training AB. Notably, only for pre-training AB, the crucial interaction effect was statistically significant (Baseline AB x Group: B = 1.89, SE = .41, t = 4.62, p < .001; Baseline AB: B = -.80, SE = .26, t = -3.09, p = .003). For motor impulsivity, the interaction effect approached significance (B = 189.97 SE = 95.63, t = 1.99, p = .052). Severity of alcohol use at pre-training showed a significant main effect of AUDIT Total (B = -101.32, SE = 49.46, t = -2.05, p = .045). The interaction effect approached significance (p = .076). Identical analyses were used to
assess the effect of the moderators on VPT AB, but none of the moderators showed the crucial Group x Moderator interaction or a main effect of the moderator.
Discussion
The current study assessed the efficacy of online VS-ABM training compared to online control training on reducing attentional biases (AB), alcohol intake and craving in a sample of heavy drinking young adults. Our results showed inconsistent effects of VS-ABM on AB, yet no statistically significant effects on alcohol intake and craving. AB were measured using a visual search task (AVST), and visual probe task (VPT). VS-ABM training resulted in reduced AB as measured by the VPT, showing a larger decline for the trained group by comparison to the control group. AB assessed by the AVST did not show significant group differences in the reduction of AB, but the significant bias towards alcohol in the trained group at pre-training was not apparent at post-training. Exploratory analyses on alcohol and non-alcohol trials in both tasks indicated that VS-ABM resulted in a general speeding for both trial types on the AVST. For the VPT, VS-ABM led to stimulus specific speeding on non-alcohol trials. This is congruent to the purpose of the training, as we predicted that VS-ABM would make it easier to look for stimuli that are an alternative to alcohol-related stimuli. Taken together it seems that VS-ABM training resulted in a significant change in VPT AB, yet not in AVST AB. The discrepancy between the two tasks underlines previously found poor convergent reliability between the two tasks (Van Bockstaele et al., 2017) which was further confirmed by non-significant correlations between AVST AB and VPT AB at pre-training, indicating that participants with high AVST AB did not necessarily have high VPT AB and vice versa. However, Bayesian analyses showed insensitive effects for both tasks. Thus, although the effects of VS-ABM in reducing VPT AB are promising, they should also be interpreted with care.
We demonstrated good split half reliability of the AVST, which extends research from anxiety focused ABM showing that the AVST might be a more reliable measure of AB than the VPT (Van Bockstaele et al., 2017) as the VPT is known for having poor reliability (for a review see Ataya et al., 2012). By contrast, task validity was slightly better for the VPT than AVST as some of the baseline measures of drinking related to the VPT, but none of these measures related to the AVST. These psychometric differences might explain the aforementioned discrepancies between the AVST and VPT. Another reason for their poor convergence might be that the two tasks do not target equivalent processes. Whilst the VPT is naturally very fast and therefore focused around implicit processes, the AVST is much slower which makes it plausible that more explicit processes could be involved in this task too. According to Bar-Haim (2010) concordant, use of explicit processing and implicit processing might beneficial in ABM training, supporting the efficacy of VS-ABM over visual probe ABM, yet these differences are still
speculative. Future research should hence focus on detecting the exact processes behind both tasks, possibly with the use of neurocognitive measures such as fMRI and EEG.
Analyses of possible working mechanisms in this study showed that high AVST AB at baseline was a significant moderator of change. Thus, participants with high AVST AB at baseline showed a larger decline in AB from pre- to post-training. In addition, severity of alcohol problems at baseline, and baseline behavioral impulsivity predicted change but only at a trend level. Thus a trend was found, indicating that VS-ABM may be more effective in reducing AVST AB in participants who showed more severe alcohol problems and higher behavioral impulsivity at baseline. Based on previous research it was hypothesized that more severe alcohol problems related to higher baseline AB (for a review see Field & Cox, 2008), through which it might increase the efficacy of VS-ABM. This was confirmed by the current research, as participants with more severe alcohol problems seemed to have benefited more from the training. However, baseline AVST AB was not significantly related to the severity of alcohol problems at baseline and only marginal significant positive correlations between baseline VPT AB and the severity of alcohol problems were found. For behavior impulsivity a similar conclusion could be drawn. Participants with higher baseline behavioral impulsivity seemed to have benefited more from VS-ABM training compared to participants with lower baseline behavioral impulsivity as was hypothesized based on previous research (Coskunpinar & Cyders, 2013). By contrast, general impulsivity and attentional control did not moderate the effect of VS-ABM on AVST AB. Participants with heightened general impulsivity or lowered attentional control did not show larger reductions in AB after VS-ABM training. It can be argued that this in line with previous research in which a less consistent role was found for general impulsivity and attentional control compared to behavioral impulsivity and the severity of alcohol problems in its relation to alcohol AB (e.g. Janssen, Larsen, Vollebergh, & Wiers 2015; Willem, Vasey, Beckers, Claes, & Bijttebier, 2013). This suggests that future research should mostly focus on VS-ABM for participants with severe alcohol problems and high behavioral impulsivity. It should be noted that different operationalizations of impulsivity and attentional control might have led to different results as in the current study only questionnaires were used to measure these behavioral constructs.
Nonetheless, it is possible that the lack of significant effects of VS-ABM on alcohol use and craving in the current study, may be due to the inconsistent effects of VS-ABM in reducing AB. A previous review on ABM in anxiety (McCleod & Clarke, 2015) has shown that most ABM studies which have not found an effect on anxiety symptoms were not able to reliably change biases towards negative stimuli. Recently, a similar argument was made for ABM in addiction research in a commentary by Wiers (2016), hence if biases towards alcohol are not sufficiently changed, no changes in craving and drinking will occur either. Possibly, in this study too, effects of VS-ABM on AB were too small to reduce drinking
and/or craving for alcohol. This might be attributed to the limitations of the current study. These limitations will now be discussed.
Firstly, although the current sample was highly valuable in multiple ways, such as having high cognitive plasticity while also being in a vulnerable period for developing maladaptive drinking, the sample used in this study was possibly suboptimal for examining clinical changes such as changes in craving or drinking behaviour. It could have been problematic that participants in our study were not actively searching for help to reduce their alcohol use. Therefore, a factor which could have been reduced in the current study is ‘motivation to change’. According to Wiers and Stacy (2006) sufficient motivation to change is a prerequisit in any intervention targeting drug abuse. Although participants indicated to be rather motivated to change their drinking, they also indicated that they mainly wanted to reduce drinking and not completely stop drinking. Above this, the current sample showed inconsistent baseline AB towards alcohol. The AVST indicated general AB towards alcohol, yet the VPT did not. Possibly, this indicates that baseline AB were not sufficiently high for the training to have an effect. Most likely, both motivation to change and baseline AB will be higher in an alcohol dependent clinical sample. Therefore, a promising line of research would be to replicate the current study with a clinical sample in order to see whether our non-clinical sample could explain the current null-results.
Secondly, a major problem in this study was that the days between training sessions showed a lot of variability within and between participants. As such, training was intense and rather structured for some participants, yet, other participants might have only done one training session per week or had a very scattered training. The variability in the number of days between training sessions might have contributed to the lack of succes of VS-ABM training for some participants. Therefore, a stricter timing of the training sessions should be incorporated in future studies, to ensure that the time between pre- and post-training assessments is equal for all participants ,and that training sessions are more equally spread out within that time. It was also indicated by an earlier review that eight training sessions in four weeks would be optimal for a clinical trial focussed on patients with social phobia (Bar-Haim, 2010). This was further confirmed by research on anxiety related AB, in which VS-ABM training was more succesful for participants who completed training sessions (De Voogd, 2016), with a maximum of eight sessions. We have no reason to assume that this guideline would not be suitable for alcohol dependent patients, suggesting a change from six to eight sesssions in a future study.
Despite these limitations our study provides a first insight into the efficacy of a new form of ABM training, VS-ABM, in heavy drinking young adults. Our study confirms previous findings that visual search AB and visual probe AB might measure somewhat different constructs. In terms of reliability, AVST AB seems to surpass VPT AB, yet more validation was found for VPT AB as this (at least to some extend) related to measures of drinking and craving, whilst AVST AB did not. To date
working mechanisms of both tasks are still to be investigated to fully understand these differential effects. This study showed that baseline AVST AB, baseline alcohol problems and baseline behavioral impulsivity might be a good starting point for further research. The lack of significant effects on craving and alcohol use can most likely be attributed to insufficient changes in AB. Insufficient change in AB might be due to inconsistent baseline AB, the use of a non-clinical sample and large variabilty in the number- and dispersion of days between training sessions. As such, replication using a randomized clinical trial is strongly advocated. Unfortunately, it cannot (yet) be proven that alcohol related VS-ABM does not suffer from the same struggles as other ABM paradigms using visual probe and alcohol-related Stroop training. However, the current study gave more insights on how VS-ABM might work succesfully in alcohol dependent patients, giving a head start to future research.
References
Ataya, A. F., Adams, S., Mullings, E., Cooper, R. M., Attwood, A. S., & Munafò, M. R. (2012). Internal reliability of measures of substance-related cognitive bias. Drug and Alcohol Dependence, 121(1), 148-151. doi:10.1016/j.drugalcdep.2011.08.023
Babor, T. F., de la Fuente, J. R., Saunders, J., Grant, M. (1992) The Alcohol Use Disorders Identification Test: Guidelines for use in primary health care (WHO Publication No. 92.4). World Health Organization, Geneva, Switzerland. doi: 10.1.1.505.4146
Bar-Haim, Y. (2010). Research review: attention bias modification (ABM): a novel treatment for anxiety disorders. Journal of Child Psychology and Psychiatry, 51(8), 859-870. doi:10.1111/j.1469-7610.2010.02251.x
Berridge, K. C., & Robinson, T. E. (2003). Parsing reward. Trends in Neurosciences, 26(9), 507-513. doi:10.1016/S0166-2236(03)00233-9
Boendermaker, W. J., Maceiras, S. S., Boffo, M., & Wiers, R. W. (2016). Attentional bias modification with serious game elements: evaluating the shots game. JMIR serious games, 4(2), e20. doi: 10.2196/games.6464
Casey, B. J., & Jones, R. M. (2010). Neurobiology of the adolescent brain and behavior: implications for substance use disorders. Journal of the American Academy of Child & Adolescent Psychiatry, 49(12), 1189-1201. doi:10.1016/j.jaac.2010.08.017
Centraal Bureau voor Statistiek. (2017). Retrieved on 30-08-2017 from
http://statline.cbs.nl/Statweb/selection/?VW=T&DM=SLNL&PA=83021ned&D1=23&D2=0-13,30-42&D3=0&D4=l&HDR=T&STB=G1,G2,G3
Coskunpinar, A., & Cyders, M. A. (2013). Impulsivity and substance-related attentional bias: a meta-analytic review. Drug and Alcohol Dependence, 133(1), 1-14.
doi:10.1016/j.drugalcdep.2013.05.008
Crews, F. T., & Boettiger, C. A. (2009). Impulsivity, frontal lobes and risk for addiction. Pharmacology Biochemistry and Behavior, 93(3), 237-247. doi:10.1016/j.pbb.2009.04.018
Crews, F., He, J., & Hodge, C. (2007). Adolescent cortical development: a critical period of vulnerability for addiction. Pharmacology Biochemistry and Behavior, 86(2), 189-199. doi:10.1016/j.pbb.2006.12.001
Crone, E. A. (2009). Executive functions in adolescence: inferences from brain and behavior. Developmental science, 12(6), 825-830. doi:10.1111/j.1467-7687.2009.00918.x
Dandeneau, S. D., Baldwin, M. W., Baccus, J. R., Sakellaropoulo, M., & Pruessner, J. C. (2007). Cutting stress off at the pass: reducing vigilance and responsiveness to social threat by manipulating
attention. Journal of Personality and Social Psychology, 93(4), 651-666. doi:10.1037/0022-3514.93.4.651
De Voogd, E. L., Wiers, R. W., Prins, P. J. M., & Salemink, E. (2014). Visual search attentional bias modification reduced social phobia in adolescents. Journal of Behavior Therapy and Experimental Psychiatry, 45(2), 252-259. doi:10.1016/j.jbtep.2013.11.006
De Voogd, E. L., Wiers, R. W., Prins, P. J. M., de Jong, P. J., Boendermaker, W. J., Zwitser, R. J., & Salemink, E. (2016). Online attentional bias modification training targeting anxiety and depression in unselected adolescents: Short-and long-term effects of a randomized controlled trial. Behaviour Research and Therapy, 87, 11-22. doi:10.1016/j.brat.2016.08.018
De Wit, H. (2009). Impulsivity as a determinant and consequence of drug use: a review of underlying processes. Addiction biology, 14(1), 22-31. doi:10.1111/j.1369-1600.2008.00129.x
Dennis, M. L., Funk, R., Godley, S. H., Godley, M. D., & Waldron, H. (2004). Cross‐validation of the alcohol and cannabis use measures in the Global Appraisal of Individual Needs (GAIN) and Timeline Followback (TLFB; Form 90) among adolescents in substance abuse treatment. Addiction, 99(s2), 125-133. doi:10.1111/j.1360-0443.2004.00859.x
Derryberry, D., & Reed, M. A. (2002). Anxiety-related attentional biases and their regulation by attentional control. Journal of abnormal psychology, 111(2), 225. doi:10.1037//0021-843X.111.2.225
Dienes, Z. (2014). Using Bayes to get the most out of non-significant results. Frontiers in psychology, 5. doi:10.3389/fpsyg.2014.00781
Dienes, Z., & Mclatchie, N. (2017). Four reasons to prefer Bayesian analyses over significance testing. Psychonomic Bulletin & Review, 1-12. doi:10.3758/s13423-017-1266-z
Duka, T., Trick, L., Nikolaou, K., Gray, M. A., Kempton, M. J., Williams, H., ... & Stephens, D. N. (2011). Unique brain areas associated with abstinence control are damaged in multiply detoxified alcoholics. Biological Psychiatry, 70(6), 545-552. doi:10.1016/j.biopsych.2011.04.006
Elfeddali, I., de Vries, H., Bolman, C., Pronk, T., & Wiers, R. W. (2016). A randomized controlled trial of Web-based Attentional Bias Modification to help smokers quit. Health Psychology, 35(8), 870. doi:10.1037/hea0000346
Ernst, M., Pine, D. S., & Hardin, M. (2006). Triadic model of the neurobiology of motivated behavior in adolescence. Psychological medicine, 36(3), 299-312. doi:10.1017/S0033291705005891
Everitt, B. J., & Robbins, T. W. (2005). Neural systems of reinforcement for drug addiction: from actions to habits to compulsion. Nature Neuroscience, 8(11), 1481-1489. doi:10.1038/nn1579
Fadardi, J. S., & Cox, W. M. (2009). Reversing the sequence: reducing alcohol consumption by overcoming alcohol attentional bias. Drug and Alcohol Dependence, 101(3), 137-145. doi:10.1016/j.drugalcdep.2008.11.015
Field, M., & Cox, W. M. (2008). Attentional bias in addictive behaviors: a review of its development, causes, and consequences. Drug and Alcohol Dependence, 97(1), 1-20. doi:10.1016/j.drugalcdep.2008.03.030
Field, M., & Eastwood, B. (2005). Experimental manipulation of attentional bias increases the motivation to drink alcohol. Psychopharmacology, 183(3), 350-357. doi:10.1007/s00213-005-0202-5
Field, M., Duka, T., Eastwood, B., Child, R., Santarcangelo, M., & Gayton, M. (2007). Experimental manipulation of attentional biases in heavy drinkers: do the effects generalise? Psychopharmacology, 192(4), 593-608. doi:10.1007/s00213-007-0760-9
Field, M., Munafò, M. R., & Franken, I. H. (2009). A meta-analytic investigation of the relationship between attentional bias and subjective craving in substance abuse. Psychological Bulletin, 135(4), 589-607. doi:10.1037/a0015843
Field, M., Wiers, R. W., Christiansen, P., Fillmore, M. T., & Verster, J. C. (2010). Acute alcohol effects on inhibitory control and implicit cognition: implications for loss of control over drinking. Alcoholism: Clinical and Experimental Research, 34(8), 1346-1352. doi:10.1111/j.1530-0277.2010.01218.x
Friese, M., Hofmann, W., & Wiers, R. W. (2011). On taming horses and strengthening riders: Recent developments in research on interventions to improve self-control in health behaviors. Self and Identity, 10(3), 336-351. doi:10.1080/15298868.2010.536417
Garland, E. L., Boettiger, C. A., Gaylord, S., Chanon, V. W., & Howard, M. O. (2012). Mindfulness is inversely associated with alcohol attentional bias among recovering alcohol-dependent adults. Cognitive therapy and research, 36(5), 441-450. doi:10.1080/02791072.2010.10400690
Geier, C., & Luna, B. (2009). The maturation of incentive processing and cognitive control. Pharmacology Biochemistry and Behavior, 93(3), 212-221. doi:10.1016/j.pbb.2009.01.021
Gunn, R. L., & Finn, P. R. (2013). Impulsivity partially mediates the association between reduced working memory capacity and alcohol problems. Alcohol, 47(1), 3-8. doi:10.1016/j.alcohol.2012.10.003
Hicks, J. A., Fields, S., Davis, W. E., & Gable, P. A. (2015). Heavy drinking, impulsivity and attentional narrowing following alcohol cue exposure. Psychopharmacology, 232(15), 2773- 2779. doi:10.1007/s00213-015-3912-3
Hofmann, W., Gschwendner, T., Friese, M., Wiers, R. W., & Schmitt, M. (2008). Working memory capacity and self-regulatory behavior: toward an individual differences perspective on behavior
determination by automatic versus controlled processes. Journal of Personality and Social Psychology, 95(4), 962-977. doi:10.1037/a0012705
Janssen, T., Larsen, H., Vollebergh, W. A., & Wiers, R. W. (2015). Longitudinal relations between cognitive bias and adolescent alcohol use. Addictive Behaviors, 44, 51-57. doi:10.1016/j.addbeh.2014.11.018
Kabat‐Zinn, J. (2003). Mindfulness‐based interventions in context: past, present, and future. Clinical psychology: Science and practice, 10(2), 144-156. doi: 10.1093/clipsy.bpg016
Kisbu-Sakarya, Y., MacKinnon, D. P., & Aiken, L. S. (2013). A Monte Carlo comparison study of the power of the analysis of covariance, simple difference, and residual change scores in testing two-wave data. Educational and psychological measurement, 73(1), 47-62. doi:10.1177/0013164412450574
Kramer, J. R., Chan, G., Hesselbrock, V. M., Kuperman, S., Bucholz, K. K., Edenberg, H. J., ... & Bierut, L. J. (2010). A principal components analysis of the abbreviated Desires for Alcohol
Questionnaire (DAQ). Journal of studies on alcohol and drugs, 71(1), 150-155. doi:10.15288/jsad.2010.71.150
Love, A., James, D., & Willner, P. (1998). A comparison of two alcohol craving questionnaires. Addiction, 93(7), 1091-1102. doi:10.1046/j.1360-0443.1998.937109113.x
Moeller, F. G., Dougherty, D. M., Barratt, E. S., Schmitz, J. M., Swann, A. C., & Grabowski, J. (2001). The impact of impulsivity on cocaine use and retention in treatment. Journal of substance abuse treatment, 21(4), 193-198. doi:10.1016/S0740-5472(01)00202-1
Neighbors, C., Lee, C. M., Lewis, M. A., Fossos, N., & Larimer, M. E. (2007). Are social norms the best predictor of outcomes among heavy-drinking college students?. Journal of studies on alcohol and drugs, 68(4), 556-565. doi:10.15288/jsad.2007.68.556
Nikolaou, K., Field, M., Critchley, H., & Duka, T. (2013). Acute alcohol effects on attentional bias are mediated by subcortical areas associated with arousal and salience attribution. Neuropsychopharmacology, 38(7), 1365-1373. doi:10.1038/npp.2013.34
Ólafsson, R. P., Smári, J., Guðmundsdóttir, F., Ólafsdóttir, G., Harðardóttir, H. L., & Einarsson, S. M. (2011). Self reported attentional control with the Attentional Control Scale: Factor structure and relationship with symptoms of anxiety and depression. Journal of anxiety disorders, 25(6), 777-782. doi:10.1016/j.janxdis.2011.03.013
Patton, J. H., & Stanford, M. S. (1995). Factor structure of the Barratt impulsiveness scale. Journal of Clinical Psychology, 51(6), 768-774. Retrieved from: https://msrc.fsu.edu/
Phaf, R. H., & Kan, K. J. (2007). The automaticity of emotional Stroop: A meta-analysis. Journal of Behavior Therapy and Experimental Psychiatry, 38(2), 184-199. doi:10.1016/j.jbtep.2006.10.008
Presley, C. A., Meilman, P. W., Cashin, J. R., & Lyerla, R. (1996). Alcohol and drugs on American college campuses: Use, consequences, and perceptions of the campus environment: Volume III: 1991-93. Carbondale, IL: Southern Illinois University. Retrieved from
http://www.ncjrs.gov/App/publications/abstract.aspx?ID=164209
Pronk, T., van Deursen, D. S., Beraha, E. M., Larsen, H., & Wiers, R. W. (2015). Validation of the Amsterdam Beverage Picture Set: a controlled picture set for cognitive bias measurement and modification paradigms. Alcoholism: clinical and experimental research, 39(10), 2047-2055. doi:10.1111/acer.12853
Reinert, D. F., & Allen, J. P. (2002). The alcohol use disorders identification test (AUDIT): a review of recent research. Alcoholism: Clinical and Experimental Research, 26(2), 272-279. doi:10.1111/j.1530-0277.2002.tb02534.x
Schoenmakers, T. M., de Bruin, M., Lux, I. F., Goertz, A. G., Van Kerkhof, D. H., & Wiers, R. W. (2010). Clinical effectiveness of attentional bias modification training in abstinent alcoholic patients. Drug and Alcohol Dependence, 109(1), 30-36.
Schoenmakers, T., Wiers, R. W., Jones, B. T., Bruce, G., & Jansen, A. (2007). Attentional re‐training decreases attentional bias in heavy drinkers without generalization. Addiction, 102(3), 399-405. doi:10.1016/j.drugalcdep.2009.11.022
Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological science, 18(5), 429-434. Retrieved from http://www.jstor.org/stable/40064634
Sobell, L. C., & Sobell, M. B. (1995). Alcohol timeline followback users’ manual. Toronto, Canada: Addiction Research Foundation. Retrieved from: https://pubs.niaaa.nih.gov/
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(5), 385-395. doi:10.1016/j.paid.2009.04.008
Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of social behavior. Personality and Social Psychology Review, 8(3), 220-247. doi:10.1207/s15327957pspr0803_1
Tapert, S. F., Schweinsburg, A. D., Barlett, V. C., Brown, S. A., Frank, L. R., Brown, G. G., & Meloy, M. J. (2004). Blood oxygen level dependent response and spatial working memory in adolescents with alcohol use disorders. Alcoholism: Clinical and Experimental Research, 28(10), 1577-1586. doi:10.1097/01.ALC.0000141812.81234.A6
Van Bockstaele, B., Salemink, E., Bögels, S. M., & Wiers, R. W. (2017). Limited generalisation of changes in attentional bias following attentional bias modification with the visual probe task. Cognition and Emotion, 31(2), 369-376. doi:10.1080/02699931.2015.1092418
Van Breukelen, G. J. (2006). ANCOVA versus change from baseline had more power in randomized studies and more bias in nonrandomized studies. Journal of clinical epidemiology, 59(9), 920-925. doi:10.1016/j.jclinepi.2006.02.007
Van Hemel-Ruiter, M. E., de Jong, P. J., Ostafin, B. D., & Wiers, R. W. (2015). Reward sensitivity, attentional bias, and executive control in early adolescent alcohol use. Addictive behaviors, 40, 84-90. doi:10.1016/j.addbeh.2014.09.004
Wagenmakers, E. J., Marsman, M., Jamil, T., Ly, A., Verhagen, J., Love, J., ... & Matzke, D. (2017). Bayesian inference for psychology. Part I: Theoretical advantages and practical ramifications. Psychonomic Bulletin & Review, 1-23.doi:10.3758/s13423-017-1343-3
Waters, A. M., Pittaway, M., Mogg, K., Bradley, B. P., & Pine, D. S. (2013). Attention training towards positive stimuli in clinically anxious children. Developmental Cognitive Neuroscience, 4, 77-84. doi:10.1016/j.dcn.2012.09.004
Weckler, H., Kong, G., Larsen, H., Cousijn, J., Wiers, R. W., & Krishnan-Sarin, S. (2017). Impulsivity and approach tendencies towards cigarette stimuli: Implications for cigarette smoking and cessation behaviors among youth. Experimental and Clinical Psychopharmacology, 25(5), 363-372. doi:10.1037/pha0000139
Wiers, R. W. (2016) Cognitive Bias Modification Does Hold Promise in the Treatment of Addiction: A commentary on Cristea et al 2016 PlosOne. PlosOne. 2016;11: e0162226 (comment).
Wiers, R. W., Gladwin, T. E., Hofmann, W., Salemink, E., & Ridderinkhof, K. R. (2013). Cognitive bias modification and cognitive control training in addiction and related psychopathology: Mechanisms, clinical perspectives, and ways forward. Clinical Psychological Science, 1(2), 192-212. doi:10.1177/2167702612466547
Wiers, R. W., Houben, K., Fadardi, J. S., Van Beek, P., Rhemtulla, M., & Cox, W. M. (2015). Alcohol cognitive bias modification training for problem drinkers over the web. Addictive Behaviors, 40, 21-26. doi:10.1016/j.addbeh.2014.08.010
Wiers, R. W., & Stacy, A. W. (2006). Implicit cognition and addiction. Current Directions in Psychological Science, 15(6), 292-296. doi:10.1111/j.1467-8721.2006.00455.x
Willem, L., Vasey, M. W., Beckers, T., Claes, L., & Bijttebier, P. (2013). Cognitive biases and alcohol use in adolescence and young adulthood: the moderating role of gender, attentional control and inhibitory control. Personality and Individual Differences, 54(8), 925-930. doi:10.1016/j.paid.2013.01.015
